Open main menu

Wikidata:Project chat/Archive/2017/07

This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

Contents

What happened to the cooperation with the US Census bureau?

I remember that a while there was an account that announced here, that they were a person from the US Census Bureau and they wanted to cooperate with Wikidata. As far as I see, most of the US population data is still not well-cited within Wikidata. Does anybody know what happened? ChristianKl (talk) 21:26, 29 June 2017 (UTC)

I don't know specifically on this but I've heard the US Census bureau is struggling with a lower budget and having trouble right now preparing for the upcoming 2020 census. ArthurPSmith (talk) 14:20, 30 June 2017 (UTC)
everybody talks about the census, but nobody wants to spend the money to fix it. we had a meetup at the sunlight foundation, but the feds did not have the support to go to open APIs. they have a publishing model. and we did not fund a wikipedian in residence to facilitate mass uploads. here is their data - https://www.census.gov/data/data-tools.html - Slowking4 (talk) 00:59, 1 July 2017 (UTC)

Exciting news about constraint reports

Hello all,

As you know, the Wikidata team has been working on a tool to improve constraint reports and display them directly on the item pages. This was a user script, and is now a gadget: checkConstraints. You can activate this gadget in your preferences. Thanks to Matěj Suchánek for the conversion!

The constraints are now checked as soon as a statement is added or edited. It just needs to be saved first.

The “unique value” constraint is now implemented. The message for this constraint still needs to be improved, but we’re working on it.

The “type” and “value type” constraints should work better now, but some problems are still occurring. If you have any trouble with one of these, feel free to ping Lucas Werkmeister (WMDE) or add a comment on the ticket.

We will soon re-import constraint templates once more, since some templates had errors.

But there’s more! We are preparing the migration to constraint statements. When this is ready, Ivan A. Krestinin (owner of KrBot) will run a bot to migrate all constraint templates to statements on the properties.

This should happen at some point in July. During the transition, some small problems may occur (it’s possible that for a short time you will see constraints twice, or not at all).

Thanks to all who provided feedback, ideas and help on this feature :) Lea Lacroix (WMDE) (talk) 13:00, 30 June 2017 (UTC)

Seems like the constraints gadget is clashing with the authority control gadget. For example, the unique value constraint on Commons category (P373) sometimes appears, but mostly doesn't. Another reason why we should have more hooks in the UI. Sjoerd de Bruin (talk) 14:58, 30 June 2017 (UTC)
Hm, I don’t think we can fix this on our end. checkConstraints appends to the statement value, while AuthorityControl completely replaces it – if AuthorityControl runs after checkConstraints, it probably removes checkConstraints’ indicator in the process. (However, I didn’t see that happen once after reloading Douglas Adams (Q42) about a dozen times – which test item did you use?)
I agree that we need more stable interfaces for the UI, though I don’t see how a hook in particular would be useful here. --Lucas Werkmeister (WMDE) (talk) 12:58, 1 July 2017 (UTC)
Do we need some constraint checks for the new method? GZWDer added Come On Over (Q522293) (a disambiguation page) to exception to constraint (P2303) of taxon name (P225). All these values are required to have taxon name (P225) themself. --Succu (talk) 17:57, 30 June 2017 (UTC)
Sorry, I don’t understand what you mean – are you saying that we should infer from the exception list that the listed values need a taxon name (P225) statement, and report a violation if they don’t have one? --Lucas Werkmeister (WMDE) (talk) 12:58, 1 July 2017 (UTC)
Hi Lea, we have now hundreds of unstructured entries of exception to constraint (P2303) related to taxon name (P225). Are there any plans (UI, gadget) how we should manage them in the future? --Succu (talk) 21:08, 30 June 2017 (UTC)
Not from our end, but apparently GZWDer has some? I just saw phab:T169374 and, through that, exceptions for test (Q21532100), and I guess that could be made to work. --Lucas Werkmeister (WMDE) (talk) 12:58, 1 July 2017 (UTC)

i wanna share my data

hello,wikipedia i wanna share my data with you which is so important and will help many people ubiquitous to simply find my biography.  – The preceding unsigned comment was added by Aawisdom (talk • contribs) at 07:43, 1 July 2017‎ (UTC).

@Aawisdom: This is not Wikipedia; this is Wikidata. Please see WD:Autobiography. For Wikipedia (at least, the one written in the English language), see en:WP:Autobiography. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:25, 1 July 2017 (UTC)

Difference between human large intestine (Q5640957) and large intestine (Q11083)

Both human large intestine (Q5640957) and large intestine (Q11083) seem to have something to do with the large intestine. Unfortunately, the first item currently has no English name. Can someone who knows Russian give it an appropriate English name? ChristianKl (talk) 12:23, 1 July 2017 (UTC)

human large intestine (Q5640957) refers to human (Q5). - Kareyac (talk) 14:58, 1 July 2017 (UTC)
Thanks, I added subclass of (P279) human organ (Q25449120) to it. ChristianKl (talk) 16:03, 1 July 2017 (UTC)

Mapping languages on Wikidata

Hi all

I'm trying to map all the languages from the UNESCO Atlas of World Languages in Danger on Wikidata before 6th July for Celtic Knot: Wikipedia Language conference. There are around 400 items left, all help would be appreciated.

There's also a matching taking place for Glottolog which is a more comprehensive database but it has over 15,000 items still to match, if anyone knows any clever ways to match them that would be amazing.

Thanks very much

--John Cummings (talk) 15:14, 24 June 2017 (UTC)

@John Cummings: You might want to check out OpenRefine - latest release has wikidata matching allowing you to correlate metadata with specific wikidata properties. I'm not sure how workable that would be for languages but it might be worth trying. Once you have a list of matches you can export a CSV file from OpenRefine that can be tweaked to run in QuickStatements. ArthurPSmith (talk) 14:25, 26 June 2017 (UTC)
Many thanks ArthurPSmith (talkcontribslogs), this is quite a lot beyond my technical ability but is anyone else knows how to do it I'm very happy to do any grunt work needed. Also you may be interested in this --John Cummings (talk) 14:35, 26 June 2017 (UTC)

Only 100 items are missing for the UNESCO Atlas of World Languages in Danger (much more for glottolog ...). Let's finish that quickly   Pamputt (talk) 20:27, 28 June 2017 (UTC)

Done :) --John Cummings (talk) 12:52, 2 July 2017 (UTC)

Use of Q7066 as a religion

Is religion (P140)atheism (Q7066) correct? Atheism isn't really a religion, so it might be a misapplication of the property. There was an enwiki RFC a while ago which resulted in the consensus that "religion: atheism" and "religion: agnosticism" shouldn't be included in infoboxes. (The consensus was also that "religion: none" shouldn't be included, but we can still arbitrarily use no value for everything so I'm unsure of the implications of that here.) Jc86035 (talk) 15:45, 24 June 2017 (UTC)

The general principle of Wikidata is plurality. If sources say that a person's religion is "atheism" we can report this in Wikidata. ChristianKl (talk) 16:26, 24 June 2017 (UTC)
We have lifestyle (P1576) for the other situations though. There is still some discussion happening on the talk page there though. Sjoerd de Bruin (talk) 16:29, 24 June 2017 (UTC)
This is fine when religion is a subclass of lifestyle. Thanks, GerardM (talk) 17:17, 24 June 2017 (UTC)
No it is not correct; just as "not collecting stamps" is not a hobby. You may though use "religion -> no value", and give a source if you have one. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:40, 24 June 2017 (UTC)
Few people make a point to publically identify as a person who's "not collecting stamps" but a decent amount of people does identify as "atheist". ChristianKl (talk) 19:39, 24 June 2017 (UTC)
@ChristianKl: But atheism still isn't a religion ("sacred belief system"), so it's strictly incorrect to say with the current definition of the property that someone's religion is atheism. Jc86035 (talk) 04:06, 25 June 2017 (UTC)
@Jc86035: You make more or less the same point as The Bible: "A god who cannot walk, cannot hear, cannot see and cannot talk is not a god." If I make that my postulate I could remove such statements from Zeus (Q34201) too. No! Let our sources guide us here, not our private opinions. -- Innocent bystander (talk) 06:51, 25 June 2017 (UTC)
@Jc86035: If you use the term "sacred" in the way it's used in modern decision science, I think there are plenty of beliefs in the cluster that the average atheists holds that are sacred for them. ChristianKl (talk) 08:39, 25 June 2017 (UTC)
@ChristianKl: Regardless of what the definition of religion (Q9174) is, atheism (Q7066) is not an instance of (P31) religion (Q9174), so using religion (P140)atheism (Q7066) is incorrect and would be better as religion (P140)no value and [some other property] → atheism (Q7066). Alternately items which are instance of (P31) world view (Q49447) could be accepted as values for religion (P140) and the definition of the property could be changed. Jc86035 (talk) 08:48, 25 June 2017 (UTC)
My opinion is that novalue cannot be used to imply Atheism. That better fit those who tell they have no opinions about religion. -- Innocent bystander (talk) 14:05, 25 June 2017 (UTC)
The indication of religion identifies association more than a firm belief. The level of religiosity is not determined in any way. Given that it is about association, it follows that atheism can be seen in this light. They consider that there is no obvious choice to make between any and all of the religions we can believe in. Thanks, GerardM (talk) 15:39, 25 June 2017 (UTC)
@GerardM: No, "they" don't--there are religious atheists. —Justin (koavf)TCM 16:05, 2 July 2017 (UTC)
What do you mean by a "religious" atheist? I do understand there being a firm belief but a religion.. Please provide sources for your statement. Thanks, GerardM (talk) 16:53, 2 July 2017 (UTC)
It's hard for me to clarify but it's just a person who belongs to a religious community and doesn't believe in God. e.g. w:Don Cuppitt or members of the w:en:World Church of the Creator or most Satanists, etc. I don't know what to tell you if you don't understand. —Justin (koavf)TCM 17:10, 2 July 2017 (UTC)

@Jc86035: English label could be changed, Russian states "attitude towards religion" I also added an exception d1g (talk) 09:44, 25 June 2017 (UTC)

  • Statements using religion (P140)atheism (Q7066) should be removed, because atheism is not a subclass of religion (as not collecting st[...]). Here it's even simpler than in en.wikipedia. We have constraints! We are a semantic database! Feel free to enact a new property called "attitude towards religion" (including values such as "religious", "atheist", "agnostic",...) (Being honest, I don't see Shia Islam as an "attitude towards religion"). Strakhov (talk) 17:28, 25 June 2017 (UTC)
1. Not before we have a separate property for world view (Q49447) or existence of God (Q1699404)
2. religion (P140)atheism (Q7066) claims break nothing, because other claims in religion (P140) are impossible for atheists. d1g (talk) 17:52, 25 June 2017 (UTC)
It breaks the very same thing as storing in child (P40) a hypothetic "Q1200000000000" labeled in English "no child". It breaks logic. We have around 4000 properties and growing, what's the problem in creating an extra one for this stuff? We can call it "post-world view", "view of very deep things beyond human comprehension", "attitude towards religion", "position towards the idea of God". It doesn't matter at all. If not, put it in a generic "lifestyle (P1576)", but not in religion because is not a religion. Strakhov (talk) 18:25, 25 June 2017 (UTC)
"life stance" was used as English alias, "отношение к религии" is used as main Label in Russian
We use abstract items for some properties anonymous (Q4233718) it isn't something clearly harmful.
Additional properties like "no author" complicate things similar how anonymous (Q4233718) complicates one property. d1g (talk) 18:38, 25 June 2017 (UTC)
Yeah, we use abstract items, it's not clearly harmful... but, I insist, what's the problem in storing different things as "religions" and "positions towards religions" in different properties? Where's the need, the reason after that mix here? We are not paper. After that, different projects will be able to display that data whatever way they want, as a religion or as a view of something or ...not showing it at all. Strakhov (talk) 19:00, 25 June 2017 (UTC)
See number 1, theism (Q620805) aren't different from agnosticism (Q288928) or atheism (Q7066), they represent existence of God (Q1699404) or world view (Q49447)
}
d1g (talk) 19:13, 25 June 2017 (UTC)
I see 1. And? Of course atheism and agnosticism are not the only isms that shouldn't be there. theism (Q620805) should not stored as a religion either. So what? Create a property for that and call it existence of God (Q1699404) or world view (Q49447) or "position" with qualifier "towards" -> "divine existence". Or rename "religion", call it "world view (Q49447)" and get the scope wider. But there's no reason for not doing things right. Strakhov (talk) 19:27, 25 June 2017 (UTC)
I added instance of (P31) religion to atheism, with an authoritative court case as source. ChristianKl (talk) 16:17, 26 June 2017 (UTC)
@ChristianKl: Then you should change this statement and create another class « ideology considered a religion in the court juriction » because a court is authoritative in law, but definitely not in theological philosophy. This jurisdiction is probably not the whole world as well and uses its own arbitrary notion of a religion. Let’s make clear which one the definition is. author  TomT0m / talk page 19:19, 26 June 2017 (UTC)
@TomT0m: Wikidata is designed to allow for different parties having different opinions. statement disputed by (P1310)-qualifiers can be used to indicate that different sources have different opinions. The opinion of the Seventh Circuit of the United States Court of Appeals is certainly not a more arbitrary notion than that of individual Wikidatians. ChristianKl (talk) 19:52, 26 June 2017 (UTC)
@ChristianKl: You misses the point entirely. It’s all about the definition of a religion in the USA. And this is where the problem should be solved : at the definition (hence item or property) level. 1 definition = 1 item. Otherwise you can add the dispute qualifier on every claim in which the item is used in that sense. This is useless data duplication. Worse, this is data loss because we for sure know this is the US notion of a religion that is used and not anyone else. This is not made explicit and you choose to use an ambiguous item instead. This is bad practice. For example, I think a correct use of disputed by would be another US court that had another opinion. Think that, as is, the statement and all the like could be disputed by virtually anyone … this is wayyyy too open. author  TomT0m / talk page 06:33, 27 June 2017 (UTC)
I found this source by looking in Google Scholar for answers to the question "Is Atheism a religion?". I consider that a good general heuristic for answering questions. Wikidata doesn't have any "definition" property. We don't have a Israel according to the Israeli and a Israel according to the Palestianians item even through both define it differently (and with different borders). Since the start, Wikidata has used "disputed by" for cases like that. ChristianKl (talk) 08:32, 27 June 2017 (UTC)
@ChristianKl: Uncompelling at all. They consciously chose a definition : see http://legal-dictionary.thefreedictionary.com/religion for example. They made a choice, they had to. But … this is a choice and has no pretention to universality. Which Wikidata has. By NPOV. The best way so do this is to avoid as much as possible item ambiguity. Hence relying on definitions instead of labels. Wikidata can have as many items with the same label as we like. There is other definition http://www.larousse.fr/dictionnaires/francais/religion/67904 of religion. Certain philosopher even consider (see https://fr.wikipedia.org/wiki/Religion#Critiques_des_d.C3.A9finitions cites Daniel Dubuisson (Q3013975)     thinks the concept of a religion is unusable because too vague or ill defined to actually be used … Which emphasize the fact we should just use more precise concepts. Lets not enter the troll, be specific and objective. What is considered a religion by the USA is objective. «The» definition of «religion» is not. It’s a trap. On the other hand, stuff like « monotheistic religion» are easier and more precise. Better use them. author  TomT0m / talk page 19:19, 27 June 2017 (UTC)
Wikidata NPOV means to report what different sources are saying and use "disputed by" when warrented. It's not a principle that's about excluding viewpoints. It might very well be that there's some sociologist who researched what's called a religion in the US or for that matter in other countries. Currently, I don't see that you linked or anybody else linked to any such sources. Wikidata does have the more precise concept of monotheistic religion (Q19842652) and I do think that it makes sense that Abrahamic religion (Q47280) subclasses it instead of subclassing religion (Q9174) directly.
The problem with definitions is that they are in a specific language. In Wikidata we don't want to depend on a specific language and thus have no definition property. If you think we should have definitions in Wikidata, feel free to propose a property for definitions.
If I look at the larousse definition Google Translate translates it into "A determined set of beliefs and dogmas defining the relationship of man with the sacred." I do think that "I don't believe in God" is a statement that verbalizes my relationship with the sacred.
For many use-case the concept of religion might be unusable because it's to vague. For the use-case that the court had, it doesn't seem to vague. We have use-cases for which it can be useful. A person might ask themselves about the religion of the politicians that sit in congress and how that shifts over time. Having that information in Wikidata is useful. For that use-case, it's useful to have a broad concept of religion that includes atheism. ChristianKl (talk) 19:57, 27 June 2017 (UTC)
The advantage of a definition is on the contrary that IT IS translatable. Always an unambigiously. A word is often not translatable by itself, first, and second, Wikidata itself as never been term based. Precisely BECAUSE words are often ambiguous. You’re totally missing this it seems.
« For the use-case that the court had, it doesn't seem to vague. » I say exacly the same thing. BUT I also say the concept is operable for them because they uses a consistent and refined definition. As such it’s an identifiable entity (and legitimate) that deserve its own item and is admissible.
We indeed don’t have a definition property but we have a « instance of » property that does the trick. You can read « Ghandhi is an instance of human » as « Ghandhi corresponds to the definition of a human ». We can even formalize a little bit the definition by using « subclass of» (human subclass of being means « a human is a being »), definition usually begin like this. Other more formal system can be vastly more expressive and can express the definition pretty precisely. And different definitions ( ≈classes) will have different claims, sometime conflicting, and different instances, which make them different entities. author  TomT0m / talk page 21:55, 27 June 2017 (UTC)
Is agnosticism (Q288928) a religion and should be modeled via religion (P140)? --Succu (talk) 20:13, 27 June 2017 (UTC)
It’s at least an « opinion », I think. I think we should either broaden « religion » scope an rename it to « opinion » on just broden it to « religion or opinion towards it ». I prefer the first one. author  TomT0m / talk page 21:56, 27 June 2017 (UTC)
Probably the intended use of religion (P140) was more related to religious organization (Q1530022). --Succu (talk) 21:28, 27 June 2017 (UTC) An example is Michael Faraday (Q8750). --Succu (talk) 21:42, 27 June 2017 (UTC)::

Not only is atheism not a religion, the analogy to stamps is incorrect as there are millions of religious atheists. This couldn't be more wrong. —Justin (koavf)TCM 16:03, 2 July 2017 (UTC)

Municipalities of Iceland – adding population en masse

Hello, my use and knowledge of Wikidata is very limited, so I don't know where exactly should I discuss my idea (I found "bot requests" but it says the request should be already discussed somewhere else), so maybe this is the place. So I've got a table of Iceland municipalities' population – the site offers various formats, so raw data shouldn't be a problem, (for human-readable source there is Statistical yearbook of Iceland) – and I know adding them one by one like I did in Q214057 isn't the way to go, but I am incapable of anything better. :) It's only a few dozens of items, because some are missing in the source, some already have P1082 filled and there is only so many municipalities you can fit on an Iceland. Thank you for any feedback and/or help.--TFerenczy (talk) 09:21, 30 June 2017 (UTC)

@TFerenczy: welcome to Wikidata! I think you can do it best with QuickStatements, the manual is here. To use this tool, you need to know the Qid's of the municipalities that you want to add data to and put it in a schema like this:
Qxxx|P1082|number of citizens|P585|point in time|S854|source-URL|S813|retreived||
Qyyy|P1082|number of citizens|P585|point in time|S854|source-URL|S813|retreived||
The S is for Source so the property will be added in the source section of the statement.
This should work, but when I ran it myself, it wouldn't add the qualifier point in time (P585)... Anyone else has any idea wh not? Q.Zanden questions? 10:33, 30 June 2017 (UTC)
it is rather picky about the date format - see the "Time values" section of the "How to" documentation. ArthurPSmith (talk) 14:22, 30 June 2017 (UTC)
@QZanden, ArthurPSmith: Thank you very much for the helpful input! So far I've added a dozen and hopefully it's alright; I would greatly appreciate if you (or anyone else reading this) could check it and "greenlight" me to do the remaining 60, I am a bit anxious.--TFerenczy (talk) 15:15, 30 June 2017 (UTC)
@TFerenczy: looks great! Thanks for your data, you get green! Q.Zanden questions? 15:43, 30 June 2017 (UTC)
@TFerenczy, QZanden, ArthurPSmith: et al. The sources look great as such, but I am confused by Vopnafjörður (Q925686) and Vopnafjarðarhreppur (Q3182530). The first is an instance of "city" and "municipality of Island". The latter is a "geographical point" but has the Dutch description "gemeente in IJsland". The Swedish article about the first is about a "populated place" while the latter is about a "municipality". This look messy as a village in Stjørdalen, Norway! What can we do about it? Innocent bystander (talk) 08:11, 1 July 2017 (UTC)
@Innocent bystander: According to the nl.wp articles Q925686 is a city and Q3182530 is a municipality. The links at Q3182530 from en.wp and de.wp are redirects to articles that belong to Q925686. This is quite messy! I don't know what to do... Q.Zanden questions? 17:32, 1 July 2017 (UTC)
Does anybody have a definition of what a "city in Iceland" is? This source tells "Vopnafjörður" is an "urban settlement". "Vopnafjarðarhreppur" probably means "Vopnafjörður Municipality". Looks like only ~100 people live outside of the "urban settlement" in this Municipality.
My advise is to ban every use of "Instance of:City" or "Town". It is too farfetched what is meant by that claim. -- Innocent bystander (talk) 17:45, 1 July 2017 (UTC)
And what would be the difference between "city", "town", "urban settlement" and "village"? Do we have a rule for the minimum population? Beside these items, we also have dozens of items per country... Maybe we should just replace it with something like "place where people live". Q.Zanden questions? 18:47, 1 July 2017 (UTC)
I see no problem with "instance of:Some subclass of city/town". What I see as problematic is that "city" as superclass is not well defined. "City in Alabama, USA" on the other hand is fairly well defined.
I do not think the size is important, since there exists citys with a population of 1 in the U.S.
In English I think they separate incorporated citys from unincorporated citys. From what I have learned the first has mayor and/or councillors and an administration of some kind. Many of them, but not all, can collect tax for example. The latter is more of a "populated place" which looks like a city.
In Sweden we had a subclass of municipality called "city" until 1971, when they all were transformed into one single class of first order municipalities. These citys were from the beginning geographically small entities. As time passed, the rural area around the citys were incorporated into the citys and they became very large. The term "city" outdated itself. To separate urban areas from rural areas, Statistics Sweden started to collect statistics about "urban areas" already in 1890. In that way, the authorities could identify which areas needed to be reorganised. In 1960 Sweden made a deal with other Nordic countries to start to collect Statistics about such urban area in a standardized way. These urban areas have a size of at least 200 persons where the inhabited buildings are located less than 200 meters from each other. I am not sure, but it is possible that these "urban settlements" in Iceland, are such statistic "urban areas". Then they can be as small as 200 inhabitants.
Today we still use the term "City" in Sweded about Swedish entities. Stockholm Municipality call itself "Stockholm City". Technically there is no difference between such a city from other municipalities. City is also used to describe an urban area with the same name as a "pre 1971 city". It can also be used to describe an urban area large enough to have all necessary services to its inhabitants.
In Sweden Statistics Sweden today collect data about so small statistical entities as 50 inhabitants. Those entities is what I have spent most of my time here at Wikidata with the last year. -- Innocent bystander (talk) 07:49, 2 July 2017 (UTC)

Use of categories

Are we now adding categories, as in this edit (part of a large batch) by User:Sic19? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:39, 1 July 2017 (UTC)

I am working to improve the Wikidata items relating to medieval churches. This is a temporary measure to help me track the items until I add instance of and main category statements to all of these items. Sorry if this is a problem for you. If so, I will remove the part of claims immediately and work on something else. Naturally, I am happy to discuss this further. Sic19 (talk) 19:50, 1 July 2017 (UTC)
part of (P361) isn't made to specify that items belong to categories. The linked item is already labeled a church, so I don't see what kind of work you are planning with it. ChristianKl (talk) 21:18, 1 July 2017 (UTC)
Sic19, please remove this kind of additions. part of (P361) is for modelling a whole/part relationship and should not be used this way. --Succu (talk) 21:45, 1 July 2017 (UTC)
OK, I will remove. Can anyone suggest an interim method for retrieving medieval churches in Wikidata until inception dates have been added? I was using part of to group these items while I'm work on them - there aren't enough statements to use queries yet - and intended to remove the part of claims on completion. I know this isn't ideal but I'm working to improve these items and the class structure, which is surely a good thing. Apologies again for any inconvenience caused. Thanks for your feedback and best regards. Sic19 (talk) 21:58, 1 July 2017 (UTC)
You can use the Query Service to collect items according to certain Wikidata statements, or Petscan to collect items according to certain Wikipedia categorizations. In case of questions feel free to ask for help; it would be useful if you describe what your input for the criticized claims was. —MisterSynergy (talk) 22:17, 1 July 2017 (UTC)
@Sic19: You're doing great work on these abbeys and churches! As above already is said, P361 isn't very suitable for such items, even not temporarily. You've found these items already in one way, so maybe it would be better if you save those pages in a notepad-file or in your usernamespace here at wikidata and from there looking for the right claims and putting them in QuickStatements. Again, you're doing a great job on improving these items! Keep up the good work   Q.Zanden questions? 22:21, 1 July 2017 (UTC)
You could use Sandbox-Item (P369).
--- Jura 02:58, 2 July 2017 (UTC)
Just to update, all of the 'part of - category:medieval churches' claims have been removed. Thanks for the suggestions - I think a list of items combined with the use of Petscan and QuickStatements will be sufficient for this task. Sic19 (talk) 07:38, 2 July 2017 (UTC)
@Sic19: Thank you for this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:13, 3 July 2017 (UTC)

SPARQL query for countrywise (US, UK, Germany, Australia) yearwise (1990 - 2017) chart positionwise song title, sale units

Urgent help sought on SPARQL (wikidata) query for countrywise (US, UK, Germany, Australia) yearwise (1990 - 2017) chart positionwise song title, sale units. Also required spotify echo nest API song title wise info on sonic attributes (energy, bpm, speechiness etc.).  – The preceding unsigned comment was added by Spadheewiki (talk • contribs) at 12:45, 2 July 2017‎ (UTC).

What? Also why don't you (whoever wrote this) have a signature? Stryn (talk) 13:03, 2 July 2017 (UTC)
@Spadheewiki: Most of the people who edit here are volunteers (and those who are, are not paid to give you this kind of support). SO if you want urgent help, you need to tell us why this is urgent (in other words, why this is more important than what they would otherwise be working on), plus what it will be used for, and by whom. Andy you need to be much more precise about what it is you're looking for. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:40, 2 July 2017 (UTC)

Help with query

Hi. I need help with this query to get only films that have only one director (P57). Any ideas? --FShbib (talk) 03:36, 3 July 2017 (UTC)

The following query uses these:

  • Properties: instance of (P31)    , director (P57)    , publication date (P577)    , genre (P136)    
     1 SELECT ?item ?itemDescription WHERE { 
     2           ?item wdt:P31 wd:Q11424 .
     3           ?item wdt:P57 ?dir .
     4           ?item wdt:P577 ?date .
     5     FILTER(NOT EXISTS { ?item wdt:P136 wd:Q336144 .})
     6     FILTER EXISTS { ?war schema:about ?item . ?war schema:inLanguage "ar" }
     7     FILTER EXISTS { ?war schema:about ?dir . ?war schema:inLanguage "ar" }
     8     OPTIONAL {?item schema:description ?itemDescription 	filter(lang(?itemDescription) = "ar")}
     9 	FILTER(!BOUND(?itemDescription))
    10 	SERVICE wikibase:label { bd:serviceParam wikibase:language "ar"}
    11   }
    12 LIMIT 200
    
@FShbib: GROUP BY ?item ?itemDescription HAVING (COUNT(?dir) = 1)
Wikidata:A_beginner-friendly_course_for_SPARQL#Grouping d1g (talk) 05:21, 4 July 2017 (UTC)

Wikidata weekly summary #267

How to find all the Q numbers(Item) of a ta.wikipedia category ?

This petscan shows article names with Page IDs. How can i get the result with Item Q numbers also?--Info-farmer (talk) 10:31, 3 July 2017 (UTC)

"Wikidata" → "Add items, where available" (or other options). Matěj Suchánek (talk) 11:08, 3 July 2017 (UTC)
Thanks indeed. By this help, i will add description to every panchayats(<10,000 from32 districts) of my state TamilNadu, India.--Info-farmer (talk) 11:33, 3 July 2017 (UTC)

Watchlist in Client is turning wa-ko

The Watchlist at Wikipedia has been useless for some time with hundreds of lines with the same edit affecting hundreds of articles at Wikipedia. Now bad has become much worse, since I cannot any longer see where ToJack has made hir edit. I can see that (s)he changed a "tg-cyrl" label, but I cannot tell where?! ToJack is not to blame here of course. The more articles on WP are affected, the more useful were probably the edits. But it is not useful for me to see it hundreds of times, when I cannot even reverse it, if I desired to! -- Innocent bystander (talk) 11:52, 3 July 2017 (UTC)

Each line should have a "diff" link. If it does not, please tell me which page(s) are shown. Matěj Suchánek (talk) 15:18, 3 July 2017 (UTC)

Number of objects in libraries, museums and archives

Hello! Is there a property to state the number of volumes a library holds? The same for museums and archives. For example it is well sourced that Museo del Prado (Q160112) collection currently comprises around 8,200 drawings, 7,600 paintings, 4,800 prints, and 1,000 sculptures. I think this info is useful, not only for articles in Wikipedia but also for our internal processes and projects as a data-driven community. Emijrp (talk) 17:26, 3 July 2017 (UTC)

I've been using has parts of the class (P2670) with qualifier quantity (P1114) for this, e.g. at Gaetano Ferolla Museum of Public Transport (Q10333944). Thanks. Mike Peel (talk) 17:30, 3 July 2017 (UTC)
I have found collection or exhibition size (P1436) (example of usage Dhaka University Library (Q630789)). For units it uses "volume" frequently. I think it makes sense to use "sculpture", "paintings", etc as units? Emijrp (talk) 19:47, 3 July 2017 (UTC)

how to fill relevant language data in a Item as No label defined ?

I saw this wikidata page. I am having more than 10,000 data for the labels like this. For example, How can i remove "No label defined" by the relevant data "எடமேலையூர் மேற்கு ஊராட்சி"?--Info-farmer (talk) 06:22, 3 July 2017 (UTC)

I think it's already there: https://www.wikidata.org/wiki/Q23741349?uselang=ta
Maybe native label (P1705) could be added with the text you mention.
"No label defined" primarily refers to the absence of the English label for the item. Maybe that would be "Edamelayur west Gram Panchayat".
QuickStatements (Q20084080) can be used to mass add statements. Please test and check first.
--- Jura 06:53, 3 July 2017 (UTC)
Is there any screencast available for the test and check?--Info-farmer (talk) 07:07, 3 July 2017 (UTC)
It would be silly to give primacy to English and not acknowledge the availability of labels in any other language. No label should mean exactly that .. NO LABEL and when it is to be for a specific language it should be compulary to include that language. Thanks, GerardM (talk) 10:17, 3 July 2017 (UTC)
The website gives primacy to English when the UI is configured by the user to use English. I don't think there's anything wrong with that. ChristianKl (talk) 11:23, 4 July 2017 (UTC)

Welcoming Sandra Fauconnier, our new Structured Data community liaison

The Technical Collaboration team is very happy to welcome Sandra Fauconnier, our new community liaison focusing on the Structured Data program. Sandra will support the collaboration between the communities (Commons, Wikidata, GLAM…)  and the product development teams involved at the Wikimedia Foundation and Wikimedia Germany. The plan is to improve Wikimedia Commons allowing users to better view, translate, search, edit, curate and use media files. To achieve that, the Commons backend will be migrated to Wikibase, the same technology used for Wikidata. Many other features and pieces are part of this plan. In the near future, as the first prototypes and tests start to emerge, Sandra will also drive the engagement with new individual content contributors, existing and new GLAM organizations, and developers interested in exploring the possibilities of the new platform. You can find more details here.--Qgil-WMF (talk) 11:58, 3 July 2017 (UTC)

An excellent appoinment. Congratulations, Sandra! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:01, 3 July 2017 (UTC)
Congrats! Qqil forgot to mention the whitepaper.
--- Jura 15:41, 3 July 2017 (UTC)
Welcome on board Sandra, I'm glad to work with you :) Lea Lacroix (WMDE) (talk) 21:01, 3 July 2017 (UTC)
  • I'm not exactly sure what the intention of the current community liaison strategy happens to be. One the one hand it seems like when the Wikidata team is unable to reply to a request for input to a RfC that touches technical issues in over a month, that there's a need for more communication but it doesn't seem to me that the quantity of request at https://www.wikidata.org/wiki/Wikidata:Contact_the_development_team seems to be that big that the solution needs to be another person. ChristianKl (talk) 11:06, 4 July 2017 (UTC)
    • I fixed a link in the original announcement, maybe it is much clearer now how this position should help us. Structured data is about be introduced at Wikimedia Commons, and this involves Wikidata as Wikimedia’s structured data backbone as well. Since Commons still has more media files than we have items here at Wikidata, this will likely have remarkably larger impact on our project than the integration of all the other Wikimedia projects had until now. —MisterSynergy (talk) 11:19, 4 July 2017 (UTC)

Inheritance for query

Hello, I need some advice concerning to way to create a query. I know we have Wikidata:Request a query but there is a in my question a more deeper problem of data modeling:

  • I have a class A
  • I have a class B, subclass of class A
  • I have one item x, instance of class A
  • I have one item y, instance of class B

What I want: All instances of class A including all instances of the subclasses of class A. So I want a query which give x and y, and not only x.

Currently I solve the problem by defining y as instance of class B AND of class A. But if I consider the inheritance of the system instance/subclass I shouldn't have to add this redundant statement.

So the deeper question is now: can we agree on the inheritance property for instance/subclass relations as general principle for WD ? Snipre (talk) 13:29, 3 July 2017 (UTC)

I thought we already did?
I have discovered one little problem. I want to describe that x started to become instance of A in 1990. In 2005 it became instance of class B who is a subclass of A. But adding an end date (2005) to "instance of:A" and an start date (2005) to "instance of:B" looks like it could be misunderstood? -- Innocent bystander (talk) 13:38, 3 July 2017 (UTC)
@Innocent bystander: We never really discuss about that kind of rules so I prefer to start once the discussion and if I feel that we have some agreement, I can start a RfC in order to generate a Wikidata page grouping basic rules.
For your case I think a split: the concept is changing by passing from an individual to a group. Better to have 2 items instead of one. Snipre (talk) 14:20, 3 July 2017 (UTC)
Well, lets take Montgomery, Alabama as an example. It is an "instance of:county seat" and county seat is a "subclass of:capital". If we (by some reason) instead choose to have a specific item for "county seats in Alabama", it would be enough if Montgomery is "instance of:county seat in Alabama". You then do not have to add "instance of:county seat". But you still maybe need to add "instance of:city" next to this claim, since a capital does not have to be a city. Depending of the circumstances and subject, this is not as obvious as it maybe look here. -- Innocent bystander (talk) 14:43, 3 July 2017 (UTC)
@Innocent bystander: Your example is a bad one: With your proposed classification how do you want to query all cities of Alabama (county seat or not) ? You never have to create an item mixing several concepts (here you the concept county seat and the concept located in Alabama in the same item "county seats in Alabama"). By using this type of items you have to create a similar system for cities ("city in Alabama") or to add a statement "located in:Alabama" which will be partially redundant with "county seats in Alabama"). By creating three individual statements "instance of:county seat", "instance of: city" and "located in: Alabama", you can query all possible combinations. Snipre (talk) 15:06, 3 July 2017 (UTC)
@Snipre: My example does not have to be a bad one. In Swedish we use completely different words to describe a "county seat" in US and "county seat" in Sweden. Technically we could even name county seats in New England differently than those in California (but normally we don't), this because the a "county" looks very different in these places. Just because two things has the same name in one language does not make them the same class. -- Innocent bystander (talk) 15:41, 3 July 2017 (UTC)
@Innocent bystander: Where do you read I wrote that Swedish county seat has to be the same item as US county seat ? You can have a item for the Swedish county seat and another one for US county seat.
Just try to answer my question "how do you want to query all cities of Alabama (county seat or not) ?" to understand my point of view. If I query all items having both "instance of:county seat in Alabama" and "instance of: city", I will miss all cities of Alabama which are not a county seat. So what is your solution to query all cities (county seat or not) of Alabama ? Snipre (talk) 16:14, 3 July 2017 (UTC)
As I said, that something is a county seat does not imply that it is a city. In fact the seat of Noraström which was located close to where I live today, had its seat in a small village. You have to look into the "city-hierarchy" to find those, not the capital-hierarchy. From what I have learned, Alabama has several classes of citys (First class city, second class city etc) and probably unincorporated citys too. "city" alone is, as I tell in another thread above, not a well defined entity. Currently we probably do not have items for "First class citys in Alabama" and maybe not even "city in Alabama". If we want to describe what each "class" of citys have for obligations and rights you have to put such information in the lower levels. Here the municipality of Gotland has other rights than any other municipality in Sweden has. We therefor need to split "municipality in Sweden" into two classes, one for Gotland, and one for the rest. -- Innocent bystander (talk) 16:30, 3 July 2017 (UTC)
Perhaps another related issue is something like Sultan of Oman (Q28478447), where the role is currently head of government (Q2285706) of Oman (Q842), but expressing that needs to have a start time, as the role itself pre-dates the modern state. Should this be a subclass of (P279) on the item with a start time (P580) qualifier, similar to how the subclasses on Bishop of Sherborne (Q680033) are entered? Or should we have separate items for the same position for when it had different meanings? Or only add this to the instances for which it applies? Or something else entirely? --Oravrattas (talk) 17:46, 3 July 2017 (UTC)


@Snipre: sorry, but am I missing something? We do this routinely now with something like " SELECT ?item WHERE {?item wdt:P31/wdt:P279* "A"} ". ArthurPSmith (talk) 15:21, 3 July 2017 (UTC)
@ArthurPSmith: I don't want the subclasses of A, only the instances (the direct ones and the indirect ones), but I have to check your query to see if it works. My fear is about the process: can your query perform some inferences ? Can it deduct that if x is instance of B and B is a subclass of A, then x is an instance of A too. There is a reasoning behind that, so I just want to have a confirmation that the query you propose can do that job. Snipre (talk) 16:14, 3 July 2017 (UTC)
ArthurPSmith’s query does indeed return direct and indirect instances only, but not the classes; if I don’t miss anything, this is exactly what you were looking for in your original request. Technically this is very robust, since SPARQL provides different means to traverse the knowledge tree along given properties during a query. If you are interested in reading SPARQL specs: in general this is called a property path, and this one in particular (the SPARQL operator /) is called a sequence path. ?item wdt:P31/wdt:P279* "A" looks for items that have P31 with a value that is connected to "A" via an arbitrary number of P279 relations. This includes zero subclass relations (expressed by the asterisk *), i.e. P31 "A" directly. —MisterSynergy (talk) 16:36, 3 July 2017 (UTC)
@Snipre: "is used to state that all the instances of one class are instances of another" - Help:Basic_membership_properties#subclass_of_.28P279.29
Wikidata:A_beginner-friendly_course_for_SPARQL#Property_paths
d1g (talk) 03:14, 4 July 2017 (UTC)
  • Subclasses inherit but I don't think it makes sense to see instances as inheriting all features. ChristianKl (talk) 11:32, 4 July 2017 (UTC)

How do I describe dialects in Wikidata?

Hi all

We have now run the UNESCO Atlas of World Languages in Danger through Mix n' Match and have some manual adjustment to make by hand which I will do asap. Some of the items are dialects of another language. I will create items for each of the dialects however I'm unsure how to describe the relationship between the dialect and the 'parent' language. I feel like there should be an equivalent of 'has part' and 'part of' properties but I'm unsure how to proceed, I looked at the English dialects for reference (British English, American English, Indian English etc) for reference but it seems like this relationship hasn't been established. Any suggestions?

Thanks

--John Cummings (talk) 20:12, 2 July 2017 (UTC)

@John Cummings: has dialect (DEPRECATED) (P134)? —Justin (koavf)TCM 20:23, 2 July 2017 (UTC)
Please make sure that you have a valid code and as important a consistent code for the dialect. Without such codes there will be chaos. Thanks, GerardM (talk) 20:25, 2 July 2017 (UTC)
Thanks very much User:Koavf (talkcontribslogs), is there anything like a property for 'dialect of'?
GerardM (talkcontribslogs), of course, it seems as though some dialects have codes but some do not, I need to do some research on this before I work out what to do. --John Cummings (talk) 23:10, 2 July 2017 (UTC)
@John Cummings: Use subclass of (P279).--GZWDer (talk) 05:26, 3 July 2017 (UTC)
I think it would be worthwhile to define the standards for dialects in some Wikiproject for languages. ChristianKl (talk) 11:26, 4 July 2017 (UTC)
Agreed. Do we have enough Wikidatans with the appropriate expertise and interest to make such a project successful? - - PKM (talk) 19:35, 4 July 2017 (UTC)
We could name it "WikiProject Languages". Visite fortuitement prolongée (talk) 20:49, 4 July 2017 (UTC)
Wikidata:WikiProject Languages already exists and need to be developed. Yet, with the coming support of Wiktionary, this project will grow up. Pamputt (talk) 20:57, 4 July 2017 (UTC)

Update statistics, please

Hello, could someone please update Wikidata:Statistics and Wikidata:Statistics/Wikipedia? The pie chart is very interesting, but it is two years old... when Wikidata had 15M items, now we are almost double that... Thank you! Renata3 (talk) 22:47, 3 July 2017 (UTC)

@Renata3: You're welcome to run the Python script below the matrix of SPARQL query links here and rebuild the data in the module yourself. Mahir256 (talk) 23:45, 3 July 2017 (UTC)
I have no idea what that means... I have poked around those links, but I don't understand how they work and I am afraid to break something. My techie skills are not up to par, there. Renata3 (talk) 04:38, 4 July 2017 (UTC)
Is it possible? I thought the updates where stopped because sparql timeouts. Emijrp (talk) 05:56, 4 July 2017 (UTC)
@Mahir256: There's no Python but JavaScript to be run in a dedicated environment. Moreover, there's no SPARQL (it was enabled after the last statistics had been generated) but WDQ (which does no longer work).
To sum up, this needs to be built from scratch. Matěj Suchánek (talk) 08:01, 4 July 2017 (UTC)
@Matěj Suchánek: Ah, yes, I severely and hastily misspoke. Here is a partial rewrite of the script to obtain the data, partial only because some of the required queries (in the case of bnwiki, which I'm testing with, those for instance of (P31) Wikimedia category (Q4167836), taxon (Q16521), occurrence (Q1190554), architectural structure (Q811979), and artificial entity (Q16686448)) are timing out and I'm not entirely sure how to rewrite them. Mahir256 (talk) 04:07, 5 July 2017 (UTC)

annotate a geographic id with a "confidence value"

On our path to Australian populations, I've come across an issue. Some State Suburb areas defined by the ABS are a bit loose, and encompass a gazetted locality plus a whole lot more area, even though they are named after the locality (for example at Aberfoyle (Q10397939) click on the links to compare SSC10006 and GL_NSW9). In most cases it is still reasonable to associate the item with this ID (GLs do not come with a population, so the SSC population is all we've got). However, it would also be good to communicate the fact that this is not a perfect overlap. The way the ABS communicates this is with a "CONF_VALUE" (see [1]). I've created items corresponding to the grades they distribute: confidence value (Q31834028) with possible values very good (according to the Australian Bureau of Statistics) (Q31833890), good (Q31833911), acceptable (Q31833947), poor (Q31833979), very poor (Q31833994). But how can I associate them with the assignment of Australian Statistical Geography 2011 ID (P4014)="SSC10006"? I've looked for a qualifier to do this, but haven't found anything appropriate. It seems odd to propose a whole new qualifier for only this task. So I'm not sure what the best way forward is. I'd appreciate any advice you've got.  – The preceding unsigned comment was added by 99of9 (talk • contribs) at 4 juli 2017 kl. 13.56‎ (UTC).

Help:Ranking is tri-valued.
We also have sourcing circumstances (P1480) and stated as (P1932). d1g (talk) 17:57, 4 July 2017 (UTC)
I can't see any of those working. Ranking isn't quite right (and only has 3 values) because that's an internal assessment, and this is a citable external assessment. sourcing circumstances (P1480) has some pretty specific options, and I wouldn't want to pollute those with mine. stated as (P1932) only accept strings, and also seems to have a clear scope that I'm outside of (misprints). --99of9 (talk) 00:44, 5 July 2017 (UTC)

Wikidata:Autobiography

The help page Wikidata:Autobiography has been nominated for deletion. Please read the page, then make your views known at Wikidata:Requests for deletions#Wikidata:Autobiography. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:19, 4 July 2017 (UTC)

Please note the precursor Pigsonthewing reverting without given reason at AN as well. --Succu (talk) 19:24, 4 July 2017 (UTC)
Yes, please do. I'm sure anyone with a clue will find it most illuminating. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:10, 4 July 2017 (UTC)

Join the strategy discussion. How do our communities and content stay relevant in a changing world?

Hi!

I'm a Polish Wikipedian currently working for WMF. My task is to ensure that various online communities are aware of the movement-wide strategy discussion, and to facilitate and summarize your talk. Now, I’d like to invite you to Cycle 3 of the discussion.

Between March and May, members of many communities shared their opinions on what they want the Wikimedia movement to build or achieve. (The report written after Cycle 1 is here, and a similar report after Cycle 2 will be available soon.) At the same time, designated people did a research outside of our movement. They:

  • talked with more than 150 experts and partners from technology, knowledge, education, media, entrepreneurs, and other sectors,
  • researched potential readers and experts in places where Wikimedia projects are not well known or used,
  • researched by age group in places where Wikimedia projects are well known and used.

Now, the research conclusions are published, and Cycle 3 has begun. Our task is to discuss the identified challenges and think how we want to change or align to changes happening around us. Each week, a new challenge will be posted. The discussions will take place until the end of July. The first challenge is: How do our communities and content stay relevant in a changing world?

All of you are invited! If you want to ask a question, ping me please. You might also take a look at our the FAQ (recently changed and updated).

Thanks! SGrabarczuk (WMF) (talk) 22:09, 4 July 2017 (UTC)

Languages on Wikidata, Glottolog catalogue matching

Hi all

Currently there are 15,000 or so unmatched items which are being done manually. Because Glottolog already has links to ISO ISO 639 codes I think it may be possible to a very large amount of automajic matching to existing items which will both speed up matching and reduce errors in matching.

The only issue I can see with match items is with dialects which may shame ISO numbers with the 'main language'. I think this could be fixed with running two queries afterwards for items that share the same ISO code and for items with multiple ISO codes.

I'm talking about the UNESCO Atlas of World Languages in Danger on Wikidata at the Wikipedia Languages Conference in early July and would be really amazing to show the Glottolog catalogue as well.

Is there any way to know who added the catalogue for matching? It would be very very good to have both data about the languages on Wikidata but also include the relationships between them e.g their language families.

Also is there any way to know if all the ISO 639 (parts 1, 2 and 3) have been added to Wikidata?

Thanks

--John Cummings (talk) 13:44, 26 June 2017 (UTC)

A language code may comprise multiple parts. The first is the language code (preferably ISO-639-3). The easiest thing to do is make a list of all the different codes when the list is not too long I am happy to have a look and come up with some recommendations. Thanks, GerardM (talk) 15:21, 26 June 2017 (UTC)
Hi GerardM (talkcontribslogs), I have got a bit confused with all the different ISO 639 standards, I've uploaded all of 639-3 into a Google Doc here. I think this points to a much larger issue with Wikidata of not having a way to know if external databases have been added, if all ID numbers from and external database have been added and if so added correctly. Please let me know if there is any non technical grunt work I can do to help with this --John Cummings (talk) 08:26, 1 July 2017 (UTC)
There is an overlap between the three relevant standards but they are not the same. When you consider these standards, the ISO-639-3 is actively maintained and the most authoritative. In the WMF we practically use ISO-639-1 when it is the same as an ISO-639-3 code. There is baggage from the past in what we have and these should not be reflected as ISO codes in Wikidata. Every year there are updates from ISO-639-3 including codes that are no longer considered to be relevant. Does this help? Thanks, GerardM (talk) 08:46, 1 July 2017 (UTC)
Thanks GerardM (talkcontribslogs), that's very helpful, thanks. I still don't understand how I should proceed with checking they have all been added to Wikidata. Should I just run the table through Mix n' Match and include the correct property on the import tool to help it automatch the languages? --John Cummings (talk) 10:05, 1 July 2017 (UTC)
Hi John Cummings, there is expected completeness (P2429) meant to track completion of import for external databases, which is here at eventually complete (Q21873974), meaning it would still be a work in progress. And indeed, I was able to import about 200 of them earlier this week from wikipedia articles. To get some actual numbers, your Google spreadsheet has 7851 current entries and 319 retired ones, while wikidata has a total of 7852 entries (including 19 explicitly marked as retired), so we may not be too far from target. The missing ones can be found through a simple spreadsheet lookup or through Mix'n'Match that would allow to go one step further by actually matching them. This is however some kind of bird's view, because when you look at items in detail, they sometimes pack together sitelinks that actually describe different realities, so they may also need some additional cleaning (i.e. merger/split).
And finally, to answer one of your first questions, according to metadata the Glottolog catalog was uploaded in Mix'n'Match by User:Dhx1. --Nono314 (talk) 13:09, 1 July 2017 (UTC)
Hi Nono314 (talkcontribslogs), thanks very much for your answer, do you know how to either the spreadsheet lookup or how to put the through Mix n' Match in a way that will maximise the automatching? (Perhaps Magnus Manske (talkcontribslogs) or Pigsonthewing (talkcontribslogs) might know also?). I feel like once we have this complete then we could be cleverer with automatching much of the Glottolog catalogue. --John Cummings (talk) 15:04, 1 July 2017 (UTC)
You can add a new catalogue using this tool, though for complex cases, I reply on Magnus' goodwill ;-) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:04, 1 July 2017 (UTC)
Thanks Pigsonthewing (talkcontribslogs), I feel like there must be a way of completing the form that will make it easier for the tool to automatch since most of the items are already in Wikidata with the relevant property. I think Magnus Manske (talkcontribslogs) is is probably the person to answer this, the Mix n' Match manual doesn't go into detail about how the automatching works or how to help it do its job. --John Cummings (talk) 17:24, 1 July 2017 (UTC)

I have imported some ISO 639-3 values from Glottolog into mix'n'match, and will use it to match entries against Wikidata, but the download from Glottolog only gives me 2000 entries. Does anyone have the full list "ISO 639-3 / Glottolog ID"? --Magnus Manske (talk) 09:53, 5 July 2017 (UTC)

Update: That added ~2% matches. --Magnus Manske (talk) 09:55, 5 July 2017 (UTC)

Outlines

Hello. I see that there are many "outline" pages without an "instance of" value, for example outline of architecture (Q7112558). Should we use outline (Q1531349) as a value or should we create a "Wikimedia outline page" in the same fashion we did for categories, templates, etc? Regards. Emijrp (talk) 19:51, 4 July 2017 (UTC)

subject > instance of (P31)   < Wikimedia outline page >
would more correct for most cases.
Then we could use facet of (P1269) for every topic, as we did other Wikimedia-related items. d1g (talk) 20:24, 4 July 2017 (UTC)
Wikimedia outlines are not facets of their topics. Perhaps there should be a dedicated property linking outlines to their subjects, similar to category's main topic (P301). --Yair rand (talk) 01:12, 5 July 2017 (UTC)
main subject (P921) might work. I'm not sure there are enough of these outline pages (e.g. under 1000 on the English Wikipedia) to justify a special property for it. - Nikki (talk) 07:32, 5 July 2017 (UTC)
There is an open proposal for one: Wikidata:Property proposal/topic's main Wikimedia outline. IMHO P921 would do. --- Jura 07:40, 5 July 2017 (UTC)
Actually, that's the other way round P921 wont do for that.
--- Jura 07:43, 5 July 2017 (UTC)

I see that it exists Wikimedia outline article (Q26884324), so I am using it for "instance of". Emijrp (talk) 08:45, 5 July 2017 (UTC)

Global translation administrators

Hello,

I opened a Request for Comments on meta to have new global translation admins. I inform you because it will impact your wiki. Do not hesitate to write your opinion on the page. CreativeC38 (talk) 09:30, 6 July 2017 (UTC)

Lua error in mw.wikibase.entity.lua

Spurious Wikidata errors sometimes appear in articles. My guess is that congestion can cause a timeout so that when a page is rendered, Wikidata is not available. An example is seen at cs:Messier 96 (last edited in August 2016); the article currently shows four big red messages in the infobox. Please don't purge the page because purging just covers up the problem and makes it harder for others to find an example. The search links I left at Wikidata:Contact the development team#Lua error in mw.wikibase.entity.lua show that the problem occurs on multiple sites. Have others noticed this? Johnuniq (talk) 10:32, 6 July 2017 (UTC)

I cannot see any problem with this page now, in cs:Messier 96 I see that the LUA-code in the page demands only 0,051 seconds of CPU-time and 1,87 MB of memory. Available for the page is 10 seconds and 50 MB of memory. So I do not think there is any timeout when rendering this specific page. The problem is something else. -- Innocent bystander (talk) 10:49, 6 July 2017 (UTC)
Thanks but I'm not referring to the 10-second Lua module limit. MediaWiki attempts to render the page. In doing that, it expands a template that invokes a module. The module uses Wikidata. Scribunto (which interfaces with Lua) executes mw.wikibase.getEntity and that ends up calling the underlying PHP code with php.getEntityId. I guess that lasts step is intermittently timing out when it tries to establish a network connection with the Wikidata database. The result is that articles end up showing an error message ("Lua error in mw.wikibase.entity.lua at line 34..."). See the search links here to see that the error happens regularly. A developer is needed but I don't like Phabricator. Johnuniq (talk) 22:31, 6 July 2017 (UTC)

If an obituary states that a notable physician, that is in Wikidata, declared a person dead, say for the data point "July 4, 2017", should you source it to the obituary, or the physician, or both?

If an obituary states that a notable physician, that is in Wikidata, declared a person dead, say for the data point "July 4, 2017", should you source it to the obituary, or the physician, or both? The question pertains to all secondary sources that are quoting a primary source. --Richard Arthur Norton (1958- ) (talk) 14:29, 6 July 2017 (UTC)

Your source is the obituary. You could use statement supported by (P3680) to indicate the physician declaring him dead, if this should be relevant. - Valentina.Anitnelav (talk) 15:01, 6 July 2017 (UTC)
There should be a reliable published source stating that the physician stated the date of death. If the reliable obituary stated this, that's fine. Of course, Wikidata does not have structured data about the contents of the obituary, other than the death date, so if we desired to present as structured Wikidata information that the physician was the one who stated the death date, statement supported by (P3680) would be suitable, but statement supported by (P3680) is not a substitute for a citation to a reliable source. Jc3s5h (talk) 16:01, 6 July 2017 (UTC)
The question cannot be answered as asked. Although a declaration of death by a notable physician is probably reliable, it can only be cited if the declaration itself is published by a publisher who can be relied upon to provide a true image or transcription of the declaration. So, was the declaration published, if so how, and did the editor adding the information read it?
As for the obituary, that might be considered a primary source or a secondary source. Sources close to the event, such as an obituary in a newspaper a few days after the event, would be primary. Obituaries in reliable publications a fair amount of time after the death would be secondary sources. For example, an obituary that appeared in Nature a few weeks after the death might be considered secondary. The question is unclear about whether the obituary is the secondary source mentioned in the question, or the secondary source is something else that hasn't been described. Jc3s5h (talk) 15:29, 6 July 2017 (UTC)

Wikidata inclusion should link to article

Please make data included from Wikidata link to the appropriate page on the wiki where it's being used. For example, if {{#property:mayor}} returns Jon Doe, I'd like it to instead return [[Jon Doe (politician)|Jon Doe]]. Thank you. --Arctic.gnome (talk) 15:32, 6 July 2017 (UTC)

That is possible already. Simply replace {{#property:mayor}} with {{#statement:mayor}}. :) --Lydia Pintscher (WMDE) (talk) 15:34, 6 July 2017 (UTC)

Add support to Outreach Wiki or not?

See outreach:Wikimedia:Village pump#Add Wikidata support here or not?, needs people who are familiar with that wiki. --Liuxinyu970226 (talk) 01:02, 7 July 2017 (UTC)

Statistics, users, languages

Ιs there any statistics about wikidata users and language preferred? Xaris333 (talk) 02:33, 5 July 2017 (UTC)

@Xaris333: This exists, but @Pasleim:'s bot hasn't updated the language/skin portions since December. Mahir256 (talk) 04:15, 5 July 2017 (UTC)
The update was broken due to a change in the database replicas [2]. Now it is fixed. --Pasleim (talk) 14:19, 7 July 2017 (UTC)

text references?

According to earlier version of c:Creator:Ernest J. Rowley (Ernest Joseph Rowley (Q31202186)) birth date come "From National Archives and Records Administration; Washington, D.C.; Manifests of Passengers Arriving at St. Albans, VT, District through Canadian Pacific and Atlantic Ports, 1895-1954; National Archives Microfilm Publication: M1464; Roll: 259; Record Group Title: Records of the Immigration and Naturalization Service; Record Group Number: 85. " How do I add it to the references of date of birth (P569) ? --Jarekt (talk) 17:02, 5 July 2017 (UTC)

If you poke around at the US National Archives website, you find that the microfilm publication has been digitized by FamilySearch.org and is available here for free. You can go search for the person, look at the image, and cite the FamilySearch.org copy of the microfilm image(s). Jc3s5h (talk) 17:26, 5 July 2017 (UTC)
Jc3s5h thanks I have done that, however now I just reference FamilySearch (Q3066228) page. I wonder if there is a way to also reference the original source. The simplest way would be a text field, but maybe just placing this information on the item talk page is enough. --Jarekt (talk) 18:59, 5 July 2017 (UTC)
There's no text field and the last time a property was proposed to provide a text field the proposal was rejected. ChristianKl (talk) 19:46, 5 July 2017 (UTC)
This is probably relevant for meta:WikiCite - we really should sort out here whether the user should create an item for this, or up to what level of detail an item is warranted. ArthurPSmith (talk) 20:01, 5 July 2017 (UTC)
The usual procedure when citing a source (not just in Wikimedia projects, but in general) is to cite where you read it. You cited the web page of a reliable source, so that fulfills the minimum requirement. You could do more, if you were afraid the web page might disappear. For example, you could create an item for the database and then cite the item, adding information such as the microfilm collection number and roll number. Jc3s5h (talk) 20:48, 5 July 2017 (UTC)
  • I have updated his date of birth, and linked seven documents to his profile that I created at Familysearch. His ggrandson also has a good page of information on him. --Richard Arthur Norton (1958- ) (talk) 01:44, 6 July 2017 (UTC)
I have linked seven documents including his birth registration and his marriage certificate, all of which contradict the 1882 birth year and match the webpage data. He is correctly identified in the Canada census as a photographer. You have identified a primary document of a border crossing with the wrong date on it, maybe even the wrong person. Your judgement on what is reliable and unreliable is a bit askew. --Richard Arthur Norton (1958- ) (talk) 02:42, 6 July 2017 (UTC)
Richard Arthur Norton (1958- ) the page at Familysearch you created is not accessible unless I register and provide my name email or phone number. Can you create page where that is not necessary or provide references to publicly available sites? By the way, for last 4 years c:Creator:Ernest J. Rowley listed date of birth as 1882 and provided the the text in the original post as the source. If that date is wrong than it is better to depreciate it instead of deleting so other will not re-import it. --Jarekt (talk) 03:33, 6 July 2017 (UTC)
Excellent idea to keep and deprecate. I am also migrating documents to Wikimedia Commons. It is worth registering for Familysearch, you don't have to give your telephone, or just give a false one if it is required. So many primary documents there for free. The crossing is of him, he lists his occupation as photographer but it doesn't match the census date of birth or his marriage certificate. The other two crossing also use different years of birth, based on the age he gave. --Richard Arthur Norton (1958- ) (talk) 03:52, 6 July 2017 (UTC)
User-contributed data or self-published websites are not reliable sources. If such a site makes use of reliable primary sources to support a statement, such as a date of birth, the primary source should be cited directly rather than citing the self-published source. Jc3s5h (talk) 04:28, 6 July 2017 (UTC)
You wrote: "User-contributed data or self-published websites are not reliable sources". Actually, I believe the exact opposite is the rule, we use the exact source. Wikipedia rules do not apply here. If an obituary states that a notable physician, that is in Wikidata, declared a person dead on July 4, 2017, you should quote the obituary as the source, not the physician. I will bring the issue up below. You should only quote a primary source named in a secondary source if you have independently viewed that primary source, otherwise you may just be compounding an error or misinterpreting the secondary source. We just spent several months correcting "New York state" vs. "New York City" that were misinterpreted by editors from the secondary source using "New York". We also do not have a blacklist of websites. Wikipedia rules do not automatically apply to Wikidata. --Richard Arthur Norton (1958- ) (talk) 14:26, 6 July 2017 (UTC)
I mostly agree with Richard Arthur Norton's post of 14:26, 6 July 2017, but feel it's off-point. Of course editors should only cite sources they have read. Of course if an editor reads a summary of a primary source in a secondary source, the editor should cite the seconthdary source as opposed to citing the primary source as if the editor had read it, when the editor never read the primary source. But to cite any source, primary or secondary, it should be a reliable source. If an editor reads in a self-published website from someone who isn't an expert in the field a summary of an obituary in a reliable source, and according to the self-published summary the notable physician said the date of death was July 4, 2017, the editor should find the published obituary and cite that. Failing that, if death certificates from the jurisdiction in question are publicly available, the editor could obtain the official death certificate and cite that. Any unrecorded oral comments made by the physician are not citable because they are not published.
It's true that Wikipedia rules do not strictly apply to Wikidata, but they do represent a consensus of a large community that provides useful guidance, in the absence of any corresponding policy at Wikidata. Jc3s5h (talk) 14:46, 6 July 2017 (UTC)
  • Why don't you start a Wikipedia article on him, an article in a Canadian paper praises him and his awards. I can clip the article and load it to Wikimedia Commons so it can be used as a reference. Do you have a free Newspapers.com account through Wikipedia?
  • Before you bother with that, you should read w:Wikipedia:Verifiability and w:Wikipedia:Identifying reliable sources. Wikimedia Commons is not a reliable source because it contains user-contributed information. But there is no requirement that references be online. It's fine to cite articles in reliable sources, even if they are on paper or microfilm. Jc3s5h (talk) 04:28, 6 July 2017 (UTC)
There is no mention of "Commons" in w:Wikipedia:Verifiability or w:Wikipedia:Identifying reliable sources, if your going to say "read the Bible, all you need to know is there", please cite the chapter and verse of the passage. The image of a news article is orders of magnitude more reliable than providing a link to text, or an image, behind a paywall. --Richard Arthur Norton (1958- ) (talk) 14:15, 6 July 2017 (UTC)
Sources that are mostly made up of self-published information (which includes wikis and other user-contributed information) just don't fit the description given in w:Wikipedia:Verifiability#What counts as a reliable source and does fit the description given in Self-published sources sub-section, which is under the Sources that are usually not reliable section of the verifiability policy.
The User-generated content section of the "Identifying reliable sources" guideline specifically says that most wikis are not considered reliable.
  • Wikipedia rules do not apply to Wikidata, and there is no Supremacy Clause at Wikidata that states that if there is no specific rule at Wikidata we default to Wikipedia rules. Wikidata was about 90% sourced from Wikipedia and IMDB when it was Freebase, before Google bought it and donated it to Wikidata. Both Wikipedia itself, and IMDB, cannot be used as sources at Wikipedia. --Richard Arthur Norton (1958- ) (talk) 17:20, 6 July 2017 (UTC)
Images provided by anonymous internet users are far less reliable than images available from a reliable paywalled source, because the images provided by the anonymous internet user may have been fabricated, altered, or incorrectly attributed. Jc3s5h (talk) 14:36, 6 July 2017 (UTC)
You are going down a rabbit hole of paranoia. Every biography of a dead person in Wikipedia has an image from Wikimedia Commons and we link to that image in Wikidata. Those images may be deliberately fake, they may be misinterpreted, the dating on them may be inaccurate, they may be deliberate propaganda from political entities ... yet we still use them. Print newspapers and magazines face the same problems, except once published, it cannot be corrected except in the next edition. Wikimedia Commons can be annotated and corrected, and we do that all the time. --Richard Arthur Norton (1958- ) (talk) 17:32, 6 July 2017 (UTC)
  • If you think secondary sources are inherently reliable, here is a correction from the New York Times that was appended to the online edition of the article, of course in the paper edition the correction appeared in the corrections column on the following day: "An appraisal on Saturday about Walter Cronkite’s career included a number of errors. In some copies, it misstated the date that the Rev. Dr. Martin Luther King Jr. was killed and referred incorrectly to Mr. Cronkite’s coverage of D-Day. Dr. King was killed on April 4, 1968, not April 30. Mr. Cronkite covered the D-Day landing from a warplane; he did not storm the beaches. In addition, Neil Armstrong set foot on the moon on July 20, 1969, not July 26. “The CBS Evening News” overtook “The Huntley-Brinkley Report” on NBC in the ratings during the 1967-68 television season, not after Chet Huntley retired in 1970. A communications satellite used to relay correspondents’ reports from around the world was Telstar, not Telestar. Howard K. Smith was not one of the CBS correspondents Mr. Cronkite would turn to for reports from the field after he became anchor of “The CBS Evening News” in 1962; he left CBS before Mr. Cronkite was the anchor. Because of an editing error, the appraisal also misstated the name of the news agency for which Mr. Cronkite was Moscow bureau chief after World War II. At that time it was United Press, not United Press International." --Richard Arthur Norton (1958- ) (talk) 15:50, 7 July 2017 (UTC)

The Library of the New York Botanical Gardens

Hoi, I am involved in a project where editathons for Wikipedia are planned for both authors and species of plants. The intention is to use a similar approach as used for the "Black Lunch Table". There will be opportunities to increase the involvement of the library; think in terms of publications images etc. Thanks, GerardM (talk) 09:08, 7 July 2017 (UTC)

GerardM, where can I find more information about this? --Succu (talk) 14:38, 7 July 2017 (UTC)
Similar to the Black Lunch Table Wikidata task list the NYBG will integrate Wikidata into its editathons and NYBG Plants And people task list. This is directly thanks to Gerard! Thank you for helping me to drink the Wikidata-Kool-Aid! Best, -- Erika aka BrillLyle (talk) 14:43, 7 July 2017 (UTC)
Thanks, but part of the announcement is „are planned for both authors and species of plants“. So what is planed to do at this side (WD) of the Wikimedia universe? --Succu (talk) 20:39, 7 July 2017 (UTC)

Input for Wikimania talk

Hey folks :)

I'll be giving a talk at Wikimania that will be mostly about the current state, exciting things from the past year and what's coming up around Wikidata. Let me know if there is anything you'd like me to include. --Lydia Pintscher (WMDE) (talk) 15:37, 7 July 2017 (UTC)

Something I'm proud of is the property Carnegie Classification of Institutions of Higher Education (P2643) for American institutions of higher education, the fully-taxonomized classification items (https://angryloki.github.io/wikidata-graph-builder/?property=P279&item=Q4223026&mode=reverse), and the addition of classifications to these universities using Quick Statements and matched via Integrated Postsecondary Education Data System ID (P1771). It's a nice example with small scope. Runner1928 (talk) 15:52, 7 July 2017 (UTC)
Lydia, there are now starting to be a few WikiProjects that are integrating Wikidata in their task lists. I think it is sort of wonderful. The idea is that the task lists are query-able, so they can be broken out by gender for gender gap editathons, and by location, like what Black Lunch Table does for all of their different location-specific events. I am also putting books that CPPC has published into Wikidata, because they have collected works in multiple languages unavailable elsewhere, and are very rich bibliographic resources (and great ready-made citations, too!)
Black Lunch Table Wikidata task list
Colección Patricia Phelps de Cisneros task list
NYBG Plants And people task list
-- Best, Erika aka BrillLyle (talk) 17:58, 7 July 2017 (UTC)
Looking at the program, it looks like there will be quite a few Wikidata sessions, which is excellent! How will they all link together? (And of course, please mention Wikidata-driven infoboxes. ;-) ). Thanks. Mike Peel (talk) 01:08, 8 July 2017 (UTC)

Regexper

The Regexper external website, linked from uses of format as a regular expression (P1793), fails on both of its uses on Property:P1793 (1, 2). Is this a problem with the website or a problem with the regular expressions? Jc86035 (talk) 11:28, 3 July 2017 (UTC)

Regex101, my preferred tool, says the former has rogue characters 1, but that the latter is valid 2. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:11, 3 July 2017 (UTC)
@Pigsonthewing: I've updated the first link from the use on IMDb ID (P345), and Regexper now says it's valid. I suppose a bug could be reported in the Regexper Github repository for the second regex, but the last commit was in October. Jc86035 (talk) 15:36, 3 July 2017 (UTC)
I've added Regex101 to P1793 as a secondary formatter URL; what do folk think of making t preferred? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:45, 3 July 2017 (UTC)
@Pigsonthewing: IMO Regexper is much easier to understand, and is also more easily accessible to the colourblind. Since the code of Regexper is on Github someone could actually fork it, put it on Tool Labs (so we don't have to link to an arbitrary external website) and maybe fix the problem with link 2. Jc86035 (talk) 17:24, 8 July 2017 (UTC)

New step towards structured data for Commons is now available: federation

Hello all,

As you may know, WMF, WMDE and volunteers are working together on the structured data for Commons project. We’re currently working on a lot of technical groundwork for this project. One big part of that is allowing the use of Wikidata’s items and properties to describe media files on Commons. We call this feature federation. We have now developed the necessary code for it and you can try it out on a test system and give feedback.

We have one test wiki that represents Commons (http://structured-commons.wmflabs.org) and another one simulating Wikidata (http://federated-wikidata.wmflabs.org). You can see an example where the statements use items and properties from the faked Wikidata. Feel free to try it by adding statements to to some of the files on the test system. (You might need to create some items on http://federated-wikidata.wmflabs.org if they don’t exist yet. We have created a few for testing.)

If you have any questions or concern, please let us know. Thanks, Lea Lacroix (WMDE) (talk) 13:04, 6 July 2017 (UTC)

It looks really cool. Excellent work. Emijrp (talk) 14:30, 6 July 2017 (UTC)
I have some questions:
  • If I get it correctly, all creators of Commons files need to have a Wikidata item with this setup. Is this correct? We would likely have to adapt the notability criteria significantly.
  • How do I link a MediaInfo item (hosted at Commons) to a file? Can I change this link once it was made?
  • Why is the MediaInfo item not displayed directly within the File page?
MisterSynergy (talk) 15:07, 6 July 2017 (UTC)
  • The creators would be described by a new datatype representing a plain text name and an optional URI (to an item, filckr or Wikimedia user, etc). See phab:T127929.
  • In the final product, the MediaInfo content would not live on a separate page, but in a "slot" of the file page itself. See phab:T107595.
  • Currently we can not transclude Wikibase entries direct to a wikitext page.

--GZWDer (talk) 15:24, 6 July 2017 (UTC)

  • Not every creator of Commons files will need a Wikidata item. I don't think that would be a good idea. My plan is to have a new datatype that can take links to Wikidata items, Commons user pages, Flickr user pages, etc. The ticket for that is phabricator:T127929
  • The MediaInfo page is automatically created right now when you add the first statement to it. This will not matter in the future though. See the next point.
  • The data is not displayed directly in the file page because for now we can not mix structured data and wiki text in the same page. We are working on fixing this with a feature called Multi Content Revisions. Once we have that they will be merged into one. The ticket for that is phabricator:T107595.
Hope that helps. --Lydia Pintscher (WMDE) (talk) 15:29, 6 July 2017 (UTC)
Hah, GZWDer was slightly faster ;-) Thanks! --Lydia Pintscher (WMDE) (talk) 15:31, 6 July 2017 (UTC)
We should have a merge tool for these two identical answers ;-) Thank you both, very clear and sounds good. —MisterSynergy (talk) 15:33, 6 July 2017 (UTC)

Currently for academic papers we have 2 types of author properties: author (P50) (when item for author exists) and author name string (P2093) (plain text, when item doesn't exist). Maybe we could have a similar approach, creating an item for uploaders when they have uploaded at least 100 photos (setting a notability threshold). Having an item for those authors would make easier to change the username and target url or even adding multiple urls (I don't think Flickr is the only photo sharing service) and other contact info (like e-mail). In cases where there are several authors (who took the photo, who fixed colours, etc), should we add a qualifier with that info? Just my two cents. Emijrp (talk) 18:23, 6 July 2017 (UTC)

If Commons items have the same author we want to save that information directly. ChristianKl (talk) 16:42, 7 July 2017 (UTC)
@Emijrp: We need way lower barriers if they have valued images or dozen of quality images.
Big image/edit count sets wrong standards for newcomers in any project. d1g (talk) 13:02, 8 July 2017 (UTC)

Struggling to use quick statements - or something like it

I spent a few years doing a human breadth-first search across biographical pages on English Wikipedia - recording my efforts to the WeRelate Genealogy Wiki. From that - I've been able to generate a lot of genealogy connection claims that are not already present in wikidata. When I started I wanted to go slow - so I didn't mind adding claims one by one. Now, I'm 4000+ claims in - and I feel safe about a larger scale load. I've tried to use Quick Statements (apparently there are both 1.0 and 1.1 versions - and I have seen two different screens). I've tried to load a snipped a couple times without success. It appears that I have to get my account set up to allow volume edits and for the operation of a remote tool - but I can't see how that's supposed work. I've looked around on my account to find appropriate settings and I'm not finding what I need.

Can someone either give me guidance - or point me in the direction of appropriate documentation pages? Thanks... --Jrm03063 (talk) 23:32, 6 July 2017 (UTC)

version 1 QuickStatements (Q20084080) may still work. I have troubles with QS2 too with Firefox. d1g (talk) 23:53, 6 July 2017 (UTC)
Did you have to do anything special to enable your Wikidata account - or to get it some sort of "grant" for high volume activity? I've tried both interfaces with equally unsatisfying results... --Jrm03063 (talk) 00:34, 7 July 2017 (UTC)
@Jrm03063: Were you not given a notice to enable WiDaR? Mahir256 (talk) 00:36, 7 July 2017 (UTC)
Yes, thanks, I'm aware of that hopeless bit of documentation. I previously found my way to the link you provided several times. (FWIW - as I look at the page now - it reports "You have authorized WiDaR to edit as Jrm03063. Congratulations! You can always log out here."). I followed it and tried to understand what I needed to do with respect to Wikidata and apparently - OAuth. It seemed like I needed to go to a special page on Wikidata proper - so I did so - and only succeeded in finding a special page that indicated the list of known remote applications (included the two variants of Quick Statements). I encountered vague representations that the account has to have certain Grants to operate with Quick Statements (high volume) but it isn't clear how (or if) I can set that on my own - where I might check as much, etc. --Jrm03063 (talk) 01:31, 7 July 2017 (UTC)
@Jrm03063: You do not need to enable anything further once you've enabled WiDaR, as pretty much anything @Magnus Manske: writes can operate under your name with it enabled—in other words, you should be able to go back to QuickStatements and use it normally once you've enabled WiDaR. The list of applicable grants that you found are what WiDaR needs to be able to do under your name, i.e. stuff that you need to allow it to do and what that dialog box with the 'Allow' button will allow you to give to WiDaR.
My apologies if I'm completely misunderstanding the problem you're having. Mahir256 (talk) 02:41, 7 July 2017 (UTC)

Last time I checked there were 3 versions of QuickStatements

About 3 months ago I switched from the original to using quick_statements 2 since the original version stop working for me. Last time I check niosh fork worked fine. If you want to check QS go to c:Category:Creator templates with Wikidata link: local authority control and open one of the pages there, if the identifiers in the Authority control field are green than it means that they are on Commons but not on Wikidata, verify that identifier link work and click (+) to use QS to add it to Wikidata. That should test if the tool is working for you. --Jarekt (talk) 11:05, 7 July 2017 (UTC)

How does QS fail for you? It still worked for me, in Firefox, quite recently. Also, I note that Magnus still says of QS2: "CAUTION! The code is mostly untested and under heavy development. It might nuke Wikidata and steal your lunch. Try it on the Sandbox item (Wikidata Sandbox (Q4115189)) only, for now." Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:09, 7 July 2017 (UTC)
In March the original version stop working for me and I wrote about it here. I did not see any errors or messages just nothing happen when you press DO it button. Since than I switch my tools to use the new QuickStatements which seems to work just fine. I will check it again next time I do a batch edit. By the way where is Magnus discussing QS2? There is nothing about it on QS2 page. --Jarekt (talk) 15:24, 7 July 2017 (UTC)
Sorry, I meant to make a link: User:Magnus Manske/quick statements2. Wasn't there a toolserver outage in March? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:49, 7 July 2017 (UTC)
That statement was made last November. The status has likely changed. I am using QS2 extensively for last few months and did not see any issues. Original QS is still not working for me. Try for example this statement trying to add VIAF identifier to Jakob Suter (Q30308487). I I press "do it" button nothing happens. Andy does it work for you? --Jarekt (talk) 20:03, 7 July 2017 (UTC)
Worked frst time and instantly, for me. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:42, 7 July 2017 (UTC)
I'm glad for you. Completely worthless for me. 5000 claims ready to go. I've given fragments to the interface - which it clears then does nothing - reports nothing. --Jrm03063 (talk) 00:25, 8 July 2017 (UTC)
Do you have tab seperated data? Can you provide example data? Maybe upload it into a google sheet? ChristianKl (talk) 12:10, 8 July 2017 (UTC)
Simply complaining that QS "does nothing" does not help to solve the problem. What browser are you using? What steps (e.g. clearing cache, deleting cookies, using private browser window/ different browser/ different machine) have you taken to attempt remedy? What is the first command that it fails on? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:33, 8 July 2017 (UTC)
I can confirm that the original QS does not work for me as well, while the Niosh fork and QS V.2 works fine. It's not something browser-related or caused by some addons/userscript, since the problem persists in multiple browsers. It simply doesn't even show the authorization status (You need to authorize WiDaR to edit Wikidata...). XXN, 22:24, 8 July 2017 (UTC)
I find it very odd that the original QS works for Andy but not for me or others. It worked just fine for me before march. Luckily the workaround is to use the other 2 versions. --Jarekt (talk) 03:12, 9 July 2017 (UTC)
And my apology for being short in my previous reply. You're trying to help me, and I appreciate that. I'll review what you've offered as my time permits, and see how it goes. Thanks! --Jrm03063 (talk) 15:12, 7 July 2017 (UTC)

Do we need a conflict of interest policy?

This deletion discussion has surfaced what seems to be a need for a formal Wikidata "conflict of interest" policy. I believe the last time such a policy was proposed was in this RFC from late 2014-early 2015, which concluded with "the proposal may be good advice to follow, there is no consensus to make it a Wikidata policy." However it seems to me the lack of consensus there was because the proposal was "preemptive", i.e. before any problem had appeared. I think we now have a problem - not so much with people obviously doing things that are conflicts of interest, but because there is serious confusion about what is allowed and what is not allowed here. Formalizing a policy that clarifies particularly how Wikidata may differ from enwiki and other wikipedias in this regard would be, I think, very helpful. What do others feel about this? ArthurPSmith (talk) 13:22, 7 July 2017 (UTC)

  Support. Wikidata has matured and it's time to revisit our community's thoughts on Conflict of Interest. I don't know if it will lead to a policy, but at the very least the discussion (or essays therefrom) could form a guideline. In the end, I'd love to see something about Conflicts of Interest appear on policies and guidelines. Runner1928 (talk) 15:48, 7 July 2017 (UTC)
Design a policy around some users aversion for one specific user and hir activity does not look like a good idea at all. Please, find other way to solve those problems, s'il vouz plaît! Come back when we have more experience of COI or more specifically its disadvantages. Such a policy could potentially harm our possibility to cooperate with many sources of good data. -- Innocent bystander (talk) 15:57, 7 July 2017 (UTC)
But perhaps *not* having a policy is harmful to cooperation. If a museum wants to enter or update data about its collection, I think that's a good thing. But if they think enwiki's rules hold here then maybe they won't. Being clear about it in a way that enables "sources of good data" to help us would be a good thing, I think. ArthurPSmith (talk) 16:06, 7 July 2017 (UTC)
I'm WikiFacilitating outreach to various GLAM partners and entities who will be doing cross-project work, including image donations and redlink task lists. I recommend they include Conflict of Interest statements on all of their User pages, including Wikidata. I use one of the WM NYC Board Member's COI statement, because I think it is very well written. Here's another example. This is of course based on English Wikipedia's policies (as I am primarily a Wikipedia editor), but I think it's fine to use that one as an overriding policy. It makes it less confusing for newer editors who might be overwhelmed by the number of sister projects. Best, Erika aka BrillLyle (talk) 17:47, 7 July 2017 (UTC)
@ArthurPSmith: To be frank: Setting a COI-policy NOW after a discussion around Pigsonthewing looks like you want to ban him, not solve any site wide problems. It would be like setting a a policy against "bald, blue eyed Swedish travellers taller than 196 cm" to get rid of this user. -- Innocent bystander (talk) 06:34, 8 July 2017 (UTC)
  Support Even when the point of the policy would be just to focus on the interaction with Pigsonthewing, it's not about banning him but agreeing on the norms about how COI's should be handled. If we would formulate a specific policy, I think Pigsonthewing would likely comply with the policy afterward. Without a clear policy, it's likely that we repeat the same conflict multiple times because we disagree about the norms.
It's also worthwhile to have a policy on selfie item creation or item creation by PR and SEO folks. Having people who participate in Wikidata and who's only interest in Wikidata is to create an item about themselves or for their clients isn't positive for the health of our project even if the created items meet our broad notability criteria. ChristianKl (talk) 11:38, 8 July 2017 (UTC)
I guess there are 2 issues:
  • item creation - and for that I would support restrictions on selfie item creation as they are likely not to meet notability criteria.
  • adding properties to existing items - and there I see no need for restricting edits which would be against Wikipedia's "conflict of interest" policy.
--Jarekt (talk) 20:18, 7 July 2017 (UTC)
I suggest we restrict ourselves to the meta question asked here, and leave the details for if-or-when the answer to that is yes, and an RfC (or whatever) takes place. That said, to address your (first) specific point now that you've made it: Wikispecies (for example) encourages taxonomists to create pages about species they name, which in turn requires pages, and thus items, about themselves. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:26, 7 July 2017 (UTC)
Please give a link for your claim „Wikispecies (for example) encourages taxonomists to create pages about species they name [...]“, Mr. Mabbett. --Succu (talk) 20:32, 7 July 2017 (UTC)
I agree it could be useful to forbid more explicitly selfie item creations, that would sound pretty simple. − Pintoch (talk) 20:23, 7 July 2017 (UTC)
Things that sound pretty simple, and depend on human values and judgement, especially on Wikimedia projects, rarely are. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:38, 7 July 2017 (UTC)
do we have a rash of COI editing? if not, then why fix it? (but then i was reported to COIN for editing at a GLAM editathon, so i'm biased) Slowking4 (talk) 01:04, 8 July 2017 (UTC)
I don't think the fact, that a COI policy can be too broad, is an argument against having a COI policy. It's rather an argument to write the COI policy in a way that makes it clear that editing at a GLAM editathon doesn't produce a problem. ChristianKl (talk) 11:38, 8 July 2017 (UTC)
  Oppose with respect to Main/Property namespace.
Users should enter data about them-selfs if they would allow such information elsewhere (e.g. in their Twitter). We shouldn't restrict people over what they do or say.
We should support plurality and not to exclude Q5 to have an opinion about them-self.
We don't have many editors who create ill-factual items about living persons (or themselves).
A policy shouldn't be based on exceptions.
We remove items which make identification of individual person impossible (too few statements) routinely. A policy wouldn't help to remove overly incomplete items or to fill missing information. d1g (talk) 03:43, 8 July 2017 (UTC)
  Strong oppose rejection of local knowledge as result of "COI" and "OR" witch-hunting. I would never able to source my edits about food. We would never have a scientific article about every topic. Some knowledge is only oral or only subjective. Yes, there are edge cases where not having such policy would be difficult - but I never saw problems of this badness so far. Questions we have about COI-like matters now are solvable. d1g (talk) 07:39, 8 July 2017 (UTC)
Having a COI policy doesn't automatically mean forbidding any edits made where a COI is involved. It can be about requiring disclosure of the COI and also about forbidding people from reverting the edits of other people when the have a COI. ChristianKl (talk) 11:41, 8 July 2017 (UTC)
Commons does not have COI policy, wikidata can go the same way, also our policy could be that there are no restrictions on edits on the basis of conflict of interest. --Jarekt (talk) 03:05, 9 July 2017 (UTC)
  Support Looks like a good idea to have a policy. - Brya (talk) 06:54, 8 July 2017 (UTC)
How many times COI occurred in last million of items? Or other edits? d1g (talk) 07:16, 8 July 2017 (UTC)
@D1gggg: Given that there's no direct disclosure of COI we don't have an easy way to measure how many times it occured in the last million of items. ChristianKl (talk) 11:38, 8 July 2017 (UTC)
We can't surely call it vandalism, but we call it COI... Doesn't seem right - at least to me.
This policy is only good against terrible cases. d1g (talk) 12:26, 8 July 2017 (UTC)

  Comment I doubt that adding facts in the main namespace, even about oneself, would classify as a "conflict of interest". What is the conflict with the project's goals?[1][2] That said we should have the guidance and ability to remove any puffery from items, and that is about identifying statements that are outside the scope of a person  – The preceding unsigned comment was added by Billinghurst (talk • contribs) at 10:12, 9 July 2017‎ (UTC).

"I doubt that adding facts in the main namespace, even about oneself, would classify as a 'conflict of interest'" We have some editors who already claim that that does classify as CoI, and is not allowed, is "discouraged", or is in some way underhand. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:55, 9 July 2017 (UTC)
I understand what some are saying, and I was saying that I doubt that sheer addition of facts is CoI, hence I gave some general/usual definitions. We are not an encyclopaedia, we are data.  — billinghurst sDrewth
For clarity, there may be a conflation "vested interest"[3][4] and the first may apply, though again the first should be fine if supported by references, and not puffery; and the second, well we all have that, I have it as I want Wikisource to do well, and to have its information presented and available.  — billinghurst sDrewth 12:43, 9 July 2017 (UTC)
  1. CoI = a situation in which the concerns or aims of two different parties are incompatible.
  2. CoI = a situation in which a person is in a position to derive personal benefit from actions or decisions made in their official capacity.
  3. VI = an interest (usually in land or money held in trust) recognized as belonging to a particular person
  4. VI = a personal reason for involvement in an undertaking or situation, especially an expectation of financial or other gain.

How can i get all the Item (Q)numbers with the relevant sub-category name ?

This ta.wiki main category contains <10k articles under 32 sub-cats. I want to add label, description, etc. I learned to get the Item (Q)numbers of a category by Petscan. How can i get all the Item (Q)numbers with the relevant sub-categories name?--Info-farmer (talk) 17:52, 7 July 2017 (UTC)

With descriptioner you can add a description, but for that you need a SparQL-query. For adding labels I'm not aware of a tool. Edoderoo (talk) 20:12, 7 July 2017 (UTC)
@Info-farmer: in the petscan' first tab ("Categories") above the category name introduced there is a "Depth" option - increase it by one or more units to get pages from subcategories of your category; in "Wikidata" tab tick the "Add items, where available" option; in "Output" tab select either TSV or CSV format to download & open it in a spreadsheet software like Excel (but you can copy also the results directly from browser). To add labels/descriptions/claims you can use QuickStatements, instructions are here, the new working version is here. For processing very big batches it's recommended to request a flood flag. --XXN, 21:16, 7 July 2017 (UTC)
வழிகாட்டலுக்கு நன்றி (Thanks for your guidelines) The result gives Title, Page ID, Namespace, Size (bytes),Last change, Item but not with the relevant sub category name. While the sub-cat changes, the input data should be change slightly. --Info-farmer (talk) 04:44, 8 July 2017 (UTC)
How can i get all the data with the relevant sub-category name ?--Info-farmer (talk) 07:08, 9 July 2017 (UTC)
SELECT ?subcat ?item ?title WHERE {
   SERVICE wikibase:mwapi {
     bd:serviceParam wikibase:api "Generator" .
     bd:serviceParam wikibase:endpoint "ta.wikipedia.org" .
     bd:serviceParam mwapi:gcmtitle "பகுப்பு:தமிழ்நாடு ஊராட்சிகள்" .
     bd:serviceParam mwapi:generator "categorymembers" .
     bd:serviceParam mwapi:gcmprop "title" .
     bd:serviceParam mwapi:gcmlimit "max" .
     bd:serviceParam mwapi:gcmnamespace "14" .
     ?subcat wikibase:apiOutput mwapi:title  .
  }
  SERVICE wikibase:mwapi {
     bd:serviceParam wikibase:api "Generator" .
     bd:serviceParam wikibase:endpoint "ta.wikipedia.org" .
     bd:serviceParam mwapi:gcmtitle ?subcat .
     bd:serviceParam mwapi:generator "categorymembers" .
     bd:serviceParam mwapi:gcmprop "title" .
     bd:serviceParam mwapi:gcmlimit "max" .
     ?title wikibase:apiOutput mwapi:title  .
     ?item wikibase:apiOutputItem mwapi:item .
     ?ns wikibase:apiOutput "@ns" .    
  }  
  FILTER(?ns = "0")
}
Try it! --Pasleim (talk) 15:35, 9 July 2017 (UTC)

Template:Delete

{{Delete}} includes the text "To contest or discuss this page's deletion, please visit its talk page." However, it seems to be current custom and practice for such discussions to take place at Wikidata:Requests for deletions. Should the template be updated? How? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:03, 8 July 2017 (UTC)

I doubt that Wikidata:Requests for deletions is a good place to have longer discussions. And adding non-Q/P-pages there at all is maybe not what was intended. There is a header "Pages tagged with {{Delete}}" at the top of that page. Now "Wikidata:Autobiography" therefor is mentioned under two headers on that page. -- Innocent bystander (talk) 10:47, 9 July 2017 (UTC)

Wikidata on frwiki

I just discovered that the Infobox Biographie2 on frwiki imports all its information from Wikidata. This seems like bigger cooperation between Wikidata and frwiki than we have with the other major Wikis. Does anybody know the background of how this happened? ChristianKl (talk) 16:34, 5 July 2017 (UTC)

@ChristianKl: There is no cooperation: Infobox Biographie2 is the result of an individual action which leaded to a long conflict to define with data from WD could be used in WP:fr. Two RfC were necessary to define in which conditions data from WD and especially Infobox Biographie2 can be used.
Currently if Infobox Biographie2 is used with big problem but under several constraints, there is a important trend to add non Wikidata version of infoboxes to prevent the use of Infobox Biographie2: one of the constraints to use Infobox Biographie2 is to add this infobox only in articles without any infoboxes. If an article contains an infobox with hardcoded data, it is forbidden in practise to replace it by Infobox Biographie2.
Opponents to Wikidata use regularly complain about Infobox Biographie2 and spot limitations of infoboxes using WD to discredit this kind of tools. Snipre (talk) 17:55, 5 July 2017 (UTC)
frwp is ahead of enwp here - but we're catching up. ;-) See Wikidata:WikidataCon 2017/Submissions/Wikidata-powered infoboxes on the English Wikipedia. Thanks. Mike Peel (talk) 18:19, 5 July 2017 (UTC)
@Mike Peel: Better read again all discussion on WP:en and propose for each criticism an appropriate answer and plan to go through a RfC before to start the implementation of lua infoboxes using wikidata at large scale. I feel there is a lot of contributors who have a bad opinion of WD and they will be very reactive if they are not aware of this implementation before it starts:
see here
see here
see here
My recommendations from our experience in WP:fr
- Infoboxes have to be well documented: contributors have to be able to find which wikidata properties is used in each row of the infobox in order to be able to correct the good property if something is wrong after displaying in the infobox.
- Hardcoded data in the WP articles have to have the priority on data from wikidata: if someone want to put an different data than the one proposed by WD, it should be possible to add the data in the wikicode of the article and this data should be always displayed at the place of Wikidata.
- To pass from the old version of the infobox using only hardcoded data to the one using WD, please get the agreement of the corresponding project before any mass changes.
- Test the new infobox on a sufficient number of articles and verify each displayed infobox manullay before any large implementation. Any bad display will be used as common example to discredit the new infoboxes (bad display of colors, of text alignment,...)
- As first attempt better display only data from WD having a real source (real source = source which not coming from WP): if someone find only one data which is wrong, everyone will speak about the verifiability issues of Wikidata.
Just keep one thing in mind: every small problem, every small error, will be spotted and used to show how the new infobox is bad and not well finished. This is my experience and I hope you will avoid that kind of attacks. Snipre (talk) 11:45, 6 July 2017 (UTC)
@Snipre: What you say completely agrees with my experience. :-) Unfortunately I've been through a lot of those kind of attacks. It seems that people care much more about the details of some infoboxes than others - so if you're working on people infoboxes, then you'll have a much harder time than if you're working on, say, telescope articles. So the approach I'm taking is to work on astronomy infoboxes, to get through most of the bugs there, before going back to people infoboxes / more complex areas. Also, it really helps to work in an area where there are enthusiastic users of the template, which seems to be the case with artworks. Thanks. Mike Peel (talk) 20:23, 6 July 2017 (UTC)

On svwiki there is a Gadget that you can add, which by a javascript adds {{Faktamall biografi WD}} to every biography which does not have such a template. When the gadget was invented, it was proposed to make visible to default users. But since it has some serious flaws for older biographies, it is only visible to logged in users who has installed it. The template in question is also used in wikitext to some extent. On svwiki there is not only resistance to Wikidata-templates, it is also some resistance against any template in my articles, wikidata-supported or not. -- Innocent bystander (talk) 20:18, 5 July 2017 (UTC)

English Wikisource is now relying largely on biographical data for its Author: namespace, using WD dates of birth and death both in the header and to categories author pages by birth and death dates. They are also relying on the WD links for interwiki, the relevant WP article, Commons category, etc., as well as for resource links such as VIAF, LoC, BNF, GND, etc. Of course, utilizing WD for books has a whole series of hurdles yet to clear, but for author data, the reliance is almost entirely upon WD now, and the header templates have been revised and tested as a community to make that possible. --EncycloPetey (talk) 20:49, 6 July 2017 (UTC)
If an infobox in an article on (any) wikipedia relies on data from Wikidata and some of that data is edited on WD, does this edit show up on the Watchlist on wikipedia? Ecritures (talk) 12:06, 9 July 2017 (UTC)
You have preferences at a target wiki to show wikidata edits, they are marked with a D per the legend.  — billinghurst sDrewth 23:34, 10 July 2017 (UTC)

Wikidata on Wikipedia in Danish

I know that on the Danish wiki there are some infoboxes that are built entirely upon WikiData, like the one for tennis players. I personally like that, as updates on fast changing data (earned prizemoney, player statistics) can be changed centrally, to update all wiki's that support wikidata. We use it for tennis players on the Dutch wiki too, also for data that doesn't look so variable at first, like player-ID's at the WTA website and FED-cup site. These ID's have been altered overnight for *all* players last year, and with a bunch of editors and some hard manual work it got updated pretty quick on Wikidata, but the sites that have hardcoded templates on their wiki often still have the old and wrong IDs. Also on French wiki there is a user that used a lot of Lua for cycling related articles, to pick up stage winners of the Tour de France from WikiData. Again pretty cool, and a good example of what can be done with a centralised database that WikiData is. Edoderoo (talk) 20:31, 7 July 2017 (UTC)

Correct. We have da:Skabelon:Infoboks Wikidata person (for basic biographies) for example. but also da:Skabelon:Infoboks tennisspiller (tennisplayer)
I have been working hard on rebuilding those infoboxes, and with the help of User:Dipsacus fullonum who created the modules and templates to use them, it works kinda neat! Kind regards,  Rodejong  💬 ✉️  20:37, 7 July 2017 (UTC)
Examples:
There are still some who fight everything new, but it works, and it's getting more and more users who are adding these Wikidata into the articles. For example by removing the parameters of the manually written data, so that Wikidata shows. And when something is undesired we can either write it manually, or hide it through the switch - for example: |wikidata=alle |ingen_wikidata=religion so that the data for this parameter doesn't show. Kind regards,  Rodejong  💬 ✉️  20:53, 7 July 2017 (UTC)

Scientific papers

hello,
I'm wondering if it is real that so little or even no scientific articles are included in wikidata about some engineering fields. For example, main subject "power engineering" returns no results and "electrical engineering" gives a single article in the output. Does this mean, that the project did not reach those fields yet or am I using inappropriate main subject items? I would like to use the query if available and to include the articles to wikidata (still don't know where to find such collections) if unavailable. Here an example: query of scientific articles with main subject: electrical engineering. Is there any suggestions or explainations? --Sky xe (talk) 13:13, 6 July 2017 (UTC)

It is possible, I remember seeing bots adding mostly papers about biology, medicine, etc. Here is the sparql sorting by topic.
select ?subject ?subjectLabel (COUNT(?work) AS ?count) where {
  ?work wdt:P31 wd:Q13442814; 
        wdt:P921 ?subject.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
GROUP BY ?subject ?subjectLabel
ORDER BY DESC(?count)

Try it! Emijrp (talk) 14:22, 6 July 2017 (UTC)

IMO there are probably many articles on this topic, they just haven't been marked as such with this property. The topic classification is far from standardized in bibliographic metadata formats. − Pintoch (talk) 14:41, 6 July 2017 (UTC)
@Sky xe: Most scientific articles imported to Wikidata come from databases affiliated with the National Institutes of Health (Q390551) and are thus unlikely to have anything about electrical engineering topics. As far as I am aware there aren't active efforts to import from places such as the ACM Digital Library or IEEE Xplore (or indiscriminately from a specific journal or set of journals). Mahir256 (talk) 15:34, 6 July 2017 (UTC)

Thanks for the answers. How can we find out now if the articles about engineering, such es electrical engineering are available but cannot be queried or if those -as Mahir256 said- probably do not exist there at all? so that an effort of working on this inclusion would be of interest. If case2 = true, then one (or I) could ask the IEEE to provide a list of their digital library, since it includes more than 4.2 Million articles (main subject: electrical and electronic engineering). I hope this can be done soon. Thanks again!--Sky xe (talk) 16:26, 6 July 2017 (UTC)

@Sky xe: Maybe you can run a query with without main subject (P921). Q.Zanden questions? 18:07, 6 July 2017 (UTC)

Q.Zanden: thanks. Unfortunately this would not help at all, too large output without any match.
I need basically scientific articles about engineering, especially electrical engineering and similar fields, since I would like to try starting a project at my university to encourage people to work on wikidata. But therefore I must show them how awesome it is for science and for our department there. Therefore, a great collection with as much and as complete data as possible is desired in order to success. Any help will be appreciated! --Sky xe (talk) 18:58, 6 July 2017 (UTC)

Sky xe - Most of the scientific articles in Wikidata have been added from PubMed and so are in bio-medical and life-sciences. But that does include some engineering - for example there are over one hundred articles published in Journal of Biomechanical Engineering (Q2331335) listed in Wikidata - this sparql query lists them, and you can check similar queries for other engineering journals:
select ?work ?workLabel where {
  ?work wdt:P1433 wd:Q2331335 .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}

Try it!

or you can check for all journals (if they have been given an "instance of" "periodical literature"):
select ?journal ?journalLabel (COUNT(?work) AS ?count) where {
  ?work wdt:P1433 ?journal .
  ?journal wdt:P31 wd:Q1002697 .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
GROUP BY ?journal ?journalLabel
ORDER BY DESC(?count)

Try it!

As to adding new scientific articles - I think the main goal for Wikidata has been supporting citations from the various language wikipedias, so if a scientific article is cited by a wikipedia article, it can be added to Wikidata. I'm not sure we are up to adding the entire corpus of particular publishers to this database at this point. ArthurPSmith (talk) 19:44, 6 July 2017 (UTC)
@Sky xe, ArthurPSmith: I believe his main complaint was about the relative paucity of, rather than a complete lack of, articles from any field not directly connected to life science. With respect to the goal you mentioned, I highly doubt that all of the articles being imported by the Research Bot at the moment have been actually cited in Wikipedia articles (unless @Daniel Mietchen: can confirm this). I'm sure there would be far, far more social science, humanities, and physical science and engineering articles on Wikidata today if that goal was kept to heart. Mahir256 (talk) 21:08, 6 July 2017 (UTC)
@Sky xe, ArthurPSmith, Mahir256: Yes, most of the papers we have indexed so far have a biomedical background, for a number of reasons. Engineering articles do exist, though:
#Get all scientific articles with "engineering" in the item label
SELECT DISTINCT ?item ?itemLabel ?_PubMed_ID
WHERE
{
  ?item wdt:P31 wd:Q13442814 ;
        rdfs:label ?itemLabel .
  OPTIONAL { ?item wdt:P698 ?_PubMed_ID. }
  FILTER(CONTAINS(LCASE(?itemLabel), "engineering"))
  FILTER (LANG(?itemLabel)="en")
Try it!
We can also filter Arthur's query above a bit further:
# Scientific articles from journals with "engineering" in the title
select DISTINCT ?work ?workLabel ?journal ?journalLabel where {
  ?work wdt:P1433 ?journal .
  ?journal wdt:P31 wd:Q1002697 ;
           rdfs:label ?journalLabel .
  FILTER(CONTAINS(LCASE(?journalLabel), "engineering"))
  FILTER (LANG(?journalLabel)="en") 
  ?work rdfs:label ?workLabel .
  FILTER (LANG(?workLabel)="en") 
}
GROUP BY ?work ?workLabel ?journal ?journalLabel
Try it!
It would be straightforward to go through the results of these queries and add suitable main subject (P921) labels to subsets thereof, e.g. by way of Quick statements. I am doing this kind of tagging for a number of subjects, again mostly around biomedicine — probably the most visible example is the Zika Corpus. For instance, when gonorrhea (Q101896) came up in the news this morning, I checked what the current state is of the relevant literature (e.g. including Neisseria gonorrhoeae (Q131129)) here on Wikidata and added some more P921 tags. Once all of the obvious papers are indexed this way (i.e. those with the word in the title), I am using the relevant profiles in Finn's Scholia (e.g. Q101896 and Q131129) to find cited or citing articles, which often are on the same subject too and can be tagged accordingly. I am then using the Scholia SPARQL queries or variations thereof and queries to publication databases to explore the topic further, e.g. to look for citing or cited articles not indexed yet, or for author name string (P2093) statements that could be converted into author (P50) statements via Magnus' resolve authors tool. On that basis, I regularly come up with lists of relevant articles that I am then feeding in here by way of Research Bot, largely through Tom's Fatameh these days.
In terms of coverage of the scholarly literature cited from Wikipedia, we currently do not have smooth workflows for that, but the articles from the Scholarly article citations in Wikipedia dataset that have a PMID, PMCID or DOI have mostly been imported. --Daniel Mietchen (talk) 10:51, 7 July 2017 (UTC)
I would personally support adding papers in bulk in this way, but I don't think we have an established consensus about the question. ChristianKl (talk) 00:33, 7 July 2017 (UTC)

@ChristianKl, Daniel Mietchen, @ArthurPSmith, Mahir256:, Emijrp:
thank you a lot for the answers and sorry if I'm a bit late. First I would like to make clear what I meant by the question: I need to do some evaluation of scientific articles about engineering, electrical engineering in particular.

  • why?: because I'm thinking about encouraging students and university teachers to allow me doing a course about wikidata and its usage for (our) scientific research for the students at my university.
  • what can be done with that? quite a lot. If a researcher or a PhD student would like to publish some scientific work, wikidata can be great for the first research steps. We can get a good view of the topic we would like to write about, by e.g. counting the number of citations and the number or related papers or by filtering certain words in the labels we are interested in. And even more is possible with wikidata, similar to the well known Zika virus query. Such requests might not be answered without wikidata.
  • where the suggested queries helpful? Unfortunately no, since it has been confirmed that almost nothing about engineering is included, and even if the label contains the word "engineering", its is usually not the major subject or the core of the paper. That's what has been confirmed by running the suggested queries.
  • so my question is: why not using wikidata to help science and research in general as it can be very useful there? IMO, There is a great potential for wikidata in science and research, that should not be neglected. Again, I'm ready to take part at this inclusion and to make the start for electrical engineering related articles, but I guess that some guidance is needed, since I'm not an expert in this awesome project. Thanks a lot again! --Sky xe (talk) 01:04, 8 July 2017 (UTC)
@Sky xe: Yes, "using wikidata to help science and research in general" is certainly a good idea, and I'd be happy to help. We do have some tools that are relatively domain agnostic, e.g. Source, M.D. that allows to create items based on a DOI (P356) (as well as the biomedically oriented PubMed ID (P698) and PMCID (P932)), , whereas Scholia's arXiv-to-Quick-Statements converter does something similar based on arXiv ID (P818), and the Source, M.D. companion tool Resolve Authors can help with author disambiguation. These workflows are continuously in flux towards more automation, and your help with that would be most welcome. Perhaps you'd like to have a look at Wikidata:WikiProject Source MetaData? --Daniel Mietchen (talk) 04:32, 8 July 2017 (UTC)
I agree that this would be good. If you want to proceed it might be useful to write the IEEE an email to ask whether they are willing to have their data hosted on Wikidata where it's in CC0. The next step would be to write request for permission for a bot to add the data. Unfortunately, I don't think the outcome of a request for permissions to add those 4 million items is clear before you actually create the request. ChristianKl (talk) 11:17, 8 July 2017 (UTC)

ChristianKl, Daniel Mietchen: Very nice, I'm glad to do that and I thank you for your support. I will look forward to contact IEEE in the first step, as you suggested ChristianKl. I will ask them first if they agree to give their content to be included in wikidata, and if "yes" then:

  • I would ask them if they can provide the following details of their items if available so far: title, author(s), publisher, publication date, original language of work, published in (journal, conference etc), volume, DOI, ISBN (if more or less please let me know)
  • or the bot should filter these properties out of what we get

how do you find that?--Sky xe (talk) 21:35, 8 July 2017 (UTC)

Here is how Source MD works. Metadata for 3.8 million papers in IEEE venues is here and CC0, so we don't have to ask them for that. If you are to ask them, perhaps ask whether they'd be willing to help? We have a few of such papers already (see also Scholia), so perhaps it would be good to review that corpus and to show it to them.
#List of articles published in IEEE venues
SELECT DISTINCT ?item ?itemLabel WHERE
{
  ?item wdt:P1433 ?venue;
        rdfs:label ?itemLabel .
  ?venue wdt:P123 wd:Q131566;
        rdfs:label ?venueLabel .
}
Try it!
--Daniel Mietchen (talk) 01:33, 9 July 2017 (UTC)

Daniel Mietchen, that is nice. If they are under CC0, then it is great and the first step would be to add them to wikidata. I would want to use the MD to get them in a format that can be worked with, but I need time to read the instructions in my free time and see how it works, since I never worked with that. An other option would be that -if it is easy for one of you experts- to give me a csv file of the whole IEEE content, from which I can construct the needed format (e.g. using vba which I'm a bit familiar with) and preparing them to be added to wikidata -after permission- using a bot or by quick statements. I'm so glad that we are coming closer to have those scientific articles in wikidata and would also like to ask IEEE after that if they welcome any cooperation or like to help. Awesome, tank you all :). --Sky xe (talk) 23:31, 9 July 2017 (UTC)

@Sky xe: The APIs for CrossRef and DataCite (REST or OAI) should be useful if you want to build a tool to add papers en masse directly using that API and Pywikibot. (Not entirely sure about other DOI registrars.) Mahir256 (talk) 03:07, 10 July 2017 (UTC)

Thanks dear Mahir256, I will read the instructions you all mentioned in the discussion and hopefully I'll be able to figure it out. Otherwise I'll need to ask here again :). --Sky xe (talk) 20:15, 10 July 2017 (UTC)

American Bryological and Lichenological Society

Mrs Carolyn Wilson Harris was a member of the "Sullivant Moss Society", a society later renamed to something more officious. It was named after "William Starling Sullivant". This is a great example where we actually fail to provide proper information. Every potential way of having this information in Wikidata will be a hack. NB I hope to be proven wrong. Thanks, GerardM (talk) 18:39, 9 July 2017 (UTC)

It's similar attributing specialized forms of occupation (P106) to Carl Linnaeus (Q1043) which did not exist at that time. :) --Succu (talk) 19:01, 9 July 2017 (UTC)
How is it similar? This society had a name that referred to a person. They renamed the organisation, functionally it is the same organisation and there is no way to refer properly to the person the organisation was named after. When Carl von Linne is considered to be a bryologist, fine it is stupid but acceptable as he did the works. Thanks, GerardM (talk) 19:26, 9 July 2017 (UTC)
It's the timeframe in question when a statement is regarded as "true". Linné worked about some cryptogam (Q333458). --Succu (talk) 19:33, 9 July 2017 (UTC)
That's not an issue at all. Apply the old name using native label (P1705) (or name in native language (P1559), with relaxed constraints) with a end time (P582) qualifier and a named after (P138) qualifier for the "William Starling Sullivant" biographical item. It only becomes an issue if you try to "shoehorn" (force) P138 as a statement at item level. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:06, 9 July 2017 (UTC)
This is what I would do. In fact, I'd say we should identify two cases: (1) An item was formerly <named after>, in which case we list all "native names" each with a start and end time, named after (as appropriate), named by (as appropriate); and {2) An item is currently named after someone, in which case we list all "native names" each with a start and end time, named after (as appropriate), named by (as appropriate) but also use P138 as a statement at the item level, since it currently applies. This doesn't bother me as redundant. - PKM (talk) 20:47, 9 July 2017 (UTC)
It's easy: create a new item for the Sullivant Moss Society. Then connect it to American Bryological and Lichenological Society (Q4064245) via followed by (P156) and add named after (P138)=William Starling Sullivant (Q1343134) to the new item. --Succu (talk) 20:22, 9 July 2017 (UTC)
That is not the correct action, if - as suggested by the OP - all that has happened is a name change. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:19, 9 July 2017 (UTC)
OP, Mr. Mabbett? Please be more talkative. --Succu (talk) 21:31, 9 July 2017 (UTC)
Please Andy, you are selective in your responses and where you reply. With a new item it shows correctly on the members of the "old" organisation. It has the added benefit that with a merger the parts are individually known and it is possible to establish those who founded it. Thanks, GerardM (talk) 06:20, 10 July 2017 (UTC)
I'm saddened that you resort to ad hominem; but where have I not replied, that you would like me to? I cannot parse "With a new item it shows correctly on the members of the "old" organisation" - what does that mean? As to the founder of the organisation, we have founded by (P112). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:39, 10 July 2017 (UTC)
When you call me out for not answering your questions / concerns when I have already done so. When you think I am not aware by "founded by", it is not applicable and I use it almost every day. What can I say. You obviously do not understand the issues here. Thanks, GerardM (talk) 17:40, 10 July 2017 (UTC)

WikidataCon: What should be in the program?

Hello all,

A short update about the WikidataCon program. The call for projects is running until July 31st, and we already have a bunch of nice submissions. But we need more!

What do you think that should be in the program? What topics, issues should absolutely be discussed during the conference? What tools should be demoed? What meetups should happen?

Feel free to add your ideas on the list. You can also add more details on the talk page. Here you can also ping people who would be great for one of the topics. And of course, if you plan to attend and you feel like taking care of one of the topics, please submit it :)

If you have any question, feel free to contact the program committee.

Thanks, Lea Lacroix (WMDE) (talk) 09:15, 10 July 2017 (UTC)

@Lea Lacroix (WMDE): Perhaps an update of Wikidata:Development plan. I thing we lost the connection between the developers and the community about the next development of WD. I remember the 2 first years where new developments of WD had a strong impact on contributors work. Now it seems that the development team is working mainly in the background.
A good solution will be 1) update the current Development plan, 2) take the list of the phabricator tasks and try to put some priorities on some bugs, 3) update again the Development plan including some new tasks.
The presentation can focus on the future Development plan and on the future challenges for WD from developers point of view. I have sometimes the impression that some contributors consider the dev team as a personal service mainly because they don't now what the team is doing. I know we have the weekly report, but this report doesn't help to have a larger view on the WD development. I know that the presentation can be very technical too, but it is necessary to "sell" the work of the dev team in order to show that the problems are solved, one by one, but continuously. Snipre (talk) 12:03, 10 July 2017 (UTC)
@Lea Lacroix (WMDE): Then it can be a good idea to present the results of WikiCite 2017 and especially the results having an impact on WD or the list of problems which have to be solved in order to promote WD as a useful database for citations. This is a high risk that Wikicite people are developing their structure on their side and this will lead to some conflicts when they will start the implementation of their models. We already had a good example with the topic Wikidata:Properties_for_deletion#language_of_work_or_name_.28P407.29_and_original_language_of_work_.28P364.29 where the change of the model for books had an impact on the movies model. The communication is important especially when specialists are working together in order to avoid future conflicts with non specialists or other contributors working in different fields but using the same properties. Snipre (talk) 12:14, 10 July 2017 (UTC)
  Comment Agree with Development plan + we can have a link to something as important in side bar (maybe it should be dynamic report, not static page somewhere)
We need such information until main development efforts are done (Lemma, Commons) d1g (talk) 12:17, 10 July 2017 (UTC)
Lots of lightening talks please - to the whole audience, not as a breakout. They're an effective way to quickly disseminate news, tips and appeals for assistance. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:43, 10 July 2017 (UTC)

Wikidata weekly summary #268

Thank you for monthly updates. I suggest that, in New Properties section, if properties are divided in Authority Control IDs and others, it will help in reading and finding which new properties other than IDs (we have lot of them) are approved. Regards and keep posting,--Nizil Shah (talk) 14:36, 10 July 2017 (UTC)
Thanks for the suggestion. Asking @pasleim: would it be easy to do with your bot? Lea Lacroix (WMDE) (talk) 15:21, 10 July 2017 (UTC)

All the dates in The English Wikipedia from the year 1000 to the present.

I would like to upload all the dates in Wikipedia from the year 1000 to the present.

All 3 files are compressed .CSV (comma separated values) files. I have included examples of their content. Eventually I would like to write an article explaining their contents and relationships. It is thought that researchers might find this data valuable for historical investigative endeavors.

I do not think any other data of this type exists.

It includes 3 files (for every article there are multiple paragraphs and for every paragraph there are multiple sentences):

1) The titles database (articles.zip). 4,477,089 titles in the English Wikipedia. 75 megabytes.

Example:

article,title,countfound,datefound,dates

"EBP4Q0ZTF","Shaun_Palmer",2,01/24/2017,19

"AEP4Q0ZTB","Paul_Hoen",10,01/24/2017,4

"CEP4Q0ZTD","Mike_Manning_(actor)",6,01/24/2017,18

"OEP4Q0ZTP","Dog-kennel",0,01/24/2017,3

"QEP4Q0ZTR","Amy_Farrington",6,01/24/2017,3

"SEP4Q0ZTT","Carlon_Jeffery",63,01/24/2017,5

"UEP4Q0ZTV","Andrew_Caldwell_(actor)",19,01/24/2017,2


2) The paragraphs (paragraphs.csv). 25,778,610 paragraphs of the English Wikipedia. 186 megabytes.

Example:

article,para,order,added,dates

"IKP6V0XTJ","OBZ7V0XTP",1,01/22/2017,1

"IKP6V0XTJ","CCZ7V0XTD",4,01/22/2017,1

"IKP6V0XTJ","SCZ7V0XTT",5,01/22/2017,3

"IKP6V0XTJ","UCZ7V0XTV",6,01/22/2017,2

"IKP6V0XTJ","YCZ7V0XTZ",7,01/22/2017,2

"IKP6V0XTJ","ADZ7V0XTB",8,01/22/2017,2

3) The sentences (sentences.zip). 38,428,8710 sentences of the English Wikipedia. 447 megabytes.

Example:

article,para,numdates,start,end,startd,endd

"IKP6V0XTJ","OBZ7V0XTP",0,1,183,09/28/1362,12/19/1370

"IKP6V0XTJ","CCZ7V0XTD",0,1,278,05/14/1310,/ /

"IKP6V0XTJ","SCZ7V0XTT",0,1,173,11/04/1327,/ /

"IKP6V0XTJ","SCZ7V0XTT",0,219,104,02/08/1334,/ /

"IKP6V0XTJ","SCZ7V0XTT",0,455,55,10/31/1342,/ /

"IKP6V0XTJ","UCZ7V0XTV",0,1,187,05/24/1352,/ /

Jroehl (talk) 20:41, 10 July 2017 (UTC)

this does not seem like something Wikidata would be able to host. However, it might be suitable for Wikimedia Commons. ArthurPSmith (talk) 20:52, 10 July 2017 (UTC)

Arthur, Wikimedia Commons does not accept raw data. It only accepts pictures and videos. Surely there is a place for all the dates in Wikipedia. This has to be important to many historians. It is like no other database in the world. Jroehl (talk) 20:59, 10 July 2017 (UTC)

Wikidata does not accept raw data either, I'm afraid - it only holds structured linked data. We don't currently have a decent way of hosting raw data within the Wikimedia projects. Your best bet would be to host this somewhere else - perhaps somewhere like figshare or archive.org? Andrew Gray (talk) 21:03, 10 July 2017 (UTC)
@Jroehl: WikiCommons does have a tabular data type these days. Wikidata on the other hand isn't a place to host .csv files. ChristianKl (talk) 21:55, 10 July 2017 (UTC)
@Jroehl: please document how CSV were created, then host them at commons with description or with scripts to re-create CSV. d1g (talk) 06:16, 11 July 2017 (UTC)
CSV files are not allowed at Commons, and tabular data allows page sizes up to 2M. Commons is probably not going to host these files either. —MisterSynergy (talk) 06:30, 11 July 2017 (UTC)

So calling this www.wikiDATA.org is a bit misleading. I would really like to share our hard work with other Wikipedians. We will look for an external place to share these files. Thank you for helping me in a timely manner. Jroehl (talk) 22:42, 10 July 2017 (UTC)

  • If the dates can be mapped to one of our time properties (Special:ListProperties/time), some might be suitable for inclusion on specific items. One of the articles you mention is w:Carlon Jeffery with 5 dates. This corresponds to Q1189654. It already has date of birth. Looking at w:Carlon Jeffery, I'm not sure what the other 4 would relate to. Maybe the dates of publication of the references and a duplicate date of birth.
    --- Jura 23:02, 10 July 2017 (UTC)

Deleting of item

Can an administrator delete the item Mihai Nedeff, which I created, not knowing that Mihai Nedef already exists? They are one and the same person. Thanks in advance. Mycomp (talk) 08:14, 16 July 2017 (UTC)

Deletion isn't necessary.
One can merge items using several tools, for example Special:MergeItems d1g (talk) 09:34, 16 July 2017 (UTC)
Duly merged Q33083182 to Mihai Nedef (Q6845092). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:19, 16 July 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 07:49, 17 July 2017 (UTC)

We need to improve the use of the "deprecated" rank

On Phabricator, Lucas made the observation:

there are only 50k deprecated statements on Wikidata overall (compared to 160M normal ones and 500k preferred ones)

and:

It doesn’t help either that only ~400 of those statements have a reason for deprecation (P2241) qualifier. (In fact, there are more normal-rank statements with "reason for deprecation" than actually deprecated ones.)

linking to this query:

SELECT ?rank (COUNT(*) AS ?count) WHERE {
  ?statement wikibase:rank ?rank;
             pq:P2241 ?reason.
}
GROUP BY ?rank
ORDER BY DESC(?count)

Try it!

I've also seen people deleting values which should have been marked as deprecated. How can we promote and increase the correct use of deprecation? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:07, 5 July 2017 (UTC)

Good question, yet I don’t have the answer as well. Some thoughts:
  • Data removal instead of deprecation is often done by experienced Wikipedia editors who are not aware of the fact that the Wikidata web frontend has a different role than the Wikipedia web frontend has. At Wikidata, data users typically do not use the web frontend (they use Query Service or Wikipedia templates/modules), thus the web frontend is basically a tool for editors only and it is perfectly fine and desirable to have “wrong” or “multiple” values in items, if ranks and sources are properly used. In contrast, Wikipedia readers use the same frontend as Wikipedia editors do, which implies to remove “wrong” values from articles in most cases. We somehow need to raise awareness for this difference, in our guidelines as well as in talks, introductions, presentations, posters, and so on.
  • I find it quite complicated to use the reason for deprecation (P2241) qualifier. There is way too little guidance, I can just use any item as a “reason for deprecation”. However, many reasons do not exist as items yet (query reasons in use), and their meaning is in many cases exclusively expressed by labels, not statements right now. I understand the desire to use this qualifier more often, but it really isn’t intuitive right now.
MisterSynergy (talk) 10:29, 5 July 2017 (UTC)
Semi-related: I just found this interesting comment about reason for deprecation (P2241) as qualifiers via Property talk:P2241#How to use. Very convincing IMHO, we shouldn’t use P2241 as qualfiers any longer and invent another solution. However, this does not solve the original question raised by Pigsonthewing. —MisterSynergy (talk) 11:16, 5 July 2017 (UTC)
From Markus' answer, I understand that reason for deprecation (P2241) should be used along the reference stating the statement is deprecated and not be a part of the statement (as a qualifier like it's mostly used). --Melderick (talk) 17:49, 5 July 2017 (UTC)
Your later comment notwithstanding (since we have neither agreement nor an alternative solution, yet), I've just started Help:Deprecation, using some of the above, text. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:25, 5 July 2017 (UTC)
I like Help:Deprecation and added some thoughts there. I think we need a few more examples of perfectly done depreciation in example section. Please check date of death (P570) of Honoré de Balzac (Q9711), which I used as an example: can that one be improved? We also need a discussion and examples when to depreciate and when to delete some values. Looking at reasons for deprecation in use some strike me as a good candidates for deletion or correction, for example link rot (Q1193907), citation needed (Q3544030), refers to different subject (Q28091153), source known to be unreliable (Q22979588) or duplicate (Q1263068). I did not look at the current items using those reasons for deprecation but imagined future use. --Jarekt (talk) 13:20, 5 July 2017 (UTC)
It seems to me that the question of when statements should be depricated is a policy question we should therefore try to find consensus through a RfC. Having a help page that isn't a policy, isn't the way to go. ChristianKl (talk) 13:45, 5 July 2017 (UTC)
That's ridiculous. We don't have RfCs for all (for any?) of our other "Help:" pages. Indeed, if you're so keen on policies, please point to the policy requiring us to have one. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:52, 5 July 2017 (UTC)
I agree with Christian because if it is not a policy accepted by majority or consensus, it is for the wd-community unclear what the value of {{Policy}} is, because the text in the template says "It is a widely accepted standard that all editors should normally follow". If there is no consensus or majority then the widely accepted standard would not be true. Q.Zanden questions? 14:07, 5 July 2017 (UTC)
I disagree: this is not about when statements should be deprecated. This page should raise awareness among users why and how to use deprecated rank; we can discuss whether this topic deserves its own page Help:Deprecation or an expansion of Help:Ranking#Deprecated rank, but that’s it. I don’t see new or changed policies here, it is a clarification of the scarce description that we have until now. —MisterSynergy (talk) 14:20, 5 July 2017 (UTC)
Currently, this page uses the word "should" multiple times. Both the words "should" and "must" give the impression of voicing policy. A help page, where statments aren't signed and thus aren't marked as personal opinion, give the impression that it presents official policy when it uses those words.
Given that the it starts with "Property values in Wikidata should be ranked as deprecated (and not removed) if they..." saying that the page is not about when statements should be deprecated ignores the content of the page. Saying 'Property value' instead of 'statement' is bad, but it doesn't change the substance of the claim.ChristianKl (talk) 14:51, 5 July 2017 (UTC)
"both the words 'should' and 'must' give the impression of voicing policy" No, they do not. Policy would be expressed by the presence of the {{Policy}} headier. From memory, this is the fourth time I've told you this, in two days. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:58, 5 July 2017 (UTC)
(edit conflict) I also think we should brainstorm about proper use of depreciation and best practices related to it, than write Help:Deprecation and than if there is a need have a policy about it. To me the purpose of the policy is to prohibit some activities or behavior and than when someone engage in it we do not need a separate discussion about it only check if the user did something against our policy to revert their edits or punish. I can not think of undesirable actions related to depreciation, that are likely to occur. --Jarekt (talk) 15:13, 5 July 2017 (UTC)
STOP with this bad habit to create a help for each problem: nobody knows which is the correct policy when contradictions appears between help pages. Why do we need a Help:Deprecation page when we have a Help:Rank. Can we have everything on the same page in order to have an unique reference ? Snipre (talk) 18:05, 5 July 2017 (UTC)

And now we have an edit war at Kinnaird Head Lighthouse (Q2514051), where User:Jura1 thinks that the attributes of a lighthouse, such as focal height (P2923) and lighthouse range (P2929), are "deprecated" when a new one opens alongside it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:51, 5 July 2017 (UTC)

Kinnaird Head Lighthouse (Q2514051) is the historical lighthouse, whereas New Kinnaird Head Lighthouse (Q31317078) is the new one. According to the english article, the known values for focal height (P2923) and lighthouse range (P2929) are those for the new lighthouse. So it seems logical to not have them on Kinnaird Head Lighthouse (Q2514051). Doesn't seem related at all to deprecation, it's more of a cleanup operation. --Melderick (talk) 17:40, 5 July 2017 (UTC)

The origin of the problem is the definition of deprecated rank: instead of creating complex structure like use of qualifier better provide the rules to deprecate a data. A rule can be one sentence and this can prevent the addition of hundreds of qualifiers so perhaps we can simplify this mess by acting smartly. Then please concentrate the information in one place and don't create new help pages or wikidata pages: people shouldn't read 10 help pages to understand how to use WD. Snipre (talk) 18:18, 5 July 2017 (UTC)

I agree with Snipre that Help:Deprecation should be merged with Help:Rank, they are too similar. --Jarekt (talk) 19:07, 5 July 2017 (UTC)

Well, if the source to a "wrong" data is "imported from Wikipedia" or some obvious user/bot-mistake (here or on Wikipedia et al), I think it is misleading to use "deprecated rank". No "real" source has ever told that this claim is true. I then prefer to remove, or rather replace the claim. In many cases the claim is in wrong item, and I then prefer to move it to the right item. (like the "sister city"-claims in Malmö (Q2211) who should be in another item with the same label.) Deprecated rank to me is more when I have a reliable source who is wrong or in dispute. I sometimes use P2241, but only when the reason is less obvious than that I have a later revised source than the original. That the deprecated rank is so little used, is maybe because few is interested in adding claims that will (almost) never be used? -- Innocent bystander (talk) 19:57, 5 July 2017 (UTC)

Deprecated rank is only for things we first thought (had sourced) to be true, but turned out to be not true (sourced again). For example a painting first thought to be by Rembrandt, but after research turned out to be by one of his pupils. Deprecated is *not* for statements that were correct at some point in time, but are no longer valid. For example Barack Obama being president of the USA is not deprecated, it's just a statement that should be qualified with a start and end date. Unfortunately I often see this being mixed up. Any people who don't agree with this? Multichill (talk) 20:38, 5 July 2017 (UTC)

Agree, a typical example for when I use deprecated rank is when I have two reliable sources claiming two different population (P1082) for one specific date. -- Innocent bystander (talk) 06:38, 6 July 2017 (UTC)
I agree with the main idea, but I'm not sure that time is the key factor. Let's say I research when John Doe died. A serious source A says he died in 1830. I add that information with the source. Later I find another serious source B that says he died in 1832. Given my reading of the situation I think that A is right and B is wrong. Even through I always believed that A was right, it can still make sense to add B and put the deprecated rank on B. If B is just a Geni page by a random person I'm unlikely to add a link but if B is a published book I'm more likely to add the data in the deprecated form.ChristianKl (talk) 18:52, 6 July 2017 (UTC)
Multichill and ChristianKl, I think we all are agreeing here. If multiple reliable sources list conflicting information but we have reasons to believe that one is wrong we keep them but depreciate one. If both are equally reliable than I guess we keep both. One thing I sometimes see is that we have one unsourced fact imported from one project and conflicting unsourced fact copied from other project. Or maybe one has a source and other does not but seems like it come from article with some references but no indication of which piece of info come from where. For example Alfred Wierusz-Kowalski (Q358110) date of death (P570) is referenced as "imported from Russian Wikipedia" which is in conflict with the date on Commons (c:Creator:Alfred Wierusz-Kowalski ) which is also unsourced. I guess that in that case one should research the issue and find sources, but if I can not find any ( I did not look in case of Q358110) than is it OK to bring the second conflicting uncourced date? --Jarekt (talk) 02:09, 7 July 2017 (UTC) --Jarekt (talk) 02:09, 7 July 2017 (UTC)
@Jarekt: I also think we are agreeing but when it comes to writing down the agreeing it's worthwhile to be clear about what it is. ChristianKl (talk) 12:41, 7 July 2017 (UTC)

Multichill, ChristianKl and Andy Mabbett, Reviving here an dying discussion. I have a dilemma related to Frederick Gutekunst (Q1452890), We have tons of references reporting year or birth and death and than we have a photo of the gravestone at Find A Grave (Q63056) with the actual dates. More precise date is better, but should we delete the year dates, depreciate them or mark day dates as preferred. I see that situation a lot. In this case I would just delete the year dates, but if we did not have such strong references on the day dates I would keep both and decide which one to mark as preferred based on strength of references. What are your views?--Jarekt (talk) 22:38, 10 July 2017 (UTC)

Another example Henry Benbridge (Q5718019) with well references year death date and purely referenced month death date. Not sure which one should be preferred. --Jarekt (talk) 22:51, 10 July 2017 (UTC)
I don't think there's a good reason to depreciate or delete the imprecise dates. Having that information can be useful for a person who has doubts about the FindAGrave data. I think giving the more precise statement "preferred status" makes sense. ChristianKl (talk) 10:15, 11 July 2017 (UTC)

Q32000000

is here--GZWDer (talk) 15:11, 5 July 2017 (UTC)

The first graph here worries me quite a bit. --Lydia Pintscher (WMDE) (talk) 07:51, 6 July 2017 (UTC)
Why so? —MisterSynergy (talk) 07:57, 6 July 2017 (UTC)
We've increased the number of items by 1.5 Million within only a few days. This is a much larger increase than we've had since we start tracking this in grafana. And over the last quarter the number of editors has stagnated or gone down slightly. That is not a healthy dynamic for the project. --Lydia Pintscher (WMDE) (talk) 08:04, 6 July 2017 (UTC)
There is the problem with duplicates that keep getting created: Wikidata:Administrators'_noticeboard#Duplicate_creation_by_bot_account, but apparently people don't mind .. or end up not maintaining them.
--- Jura 08:50, 6 July 2017 (UTC)
Do we have a clue where these new items come from? Thanks, GerardM (talk) 11:16, 6 July 2017 (UTC)
At least Punta Barrica (Q32000000) is a bot-created article on Cebuano Wikipedia (Q837615), imported from Geonames, which in turn imported that one from GEOnet Names Server (Q1194038). As there are still a lot of bot-created articles on ceb without Wikidata link, there might be potential for lot more to come... Ahoerstemeier (talk) 12:42, 6 July 2017 (UTC)
@Ahoerstemeier: The right side of this graph seems to track Mr.Ibrahembot's progress pretty well. Mahir256 (talk) 00:35, 7 July 2017 (UTC)
I have just been pinged about a delay of changes arriving on Wikipedia's recent changes and watchlists. The dispatching lag here is getting significantly high: https://grafana.wikimedia.org/dashboard/db/wikidata-dispatch?refresh=1m&orgId=1 Based on the timing I believe this is related and pretty bad because Wikipedia editors are not able to see changes relevant to them. --Lydia Pintscher (WMDE) (talk) 13:21, 6 July 2017 (UTC)
@Lydia Pintscher (WMDE): Pinging Mr. Ibrahem, whose bot is running at 300 edits per minute. It might help to slow it down. Jc86035 (talk) 05:32, 7 July 2017 (UTC)
@Mr. Ibrahem: The bot is also introducing errors which may be from the GeoNames database. The coordinates of Motuhaua Rock (Q32283472) (instance of island) are within the North Island of New Zealand and are located in a forest. Jc86035 (talk) 08:56, 7 July 2017 (UTC)
And the coordinates for Motuharakeke Island (Q32283451) (instance of island) are in the ocean. Jc86035 (talk) 09:09, 7 July 2017 (UTC)
Interestingly GNS had the coordinates right, when it was imported to geonames it was moved into the ocean, then that coordinate was used to import to ceb, and then came here. As I said in a thread above, I don't like the idea of mass-importing from geonames as the quality of data is not good. Ahoerstemeier (talk) 10:00, 7 July 2017 (UTC)
Geonames' rounding is a problem. It makes it hard to match these with other items or sources, especially when P31 is vague and the name identical. Makes me wonder how they handle this in svwiki.
--- Jura 11:19, 7 July 2017 (UTC)
@Jura1: How much does Geonames round coordinates? If there are more accurate (and freely licensed) coordinates available then maybe they could be reimported to cebwiki and svwiki. Jc86035 (talk) 17:16, 8 July 2017 (UTC)
I'm not sure. Besides the problem at svwiki, the problem here is mainly with items and coordinates for such places already created. If one wants to create an item for a place one can identify with fairly accurate coordinates what should one do? Add it to one or several partially matching items based on svwiki? Create a new item? If one creates it, should one remove coordinates and statements from such "partial duplicates"?
--- Jura 18:09, 8 July 2017 (UTC)
@Lydia Pintscher: looks like 2 million in three months (not within days). - Brya (talk) 10:48, 7 July 2017 (UTC)
Looks to me like 2 million since 2017-05-19, 1.5 million since 2017-06-24 and 1 million since 2017-06-29. Thincat (talk) 11:08, 7 July 2017 (UTC)
Correct. Which as I said is a much larger increase than we've ever had since we started tracking it in grafana. --Lydia Pintscher (WMDE) (talk) 15:09, 7 July 2017 (UTC)
OK, I misread the graph. I apologize. - Brya (talk) 09:11, 8 July 2017 (UTC)
I stopped my bot from importing new items or improving other items for a time --Mr. Ibrahem (talk) 12:02, 7 July 2017 (UTC)
My bot made 543179 items from cebwiki. --Mr. Ibrahem (talk) 12:09, 7 July 2017 (UTC)
Thank you. --Lydia Pintscher (WMDE) (talk) 15:09, 7 July 2017 (UTC)
Re-examining the graph I am worried too. Perhaps it is time to revisit the notablity criteria. The "an instance of a clearly identifiable conceptual or material entity. The entity must be notable, in the sense that it can be described using serious and publicly available references." leaves a lot of room to include anything mentioned in "serious and publicly available references". And some users do just that. - Brya (talk) 06:57, 9 July 2017 (UTC)
Most actually have a sitelink. It's just that he hasn't done it since January. Personally, I think the regular creation of such items is a good thing. The question is mainly how. Last year, we excluded templates with a single sitelink, as they didn't add much and ended up filling up deletion requests. Personally, I think the main problem is that many items don't include any statements at all, but apparently this is a concern that isn't shared (Wikidata:Project_chat/Archive/2017/06#Bulk_create_items_without_statements.2C_but_only_a_single_sitelink_.3F) and that some items are deliberately created duplicates (apparently this isn't seen as a problem either: Wikidata:Administrators'_noticeboard#Duplicate_creation_by_bot_account and the operator assumes that someone else cleans it up).
--- Jura 08:50, 9 July 2017 (UTC)
There is a good reason why the number of items without a statements is not really a problem. Check the statistics the percentage of items without statements has never been this low. When you look at the statistics, Wikidata is growing steadily in quality. It takes time and there is much to be done. The fact that 18.55% of our items has 10 or more statements is exciting! Thanks, GerardM (talk) 09:00, 9 July 2017 (UTC)
Maybe some contributors actually tried to improve the situation. Notably nlwiki that had the bot exclude it from creating new items. Personally, I don't think your interpretation of that chart is correct. Many items are likely to have only 1 or 2 statements and items with sitelinks to Wikipedia are different than those to other wikis.
--- Jura 09:15, 9 July 2017 (UTC)
@Lydia Pintscher (WMDE): Can you explain your concern? You are product manager for Wikidata, if you say that you are worried for a big increase of numbers of items, the first thing that I think is that we may have a technical problem. There is some constraint about number of item or number of item per period? If the answer is no, I don't see problem, if the answer is yes, you must explicit the constraint. --ValterVB (talk) 09:23, 9 July 2017 (UTC)
As I have said above I am worried for social and technical reasons. Social: We are growing the number of items in a way that is absolutely out of step with the number of editors. Technical: It leads to a backlog in dispatching changes to the other Wikimedia projects. Changes from Wikidata do only show up with a considerable delay in their watchlist and recent changes. This makes the editors there unable to help keep an eye on what is going on to their content. --Lydia Pintscher (WMDE) (talk) 09:28, 9 July 2017 (UTC)

@GZWDer: The growth does not stop and it mostly is coming from you now :( --Lydia Pintscher (WMDE) (talk) 09:28, 9 July 2017 (UTC)

@Lydia Pintscher (WMDE): I have asked GZWDer how many categories are still missing in Wikidata, his new items are mainly categories. Will the backlog stay even if we stop creating new items? Are 32,000,000 items a bottleneck with the current technical resources? Emijrp (talk) 10:18, 9 July 2017 (UTC)
Thank you! The backlog will go down. You can see the current state of it here: https://grafana.wikimedia.org/dashboard/db/wikidata-dispatch?refresh=1m&orgId=1 --Lydia Pintscher (WMDE) (talk) 10:21, 9 July 2017 (UTC)
User:GZWDer (flood) uses pywikibot with PAWS. Doesn’t pywikibot have mechanisms to throttle down if the servers are uncomfortable with the edit rate? I’m asking because the bot runs at a surprisingly high rate for PAWS, and there is code in several of it’s scripts (code here) such as
def empty(lagtime):
    pass

pywikibot.throttle.time.sleep=empty
which seems to circumvent throttle mechanisms by overwriting a built-in function with a new empty one. The median lag is steadily above 60 right now (api call), but the bot continues to create items at a high rate. However, I am not yet experienced enough to tell whether this is still in line with our bot policy, so pywikibot experts please have a look at this and take action, if required. —MisterSynergy (talk) 10:52, 9 July 2017 (UTC)
pywikibot backs off if an API request is denied due to rate limit, and sleeps for the necessary amount of time before retrying the request – but since this account has a bot flag, the API doesn’t throttle it, and so it chugs along at the maximum rate that the PAWS server supports. (I think.) —Galaktos (talk) 14:27, 9 July 2017 (UTC)
The throttle logic is in Pywikibot to make sure bots are not editing too fast and shouldn't be tampered with.
The account was still creating a lot of new items causing the lag to rise even more. I disabled the bot. Multichill (talk) 16:54, 9 July 2017 (UTC)
@Multichill: Pywikibot have a default throttle (10s) even if there're no other bots running, which means at most 8640 new items per day. I may try to set up another throttle if needed. In comparison, PetScan does not have a throttle by default. About dispatch lag: per Special:DispatchStats, most of the lag is about cebwiki, which I never do anything about it. (the related edits is by Mr.Ibrahembot)--GZWDer (talk) 17:07, 9 July 2017 (UTC)
So instead of setting the put_throttle in your user-config.py you completely removed it? You do realize this breaks maxlag throttle?
You seem to be running multiple Petscan instances at the same time. You've been told multiple times not to do that and also blocked for it several times. Why do you keep doing this? Multichill (talk) 17:47, 9 July 2017 (UTC)
@Multichill: I will run only one Petscan instances at one time. Does anyone know how can I set put_throttle in PAWS? Yes currently the removal of the throttle is only a hack. I am trying to set -pt:1 but failed.--GZWDer (talk) 18:03, 9 July 2017 (UTC)
By the way, previously I also monitor Special:DispatchStats and job queue size when I runs multiple Petscan instances. When the job queue grows to 1M jobs I reduced the running threads (not only instances) of Petscan to 1 and later stop it, but this does not make job queue better (Mr.Ibrahembot is also running, with 200+ edit/m at that time). So I restarted Petscan instances.--GZWDer (talk) 18:08, 9 July 2017 (UTC)
@GZWDer: unblocked the account. Please go slow as long as the dispatch stats are lagging. Multichill (talk) 20:28, 10 July 2017 (UTC)
@Lydia Pintscher (WMDE) and others: there wasn’t so much bot activity going on today as far as I can see, but the dispatch lags have not improved according to the Grafana charts. How long do you think it will take to catch up? —MisterSynergy (talk) 19:31, 10 July 2017 (UTC)
It will unfortunately still grow for a while even after the edits are stopped. I don't know how long it will take to get back to normal unfortunately. It depends on the kinds of edits that were made and how many pages they affect on how many wikis. A change to a label for example could affect 1 or Millions of articles. -- – The preceding unsigned comment was added by Lydia Pintscher (WMDE) (talk • contribs) at 01:54, 11 July 2017 (UTC).(UTC)

@GZWDer (flood), GZWDer, Multichill, Lydia Pintscher (WMDE): "GZWDer (flood)" is now creating 100 items/minute with PetScan. At this rate, we will reach 33,000,000 in 10 hours. What the...? Emijrp (talk) 22:32, 11 July 2017 (UTC)

For admins: I'm only running one PetScan instance. I will complete current task in one hours and will have no further edit until ~6:00 UTC. If there're any issue please comment in this period.
@Emijrp: After completing current task all Wikipedias with 100000+ articles except ceb, nl and zh-min-nan (56 wikis in total) will be finished.

--GZWDer (talk) 22:41, 11 July 2017 (UTC)

...and tens of thousands of duplicated items were created. At least these could be avoided. Do the creator of these items plan to spend some time working on duplicate detection and merging? --XXN, 16:40, 12 July 2017 (UTC)

Need more flexible Formatter URL

Many external databases hold several types of entities, eg artworks and artists. Since FormatterURL takes only one placeholder $1, one can use a single external-id for the different kinds only if "type/id" come next to each other in the URL (eg "artwork/1" vs "artist/2"). Often that is not the case, which leads to a proliferation of external-id properties, or giving up on some of the less important kind of entities. For example:

So I'd like to plead again that we need a replace transformation in FormatterURL, or at least two replace fields. Eg for Atheneum a FormatterURL like http://www.the-athenaeum.org/$1/detail.php?id=$2 would do the trick, where $1 is id type (art/people/sources) and $2 is the id. That would buy us 3 external-id props at the price of one :-) --Vladimir Alexiev (talk) 15:13, 9 July 2017 (UTC)

according to Phabricator:T151329 a Lua module will be provided in future to handle such cases. --Pasleim (talk) 15:46, 9 July 2017 (UTC)
Why do we need to avoid having multiple properties? There isn't a limit on the number of properties we can have. I think inventing ways to store multiple pieces of information in a single string is a bad idea. It complicates things which creates unnecessary problems and makes the data harder to use. It does not even accurately represent the real data ("type/id" is not really the ID, it's a Wikidata invention). Separate properties have benefits too, such as better constraints. - Nikki (talk) 19:56, 9 July 2017 (UTC)
@Nikki: Because the fewer things to deal with, the better: "Occam's Razor". "Complicates": prove this (if Mix-n-Match had a split by kind of entity that was applied on the WD side I would ageee, but it does not). Not sure what you mean by "really": most sites don't have a definition of their IDs, so whatever stable means we find convenient to identify one of their items, we can use --Vladimir Alexiev (talk) 14:43, 11 July 2017 (UTC)
I agree: „"type/id" is not really the ID, it's a Wikidata invention“. --Succu (talk) 20:07, 9 July 2017 (UTC)
There might be uses for this outside of "type" - for example Dictionary of Welsh Biography ID (P1648) has language hardcoded into the identifier (either "c" or "s" as first letter, Welsh or English content; they all currently use English). If we had a more complex formatter URL we could have one placeholder for the variable and one for the desired language. Andrew Gray (talk) 12:25, 10 July 2017 (UTC)

WikiArt

T.seppelt (talk) 21:00, 18 February 2016 (UTC) Vladimir Alexiev (talk) 11:59, 13 March 2017 (UTC) GerardM (talk) 15:58, 26 March 2017 (UTC) Jonathan Groß (talk) 17:52, 26 March 2017 (UTC) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits Jneubert (talk) 13:47, 29 April 2017 (UTC) Framawiki (please notify !) (talk) BrillLyle (talk) 10:09, 10 July 2017 (UTC) Sic19 (talk) 20:42, 12 July 2017 (UTC) Wikidelo (talk) 21:15, 8 May 2018 (UTC) salgo60 Salgo60 (talk) 07:09, 10 June 2018 (UTC) ArthurPSmith (talk) 19:52, 22 August 2018 (UTC) PKM (talk) 19:40, 23 August 2018 (UTC) Ettorerizza (talk) 06:44, 8 October 2018 (UTC) Fuzheado (talk) 03:47, 19 December 2018 (UTC) Daniel Mietchen (talk) 16:30, 7 April 2019 (UTC) Eihel (talk) 15:13, 19 June 2019 (UTC) NAH Iwan.Aucamp

  Notified participants of WikiProject Authority control

Beat Estermann Vladimir Alexiev Ilya Sadads Astinson Strakhov Zeromonk Spinster Wittylama Daniel Mietchen Susannaanas Sic19 Jason.nlw Carlojoseph14 YULdigitalpreservation MB-one Ouvrard MartinPoulter Missvain VIGNERON Ainali Birk Weiberg Pmt Mauricio V. Genta Smallison ProtoplasmaKid 2le2im-bdc Rodrigo Tetsuo Argenton Ivanhercaz VisbyStar Patafisik Beireke1 Vahur Puik Ettorerizza Sp!ros Alexmar983 Epìdosis Buccalon Mrtngrsbch Eothan Giaccai NAH

  Notified participants of WikiProject Cultural heritage

Multichill (talk) 11:28, 8 August 2014 (UTC), focus on the Netherlands Husky (talk) 11:38, 8 August 2014 (UTC) - Cool, i'd like to focus on building tools to visualise progress. Spinster (talk) 07:00, 9 August 2014 (UTC) Happy to help with manual finetuning that can't be done by bots, and anything else on the 'soft/wet' side of this project. I'm dreaming of complete artists' oeuvres on Wikidata! Rich Farmbrough (talk) Time to learn2Wikidata Jheald (talk) 12:33, 17 August 2014 (UTC) Kippelboy (talk) 07:01, 21 August 2014 (UTC) (Focus on Catalan paintings (subdivision of Spain) Mushroom (talk) 12:27, 21 August 2014 (UTC) Jane023 (talk) 09:11, 3 October 2014 (UTC) work on Dutch 17th-century paintings and landscapes of Haarlem; Most recently, the sum of all "attributed" paintings by Frans Hals, which is nearly done Missvain (talk) 18:51, 18 October 2014 (UTC) (talk) 13:27, 15 November 2014 (UTC) Zolo (talk) 14:57, 23 November 2014 (UTC) Beat Estermann (talk) 10:33, 3 December 2014 (UTC) (Focus on Swiss heritage institutions) Vladimir Alexiev (talk) 15:07, 23 January 2015 (UTC) KRLS (talk) 11:26, 11 February 2015 (UTC) (Focus on Catalan area museums) DivadH (talk) 11:35, 1 March 2015 (UTC) ,happy to help out with any questions in regards to the Europeana API, how to best query it, and/or our metadata Xcia0069 (talk) 11:49, 8 March 2015 (UTC), Work on data related to Gianlorenzo Bernini and Artemisia Gentileschi. Work at Europeana too ! Susannaanas (talk) 07:29, 9 March 2015 (UTC) Wittylama (talk) 17:29, 20 March 2015 (UTC) Fabrice Florin (talk) 02:35, 26 June 2015 (UTC) I can help in California later this year. Vaughn88 (talk) 15:58, 15 July 2015 (UTC) I can help! Raymond Ellis (talk) 19:31, 17 August 2015 (UTC) Hsarrazin (talk) 14:11, 29 August 2015 (UTC) - will give a hand with Creators and AC :) louis-garden (talk) 14:21, 31 August 2015 (UTC) for italian paintings (XIIe-XVIIe) Olivier (talk) 21:46, 8 September 2015 (UTC) Kopiersperre (talk) 11:33, 20 November 2015 (UTC) ProtoplasmaKid (talk) 03:49, 23 February 2016 (UTC) Micru (talk) 11:19, 29 February 2016 (UTC) Stuart Prior (WMUK) (talk) 11:04, 28 April 2016 (UTC) Hannolans (talk) 23:14, 22 October 2016 (UTC) Geraki (talk) 09:52, 24 October 2016 (UTC) (Focus on Greece) PatHadley (talk) 12:16, 3 January 2017 (UTC) MartinPoulter (talk) 14:54, 11 January 2017 (UTC) Working to get data from the University of Oxford (Q34433) and its component institutions shared on Wikidata. Pablísima (talk) 18:07, 8 February 2017 (UTC) Carl Ha (talk) 22:10, 9 February 2017 (UTC) Marsupium (talk) 19:44, 22 May 2017 (UTC) Mauricio V. Genta (talk) 16:15, 26 June 2017 (UTC) Shani Evenstein (talk) 10:26, 26 July 2017 (UTC) Nasty nas (talk) 07:45, 24 August 2017 (UTC) Bodhisattwa (talk) 14:28, 28 October 2017 (UTC) Joalpe (talk) 18:39, 9 November 2017 (UTC) Fuzheado (talk) 18:33, 30 November 2017 (UTC) Sarasays (talk) 20:00, 1 December 2017 (UTC) Thierry Caro (talk) 07:30, 9 December 2017 (UTC) John Samuel 18:29, 21 December 2017 (UTC) Jklamo (talk) 12:06, 31 December 2017 (UTC) Reosarevok (talk) 10:28, 15 February 2018 (UTC), focus on Estonia Ambrosia10 (talk) 19:48, 19 February 2018 (UTC) Subsublibrary (talk) 03:17, 22 February 2018 (UTC) Martingggg (talk) 07:00, 22 February 2018 (UTC), focus on Argentine and Hispanic America Kruusamägi (talk) 16:42, 13 March 2018 (UTC), focus on Estonia SIryn (talk) 10:36, 9 June 2018 (UTC) Jarekt (talk) 13:49, 7 September 2018 (UTC), focus on moving metadata from Commons to Wikidata Walkuraxx (talk) 10:00, 30 November 2018 (UTC) Omotecho (talk) 22:11, 21 January 2019 (UTC), focus on Japan GualdimG (talk) 16:19, 19 February 2019 (UTC) Léna (talk) 08:57, 4 April 2019 (UTC) Yann (talk) 09:53, 9 June 2019 (UTC) Paul Cézanne (Q35548) for a start... Abbe98 (talk) 19:25, 18 June 2019 (UTC) Daniel Mietchen (talk) 15:36, 23 July 2019 (UTC) Have been doing small things here and there — especially improving items around what is depicted on paintings — so might as well sign up. Jbandrews (talk) 02:14, 16 October 2019 (UTC) Product Manager at the Art Institute of Chicago. Amadalvarez (talk) 06:19, 19 October 2019 (UTC) Meresquared (talk) 14:51, 22 October 2019 (UTC) I'm interested in adding depiction descriptions to paintings, and improving data around women and POC.   Notified participants of WikiProject sum of all paintings

Do you think we should make a prop for WikiArt? eg

 – The preceding unsigned comment was added by Vladimir Alexiev (talk • contribs).

It's a wiki without sources, not sure if that is useful. Sjoerd de Bruin (talk) 16:15, 9 July 2017 (UTC)
What do you mean "without sources"? When not having access to a particular database stopped us from linking to a rich information site? First it's a useful resource on its own. Second, they may have a permissive license allowing scraping. Third, we could talk to them to get a data donation --Vladimir Alexiev (talk) 15:07, 11 July 2017 (UTC)
How much content does that Wiki host?
2672 artists (156 movements in 2-level hierarchy, 253 schools/groups, 50 genres, 28 fields, 87 nationalities)
Looking at Dali: 1102 artworks, his Periods, related artists. In comparison, his Commons gallery is very poor because " the artist has not been dead for at least 70 years. Please do not upload photographs or scans of works by this artist, unless...". His Commons category has a lot more images, but this is not a list of works. Catalogue Raisonné of Salvador Dalí Paintings (Q24009539) just points to https://www.salvador-dali.org/en/research/cataleg-raonat/.
146509 artworks (160 styles, 54 genres)
To what extend is it a mirror of Wikipedia? ChristianKl (talk) 18:02, 9 July 2017 (UTC)
Looking at Dali, I see no evidence this has *anything* to do with Wikipedia. What gave you that idea, the site's name? --Vladimir Alexiev (talk) 15:07, 11 July 2017 (UTC)

IUCN-ID (Property:P627)

In the latest ten or so online versions of the IUCN Red List of Threatened Species the links have ended with the IUCN taxon ID (entered as a numerical value for P627 here at Wikidata) immediately followed by "/0". This works well with property P627: see Wikidata item Q208549 (Mediterranean monk seal) for an example of such a link, where the IUCN-ID "13653" is listed and properly linked to IUCN about halfway down the page. However, in the most recent version (IUCN 2017.1) they have changed their system, and now some of their links ends in "/1" instead. This brakes the links rendered by P627: see Wikidata item Q2275770 (Geoffroy's Blood Bee) for an example, where the correct IUCN-ID "19198964" is listed, but the link generates an HTTP 404 error code: iucnredlist.org/details/19198964/0. The proper link is iucnredlist.org/details/19198964/1 with the last digit being the only difference.

Is there any way we can fix this, or are we at a loss? Best regards, Tommy Kronkvist (talk), 21:54, 9 July 2017 (UTC).

Some options -
  • If all items have switched to /1 we can change the URL formatter. This is easy.
  • If some items have switched to /1, but not all, we can change the standard form in Wikidata to be xxxxxxxx/y - so 19198964/1 rather than 19198964.
This second option will take a bit of work but it's not impossible - we just add all the new values and then remove the old ones. Do we have any way of predicting which will end in /0 and which in /1? If we do, great, if not... hmm, the best approach might be to run a script looking at every value using the /0 and /1 formats and seeing which returns a 404 error. Andrew Gray (talk) 22:27, 9 July 2017 (UTC)
A third ("nuclear") option is to replace the property with an equivalent, but using the URL datatype. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:30, 9 July 2017 (UTC)
It might be worthwhile to ask per email them why they made the change, to better understand whether they want to express something with the change. ChristianKl (talk) 13:36, 10 July 2017 (UTC)
They write on their webpage "It is advisable wherever possible to use the taxon name (species name) to make your API calls, rather than using IDs. IDs are not immovable are expected to be used mainly by organisations that work closely with the IUCN Red List." --Pasleim (talk) 14:15, 10 July 2017 (UTC)
Hi Tommy: For some time now IUCN supports regional assessments besides global assessments. See e.g. Ceriagrion glabrum The general URL format is http://www.iucnredlist.org/details/iucn_id/region_identifier/ with the following region identifiers:
Global (=0), Northern Africa (=20), Eastern Africa (=17), Persian Gulf (=26), Europe (=1), Southern Africa (=18), Mediterranean (=3), Western Africa (=7), Central Africa (=15), Pan-Africa (=9), Gulf of Mexico (=25), Northeastern Africa (=16).
Mediterranean monk seal (Q208549) has two regional assessments and global one. Geoffroy's Blood Bee (Q2275770) has only one regional assessment. At the moment only global assessments are supported by IUCN taxon ID (P627). --Succu (talk) 12:45, 11 July 2017 (UTC)
Thanks Succu for that useful update. Sadly it doesn't help us much – if anything, it makes the issue more complicated. Using the taxon name (per Pasleim's input) or the URL datatype (per Andy Mabbett) might be the way to go. However the problem with using taxon names is pretty much the same as with using IUCN IDs: they are not static, and some of them are changed fairly often. –Tommy Kronkvist (talk), 13:41, 11 July 2017 (UTC).
A possible solution is related to #Need more flexible Formatter URL. If I'm right the are around 12,000 species and subspecies that are not globally assessed. --Succu (talk) 15:10, 11 July 2017 (UTC)
Note: IUCN taxon ID (P627) is refering to a taxon concept evaluated by IUCN. A taxon concept is labeled by the prefered name of a taxon (=taxon name (P225)). A taxon concept can change by lumping/splitting/etc. taxa. Then the IUCN-ID is droped and a new one is assigned to the evaluated taxon concept. --Succu (talk) 20:23, 11 July 2017 (UTC)

Disambiguation items

I have a few questions about disambiguation items:

  • Help:Statements says "Disambiguation pages should have only the property 'instance of (P31)' with the value 'Wikimedia disambiguation page (Q4167410)'". However, the disambiguation page guidelines don't restrict properties to only P31. Is the use of said to be the same as (P460) on items such as Cat (Q1022892) valid? It seems wrong anyway, since "this item is said to be the same as that item, but the statement is disputed" doesn't reflect what's going on, since disambiguation items with different labels can't be merged.
  • Help:Label says labels begin with a lowercase letter unless the upper case would be used in the middle of a normal sentence. Does that apply to disambiguation items? What if the disambiguations are a mixture of proper nouns and ordinary nouns? What if none of them are proper nouns?
  • The disambiguation page guidelines say that site links must be identical, except that special characters can be added/removed (e.g., accents). Does that mean we can merge Cat (Q1022892), Çat (Q1211764) and Cát (Q1307187)?
  • The disambiguation page guidelines say that transliterations are allowed. In practice, it doesn't seem easy (when you aren't familiar with a script) to distinguish a transliteration from a translation. E.g., fa:گربه_(ابهام‌زدایی) is linked to Cat (Q1022892), although it seems to be a translation. I'm hardly certain about it though.

Thanks for any insight. Ghouston (talk) 23:23, 9 July 2017 (UTC)

Full name disambiguation pages are quite common and shared among languages, so I would keep them for the interwiki links. I have started using the property "Has part" to include the parts in the disambiguation items. This also helps for cases where on some Wikipedias no red links are allowed in disambiguation pages. Jane023 (talk) 19:08, 10 July 2017 (UTC)
  • Working on them with a bot seems like a great idea. Depending on the exact rules, perhaps most of the maintenance could be automated. Ghouston (talk) 01:37, 10 July 2017 (UTC)
(personal opinion) Guidelines are wrong, a disambiguation is based on the characters of the term, it use isn't often a word, so cannot be said to be the same as anything; they don't even mean the same thing in the one language, let alone many. It is my opinion that it is, and should only be, characters for one character set. I wouldn't add the Cyrillic version of the same, though at the same time I would only remove a link for a wiki if there was a better match, eg. "Gordon" is not equal to "Гордон", though if it was linked, I would leave it unless "Gordon" disambig appeared on the wiki in question, in which case I would swap it in.  — billinghurst sDrewth 01:03, 10 July 2017 (UTC)
  • Another difficulty is that transliterations aren't always unique, you can convert a single Cyrillic word into various Latin equivalents, and there are multiple systems for Chinese. Would you also want to stop merging the characters with different accents? Ghouston (talk) 01:21, 10 July 2017 (UTC)
  • It would also be a problem if a wiki has separate disambiguation pages that differ only in accent. Ghouston (talk) 01:23, 10 July 2017 (UTC)
  • There is also a property proposal and an RFC about disambiguations.
    --- Jura 01:29, 10 July 2017 (UTC)
  • I think regarding my label question, it's easiest to capitalize every disambiguation label. I can't imagine wanting to check every wiki and decide if any of the meanings are proper nouns, and a bot wouldn't want to do it either. Ghouston (talk) 01:37, 10 July 2017 (UTC)
As for the use of said to be the same as (P460), or different from (P1889) on disambiguation items, I'd say they are misguided attempts to sort the pages by topic, and can be deleted. Ghouston (talk) 01:37, 10 July 2017 (UTC)
As for merging Cat (Q1022892), Çat (Q1211764) and Cát (Q1307187), it seems they were separated partly due to topic sorting, a desire to keep "Cat" surnames separate, which also seems misguided. If we want to merge accents, these items should be merged. zh_yue wiki has links to two of them, but it's a transliteration issue again and I'm not sure if they are valid. Likewise for Gato (Q5447646) vs Gató (Q11923863). Ghouston (talk) 01:37, 10 July 2017 (UTC)
Should it be possible to merge items that differ only in captialisation too? E.g., TAC (Q32550) and Q711595. There are also items where it would be a shame to de-merge them, because the sitelinks are quite compatible in translation, e.g., Oakland Township (Q2008516). Ghouston (talk) 04:02, 10 July 2017 (UTC)
  •   Comment I do not think disambiguations makes much sense to keep here at Wikidata at all. I haven't seen any use for them. Adding them to almost any property other than "said (not) to be the same as:" is always wrong. The only real reason for us to keep them here is the Interwiki. We could choose to exterminate them from Wikidata without loosing anything of value for us. The only problem is that we in such cases would suffer a riot (Q124757) from Wikipedia. -- Innocent bystander (talk) 08:36, 10 July 2017 (UTC)
  • Yeah, I assumed they were just for interwikis. So assuming they are going to be kept, the only things to consider are a) what's useful for interwikis b) what makes maintenance easier, preferably with simple rules that can be implemented with a bot. Ghouston (talk) 08:56, 10 July 2017 (UTC);;
  • Or just simply let people who, by some strange reason, have interest in them, do whatever they like, except from mixing this set with the "real Wikidata". There are too many exceptions in this set to make it worth maintaining with strict rules. -- Innocent bystander (talk) 09:15, 10 July 2017 (UTC)
  • Well, this is where I started running into problems, when I was trying to untangle some disambiguation links from a "real" item. It was only when I started getting confused that I discovered that there were different rules for disambiguation items, label-based vs concept-based. Ghouston (talk) 09:21, 10 July 2017 (UTC)
  • I am not so sure about that, the key is that they are all treated as homonyms. In that perspective they all have exactly the same concept. What differs them is the label. But of course, there are many disambig-articles that have more than the spelling in common. The articles in Slaget vid Öland for example has the concept of sea battles in the Baltic sea including the word "Öland" in common. The article is the result of a very long and bloodthirsty edit war about which article should have this label. The disambig is a compromise in the peace treaty to that edit war. Dare to change that article! -- Innocent bystander (talk) 09:44, 10 July 2017 (UTC)
  • From a Wikidata perspective, personally, I think the priority is to ensure that such items don't include sitelinks to articles (and other items don't link disambiguation pages at Wikipedia). This can happen through pagemoves, re-purposing of pages at Wikipedia, etc. (see point 7 on Wikidata:Bot_requests#Take_care_of_disambiguation_items). There should be an automated way to do this. Also disambiguation items shouldn't be used as property values (constraint reports and Wikidata:Database reports/Most linked disambiguation page items ensure this).
    --- Jura 09:50, 10 July 2017 (UTC)
    I have noted several times the last weeks that page moves on Wikipedia, over an existing (and thereby deleted) article tends to result in wrong sitelinks. This often affects disambigs.
    Another problem is that Wikipedia often have a disambig-template in articles we here at Wikidata consider "real" concepts, like names and lists. -- Innocent bystander (talk) 09:57, 10 July 2017 (UTC)
  • Like on Maximilian II (Q227885), "Maximilian II (disambiguation)" could also be done as "List of people named "Maximilian II". But "people named Maximilian II" is a sort of arbitrary list, like any other list of people who share the same name. Ghouston (talk) 10:11, 10 July 2017 (UTC)
    • I'm sure it's possible to find a solution for most issues (please add at Wikidata:Bot_requests#Take_care_of_disambiguation_items if needed). Obviously, it's a still a problem that Wikidata doesn't update sitelinks if pages are moved by people who haven't edited (or at least logged into) Wikidata ever. It just keeps deteriorating data.
      --- Jura 10:13, 10 July 2017 (UTC)

Help fix occupations

I am unclear about the status of some occupations (P106), but e.g. some are just wrong. So an archbishop is not an occupation, but a position held. Are there constraints for these? Could it be that I see historical mistakes (so dating from 2013-2015)? I ran a list of totals of women per occupation here: User:Jane023/Number of women per occupation. They are listed in descending order of popularity with an enormously long tail of near-single usages. Some of them are just wrong. Please help clean them up! I tried running this for men but it timed out. I am starting a new thread here, because my earlier post has to do with incorrect use of a different property, not P106. Jane023 (talk) 15:44, 10 July 2017 (UTC)

An example of a profession that I have serious doubts about is erudite (Q20826540), which is only clarified by stating it is different from scholar (Q2248623). Jane023 (talk) 15:59, 10 July 2017 (UTC)
I can take a look at these. - - PKM (talk) 21:59, 11 July 2017 (UTC)
Thanks! I have started looking at the arts-related professions (as well as arts-related professions that don't show up in the list for some reason). Jane023 (talk) 11:58, 12 July 2017 (UTC)

A gadget to delete all labels, descriptions and aliases in all languages of an item

Is there a gadget or a tool allowing to blank all labels, descriptions and aliases in all languages of an item ? The item metasilicate (Q6823230) was defined as a disambiguation page but it is not. I am correcting the content, but I need to delete all labels in order to be sure I can start from a clean item. I have currently 15 languages displayed in my interface but I want to clean all languages. Thanks Snipre (talk) 18:09, 10 July 2017 (UTC)

@Snipre: Yes, see MediaWiki:Gadget-dataDrainer.js. Jared Preston (talk) 18:23, 10 July 2017 (UTC)
QIDs are meant to be stable. If you delete all descriptions, labels and statements, you are re-purposing the item and confused our data users.
--- Jura 21:21, 10 July 2017 (UTC)
@Jura1: Just explain me what is the concept of a disambiguation page ? There is no concept behind a disambiguation page: this is just a organization item for WP mixing different concepts having a similar name. Then if you take the time to study the problem, you will learn that the choice of a disambiguation page for the concept metasilicate on WP:en was in contradiction with all other similar pages like en:silicate or en:Permanganate. This was checked with the Chemistry project on WP:en. Finally I corrected labels which were all in English and started the translation in order to offer the possibility to all users to understand what is the concept of the original English article. I am working on WD since sevral years so I know the basics. Snipre (talk) 21:41, 10 July 2017 (UTC)
I know the idea of relative stability of ids is hard to explain and some struggle with it for years. Maybe someone else can explain it better.
--- Jura 22:13, 10 July 2017 (UTC)
The bots and tools who mass-adds these descriptions maybe should check that every sitelink really is a disambig or category before they completely mess up the items. -- Innocent bystander (talk) 08:15, 11 July 2017 (UTC)

Strategy discussion, cycle 3. Let's discuss about a new challenge

Hi! It's the second week of our Cycle 3 discussion, and there's a new challenge: How could we capture the sum of all knowledge when much of it cannot be verified in traditional ways? You can suggest solutions here. You can also read a summary of discussions that took place in the past week. SGrabarczuk (WMF) (talk) 12:28, 11 July 2017 (UTC)

Improving toolbar on the Query Service

Hello all,

You may have noticed some changes on the Query Service interface :) We added a toolbar so you can access easily the main functions of your favorite SPARQL endpoint.

 

  • the expand icon allows you to switch to full-screen mode. You can exit this mode by pressing the "escape" key
  • the pin icon shows you the useful prefixes you may want to use in your query
  • the folder icon is another way to display the queries examples
  • the previous icon allows you to go back to the previous query you wrote (the content is stored in the local storage of your browser, so you can also find back the last query you wrote before closing the tab or browser)
  • the bin icon deletes the whole content of the editor (if you do this by mistake, don't panic, use the previous button to get your query back ^^)
  • the chain icon displays the short URL of your query
  • and last but not least, the "run" button has been replaced by a simple play button

The rest of the interface remains the same. We're also continuing working on the Query Helper box, on the top right of the screen, that allows beginners to edit queries without touching the code. We will soon introduce to you the new filter tool based on context aware suggestions.

What do you think? If you notice any bug or want to make a suggestion, feel free to add a comment on the ticket, or to ping me or Jonas Kress (WMDE). Lea Lacroix (WMDE) (talk) 15:40, 11 July 2017 (UTC)

Looks good, thank you. One small thing - in the expanded view, there's no "play" button. Also, I've also previously suggested that there should be a "switch to 'embed' view" button. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:47, 11 July 2017 (UTC)
small but great imrpovement! Just one big remark, the space bar to escape is good but not enough, and why is there no keyboard shortcut to enter full-screen mode? the asymmetry is quite counterintuitive. Seing Andy comments, I agree and think that probably all these new icons should stay in full-screen mode. Cdlt, VIGNERON (talk) 15:58, 11 July 2017 (UTC)
It would be really great to be able to save and manage queries in the Query Service. — Ayack (talk) 19:35, 11 July 2017 (UTC)
@Ayack: see phab:T104762.--GZWDer (talk) 21:03, 11 July 2017 (UTC)
@Jonas Kress (WMDE), Lea Lacroix (WMDE): The Phabricator task, where you asked us to leave feedback, has just been closed. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:28, 12 July 2017 (UTC)

Authors for PlantsAndPeople

Hoi, I have completed the first part showing the people that are known for now to be of interest to the Platns and People project.

The next thing I want to do is add the plants that are targeted for the editathons.

My question to you is, what is it that you are interested in from a library that is associated with a botanical garden. Personally I would love to see images of the types used for taxons. There will be issues where we may discuss the need for additional properties. For instance we use an identifier for botanists but there are two. One is for those who published taxonomy and the others is for any and all botanists.. Are we happy for the people who make a career out of taxonomy to help us/take part in these discussions? Thanks, GerardM (talk) 16:56, 11 July 2017 (UTC)

Redirect creation failures…

I see many incomplete mergers recently, where the redirected item is not properly changed to a redirect after blanking. Since the mergers are performed by lots of different users, I suspect that some script is not working properly. Has this been observed by someone else as well? What could be the reason? —MisterSynergy (talk) 17:18, 11 July 2017 (UTC)

I have never had a problem with incomplete merges using the merge gadget in the web UI, but I HAVE recently seen many failed mergers when I tried doing merges via QuickStatements - there may be a problem with how that is interacting with wikidata's API somewhere? It works most of the time, but around 10% failed when I was doing this a few weeks back, and I had to go back and fix the merges by hand. ArthurPSmith (talk) 17:49, 11 July 2017 (UTC)
I think this happens when the item to be merged is not empty after the merge. It is possible to tell the API to ignore conflicting data, eg. sitelinks or descriptions, then the conflicting data stays in the item, which is not possible to redirect since it's not empty. Matěj Suchánek (talk) 18:17, 11 July 2017 (UTC)
Also me, I find a lot of incomplete merges, in this case I redo the merge, but I don't know the reason ... --ValterVB (talk) 18:37, 11 July 2017 (UTC)
Good hint, I looked at a couple of cases and they are in fact not empty, there are some descriptions and aliases left after the merge (and before redirect creation). How is this caused? (I don’t see any Widar tags whatsoever, btw.). —MisterSynergy (talk) 18:52, 11 July 2017 (UTC)
I suspect the client "add links" widget (no edit summary or tag, right?). Matěj Suchánek (talk) 18:56, 11 July 2017 (UTC)
Correct, no edit summary upon “blanking”. Two examples would be (without blaming the editors): [3][4]. —MisterSynergy (talk) 19:06, 11 July 2017 (UTC)
Can the "add links" widget merge items? If the page is linked to a Wikidata item, the "Add links" link takes me to Wikidata. I was thinking it's probably Special:MergeItems (see phab:T127213). - Nikki (talk) 05:00, 12 July 2017 (UTC)
Of course it can. I often use it to merge duplicates. Special:MergeItems does not ignore conflicting data, so it cannot be the cause. Matěj Suchánek (talk) 06:39, 12 July 2017 (UTC)
Are we talking about the same thing? The "Add links"/"Edit links" link below the interwiki links on other wikis? If we are, how do you merge items with it? It just takes me to the Wikidata page when there's already a Wikidata item. - Nikki (talk) 08:05, 12 July 2017 (UTC)
Yes, we are. If a page has no interwiki, you get the widget instead. If you connect it to another page and both pages have items, they are merged. Matěj Suchánek (talk) 08:13, 12 July 2017 (UTC)
Example: Q20079942 merged with Q9304368 failed, --ValterVB (talk) 19:15, 11 July 2017 (UTC)

Being unwelcoming by deleting items without talking to their creators

Despite notability policy saying "It refers to an instance of a clearly identifiable conceptual or material entity. The entity must be notable, in the sense that it can be described using serious and publicly available references. If there is no item about you yet, you are probably not notable." and not "It refers to an instance of a clearly identifiable conceptual or material entity. The entity must be notable, in the sense that it is described using serious and publicly available references. If there is no item about you yet, you are probably not notable." ValterVB seems to have decided to delete items that have no statements or sitelinks without checking whether they fulfill our notability criteria and are about clearly identifiable entities. Even worse, he has decided to delete them without first engaging in dialog with the people who create the items.

If we want to be welcoming of new users deleting their work instead of talking to them seems a bad strategy. In the current case of https://www.wikidata.org/wiki/Wikidata:Data_Import_Hub#Gatehouse_Gazetteer_.28Wales.29 ValterVB seems to be set on using his powers to make it harder for a new user to contribute to Wikidata.

In this case, I went through the list of items without statements and sitelinks and found the items about the castles that were created. I created a property proposal for a corresponding external-ID and I talked with it's creator. Then ValterVB deleted the item in the property proposal despite it not only referencing a clearly identifiable entity but also serving a structural need for the property proposal discussion. I think this type of deletionism is harmful for our community and should stop. ChristianKl (talk) 15:57, 10 July 2017 (UTC)

I have discovered castles can be tricky. Sometimes they are non-notable, even when you would assume all castles are. An example is when only their location remains because they have been "moved" to make way for a highway and therefore there is only an abract dot on a map and a set of coordinates. I would still say it's notable for historic reasons, but not everyone would agree. Jane023 (talk) 16:03, 10 July 2017 (UTC)
When a castle has been moved or whatever, the notability of the castle at this time may be diminished but its historical significance is still there. Notability is not about only the present, it is also its notability in the past. Thanks, GerardM (talk) 06:23, 11 July 2017 (UTC)
This is difficult. A general comment, unrelated to the case you mentioned:
  • We have lots of items about “clearly identifiable conceptual or material entities” without references or identifiers, where only the creator can do the identification (possible reasons: requires special knowledge about the entity, requires special access to sources, ambiguity of the labels, etc.). I do expect editors to provide these references/identifiers soon after the items were created, particularly if there are no sitelinks or backlinks. Admins can’t do that, we have way too many of such cases. However, this is (unfortunately) not expressed in WD:N right now, but we should consider doing so.
  • As an admin, I tried to get in contact with creators a couple of times, but this was not very successful. I even got the impression that editors are rather scared about admin presence on their talk page, although I tried to be as kind as possible and offered help explicitly. Most times there was no response at all (and I had to spent considerable effort into the repair job by myself), or the creators indicated that the items were indeed tests, accidental duplicates, abandoned leftovers, even unknowingly created items through some tools, and so on. It is rarely the case that an item could be rescued this way, but I’d estimate that we collect at least 100 new ones with this problem each and every single day. Since only very few admins look for them (clearly less than 5 each day I’d say), there is no way to talk to all editors individually.
  • Items not identifiable by “every” editor tend to be forgotten by the community. If there is edit activity at all, it is mostly by bots, involving the addition of information you can already derive from existing content (example: add P735 based on label string).
  • There is also unclarity about the seriousness of references. A curated high-quality online database is typically regarded as serious, while a Facebook profile probably isn’t. Yet there is a large region in between these poles where it is very unclear whether a reference is serious enough to proof notability. While this is not the main problem here, it does at least add extra work to the editors (admins) who try to identify all kinds of items which aren’t unambiguously identified yet.
MisterSynergy (talk) 16:54, 10 July 2017 (UTC)
In this case, a simple google search on the name of the items shows the corresponding database. They are not items that are not identifiable by “every” editor.
If we wanted to delete all items without sidelinks or statements in an automated fashion we could start with a bot that writes a note to the talk page of the item and pings it's creator. If there no reason within two weeks the item gets deleted, otherwise it gets in a queue for human review. ChristianKl (talk) 21:11, 10 July 2017 (UTC)
In this particular case it might have been possible to perform a Google search, but this is not the case for most affected items (label ambiguity, language barriers). Unfortunately you don’t know whether a Google search is successful until you perform it. There are just way too many items, or way to little number of admins to establish such a procedure—or we expect this job to be done by the creators. Whether or not we support them by bots or policies is a matter of discussion, and I would agree with you that the current solution is far from being optimal for item creators as well as for deletion administrators.
I would also like to emphazise the importance of the deletions by itself. If we didn’t watch for unidentifiable items, we would basically give up all notability criteria. One could simply create an item about oneself, any company, something fictional, faked content, and so on without an identifier and nobody would ever take notice of this. In case of existing sitelinks the notability check is done by the connected projects, and used items are somewhat bound to the knowledge tree as well (although they better should have claims of course), but the items under discussion are not checked for notability at all. —MisterSynergy (talk) 21:37, 10 July 2017 (UTC)
We have a very low threshold for notability, and we have a very high n° of items to check. I follow 3 rule before delete item and all these rules must be respected:
  1. It does not have backlink
  2. It does not have sitelink
  3. It does not have reference, and don't have identifier except social link, not useful for notability.

After this I check the history to detect if someone has deleted sitelink, if the page exist then I restore; if someone has moves the sitelink without "trace" (deletete and add) then I merge. If someone ask to admins to search and add reference, start an RFC and then add to the rules. The problem for user alerts is that they are too much, isn't possible add all this item to RFD, if someone is interested exist a lot of report full of item candidate to deletion. After deleted if someone write me to restore the item so he/she can add reference, I restore without problem, but if I restore and none add referece, I delete again. --ValterVB (talk) 20:24, 10 July 2017 (UTC)

Your steps are flaved in that you will remove items, when author forgot "stated in" statements
Information about castles is not relational unless a historian was editing it.
Normally we should stop and contact authors when we see unsourced imports.
IMO proper way is to create item "unknown source 2342342" and make "stated in" claim and keep them until we have better resolution or more editors with more information. d1g (talk) 06:04, 11 July 2017 (UTC)
We have Watchlists, I can see if somebody has edited or deleted items I have created. I can ask the sysop who deleted the page what it was she deleted and why. I do not want a message at my talk_page every time somebody has an opinion about my items. In my opinion this habit of constantly leaving messages to talk pages for the smallest of reasons, is one thing that makes Wikimedia Commons a very awkward place to be at! And I do not own the items I create here, they are in public domain. -- Innocent bystander (talk) 06:35, 11 July 2017 (UTC)
Watchlist won't help to fill sources. It is enough to discuss one item or whole dataset.
Wikidata items are either fictional or real and attributable (or items with common sense)
d1g (talk) 08:16, 11 July 2017 (UTC)
The watchlist tells you that an item was deleted and it gives you the ID of the items. Users without admin rights often don't know which exact item was deleted. Experienced users might be able to ask a sysop/admin about items but I don't think that's an option that a new user thinks about. ChristianKl (talk) 17:49, 11 July 2017 (UTC)
I see no problems of that kind for newbies at Wikipedia. I there daily see requests to undelete and review of deleted pages from newbies there. It could be tougher here at Wikidata, yes, since you maybe not find the red item if you do not include it into your watchlist in the first place. And if you do not identify the item, you will not find the deleting sysop in the log. -- Innocent bystander (talk) 05:14, 14 July 2017 (UTC)
@ValterVB: You recently deleted an item that was used as an example in a property proposal. When another admin undeleted it, you deleted it a second time. How is that possible, if you check for the absence of backlinks, before deleting? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:48, 11 July 2017 (UTC)
I use different strategy to check the item, for backlink there is 2 different tools: SPARQL, but work only in ns0 so can't help in this case, and template:Rfd links but sadly also this work only on ns0, some time i generate long list of link to the history with excel and open the links with "Bulk URL Opener" extension, in this case I can have more control and I can add also other namespace. I never thinked at link to property proposal so I will add this check to my workflow. --ValterVB (talk) 18:48, 11 July 2017 (UTC)

Quick translation needed: multiple languages

The {{User Wikidata birthday 2016}} template has text in these languages: cy, el, en, de, fr, it, nl, pt, ru, sco, sv, tt. The English version is This user is celebrating Wikidata's [[Wikidata:Fourth Birthday|4th birthday]]..

I shall shortly begin making this year's template.

If the list above omits a language you can write, please add the translated text to last year's template, from where it can later be copied and updated for this and future years. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:56, 12 July 2017 (UTC)

Logical operators on statements

We know that the 2024 Summer Olympics (Q995653) will be held in either Los Angeles (Q65) or Paris (Q90). I'd like to add a statement for country (P17) and located in the administrative territorial entity (P131), but I cannot add both LA and Paris as statement objects since they won't _both_ be true. Rather, _one of_ the two is true, we just don't know which yet. I'd love to have some way of encoding a logical OR. We already have a way of encoding logical AND: just add both objects. Is there a way that Wikidata would currently accept a statement that the 2024 Olympics will be in either the US or France but not both? Runner1928 (talk) 19:01, 12 July 2017 (UTC)

I believe that's what disjoint union of (P2738) is for. ArthurPSmith (talk) 19:26, 12 July 2017 (UTC)
No, I'm pretty sure that's only for delineating classes. --Yair rand (talk) 19:37, 12 July 2017 (UTC)
What about candidate (P726)? --Pasleim (talk) 21:42, 12 July 2017 (UTC)
Or you just wait until september when the decision will be made. in the meantime, candidate (P726) is a good option. Q.Zanden questions? 00:38, 13 July 2017 (UTC)
I agree that it would be nice "to have some way of encoding a logical OR" and XOR. There are a lot of cases where specific properties like candidate (P726) don't work and it the OR won't be resolved in the future. --Marsupium (talk) 01:05, 13 July 2017 (UTC)
I very strongly suspect that adding logical operators will result in the creation of an entire extremely complex language used for describing things that no one imagined would ever be on Wikidata. It's just one of those things that seems to happen, especially on Wikimedia. (Some examples include the if-templates hacks -> parserfunctions -> endlessly convoluted 12th-level metatemplates -> Lua, the template-data systems on some Wikipedias, ENWT's old category-generating mechanisms and bizarre language-database).
I'm not saying this is certainly a bad thing, but if we go down this road, we should plan it properly instead of piling things on bit by bit and then being all surprised when the system inevitably becomes quite messy. --Yair rand (talk) 02:15, 13 July 2017 (UTC)
I agree. --Marsupium (talk) 13:23, 13 July 2017 (UTC)

Please note: There is no implied logical relationship between multiple statements given for a single property. The data model intentionally does not specify this. Giving two (or more) statements may mean both are true, or it may mean one used to be true and the other is now true, or it may mean either may be true according to different sources. The interpretation depends on the inherent semantics of the property and can be specified further using qualifiers, ranks, and references. -- Daniel Kinzler (WMDE) (talk) 16:30, 13 July 2017 (UTC)

we want to have "and" and "or"s within one property, so multiple properties will double property count
Most common cases is "part of" and "made of" statements
piece 1 and piece 2, but piece 3 or piece 4
ingredient 1 xor ingredient 2, and ingredient 3
I don't know if qualifiers would make sense here or how.
We possibly want rdf:bag + operator marks here to write simple or/and/xor sentences d1g (talk) 17:40, 13 July 2017 (UTC)

Back to the original question! There is an item Selection of the host city Summer Olympic Games 2024 and 2028 (Q20827507) about the candidacy-procedure. Instead of adding several values to 2024 Summer Olympics (Q995653), why not list all candidates to that item instead? There you can also specify when each of the candidates withdraw their candidacy. -- Innocent bystander (talk) 17:58, 13 July 2017 (UTC)

+1 they are still only candidates at this point of time.  — billinghurst sDrewth 04:29, 14 July 2017 (UTC)

Facto Post – Issue 2 – 13 July 2017

Facto Post – Issue 2 – 13 July 2017
 

Editorial: Core models and topics

Wikimedians interest themselves in everything under the sun — and then some. Discussion on "core topics" may, oddly, be a fringe activity, and was popular here a decade ago.

The situation on Wikidata today does resemble the halcyon days of 2006 of the English Wikipedia. The growth is there, and the reliability and stylistic issues are not yet pressing in on the project. Its Berlin conference at the end of October will have five years of achievement to celebrate. Think Wikimania Frankfurt 2005.

Progress must be made, however, on referencing "core facts". This has two parts: replacing "imported from Wikipedia" in referencing by external authorities; and picking out statements, such as dates and family relationships, that must not only be reliable but be seen to be reliable.

In addition, there are many properties on Wikidata lacking a clear data model. An emerging consensus may push to the front key sourcing and biomedical properties as requiring urgent attention. Wikidata's "manual of style" is currently distributed over thousands of discussions. To make it coalesce, work on such a core is needed.

Links


Editor @Charles Matthews:. Please leave feedback for him.

If you wish to receive no further issues of Facto Post, please remove your name from our mailing list. Alternatively, to opt out of all massmessage mailings, you may add Category:Opted-out of message delivery to your user talk page.
Newsletter delivered by MediaWiki message delivery

Charles Matthews (talk) 08:27, 13 July 2017 (UTC)

  • @Charles Matthews: "In addition, there are many properties on Wikidata lacking a clear data model. An emerging consensus may push to the front key sourcing and biomedical properties as requiring urgent attention. " could you expand on that point? Which biomedical properties do you believe lack a clear data model? When you speak of an "emergent consensus" do you refer to particular discussions? ChristianKl (talk) 17:47, 13 July 2017 (UTC)
For example, I know there has been some difficulty with medical condition treated (P2175), and that you are probably familiar with that issue. It is better to think through semantic lack of clarity at the data model level. There is the process that enWP went through starting more than 10 years ago, and ending with an explicit, community-written manual. When I working on How Wikipedia Works (Q2763515), the editor at No Starch Press (Q7045008) told me that he was impressed with it: because, I suppose, publishers and newspaper typically impose style guides top-down.
So I think we have plenty of work to do, to clarify both semantic points and the better choices in presenting information.
On the referencing side, "emerging consensus" would just mean the progress made at WikiCite. Actually, I know there, from my own experience, that the top-down approach hasn't immediately worked. But people are willing to work on detailed issues, and things will probably happen in a divide-and-conquer, community-led piecemeal way. My language is obviously somewhat journalistic, but I don't think it is actually misleading. Charles Matthews (talk) 18:10, 13 July 2017 (UTC)

Prototype for improving the Query Helper

Hello all,

Thanks for all your feedback regarding the toolbar of the Query Service. We will provide answers and fixes in the next days.

We would like to share with you a prototype for further improvement. As you may have noticed, the Query Helper is a feature of the Query Service that is designed for beginners, or people who don't know how to use SPARQL. With a visual interface, it helps people to create and modify queries, by selecting filters and things to show. This tool is not able to display all the features of SPARQL yet, but should do in the future.

  • We made the tool more visible and better integrated in the page: the Query Helper is on the left of the screen, the text editor on the right. The size increases depending on the lenght of the query. Note that if you don't want to display the Query Helper, you can hide it, and continue your workflow as usual. If you want to show it again, you can click on the "information" icon in the left menu.
  • We improved the Query Helper by adding some context-aware filters. If you click on "Filter", you can start entering the value (for example, cat). When you select it, the relevant property will automatically added and the filter will be "instance of cat" (a line "?variable wdt:P31 wd:Q146." will be added in the code). This workflow is a bit different from what the SPARQL experts are used to, but should be easier for the beginners.

You can try the prototype here. Feel free to give us feedback or suggestions here or in the ticket. Thanks, Lea Lacroix (WMDE) (talk) 10:54, 13 July 2017 (UTC)

Some issues:
  • The horizontal positioning doesn't work so well with smaller screens. I recommend using some media queries to reposition things on such screens.
  • The various little icons (especially the "tag") really need explanatory tooltips. I still haven't figured out what the tag button does.
  • The little popout bubbles should disappear when the user clicks elsewhere. Can be rather frustrating otherwise.
--Yair rand (talk) 20:26, 13 July 2017 (UTC)

Last days to apply for a scholarship for the WikidataCon

Hello all,

As a reminder, the WikidataCon scholarship process ends on July 16th. After this date, you can only rely on external resources (local organizations, other sponsoring process) for funding your trip and accommodation.

The registration process is currently closed, but we will release some tickets after the scholarship recipients are selected so they can register. Another bunch of tickets will be released later.

Thanks, Lea Lacroix (WMDE) (talk) 12:05, 13 July 2017 (UTC)

About Wikidata:Notabily

I write this to resolve a doubt I have about the scope of Wikidata:Notability. Costa Rica Skies (Q5174859) was a Costa Rican charter airline that only existed for a few months, never had airline identifiers but did permit to make international flights. According to its ex-president and founder, obtained its permission on December 22, 2007 and began operating in early 2008. In short: it existed and made international flights. The reason it closed prematurely was that the DEA discovered that it was a "screen" to transport cocaine from Venezuela to Costa Rica and from there to the city of Atlanta. Due to this I did not get to be recognized by the IATA or ICAO. Source in the external link. Except news articles about its link with drug trafficking, all the internet hits that I could find took its info from es.wikipedia (circular reference) that had the article since October 2007 (self-promotion of a company that did not yet exist...) If the element had no links to Wikipedia, would such an airline be "notable" for Wikidata? Why? I will be grateful for any guidance. (English lvl 1). --Metrónomo (talk) 01:05, 14 July 2017 (UTC)

If it exists at a Wikipedia in a satisfactory namespace, then it is notable. If it doesn't exist, or no longer exists at a Wikipedia, then utilise the other criteria to determine whether it is notable and suitable for WD, and start a discussion for its deletion. It is not our job to out-think the sister wikis, they choose their notability criteria; and if you think that it is not notable for the wiki, they have their deletion discussion options.  — billinghurst sDrewth 04:35, 14 July 2017 (UTC)
If they made at least one flight then it is a company, everything else is details - we have a separate properties for them
use (P366) approved by (P790) ICAO airline designator (P230) IATA airline designator (P229) d1g (talk) 09:57, 14 July 2017 (UTC)
Despite the Wikipedia links which confer notability under #1, it's notable under #2 because it's clearly identifiable conceptual entity that can be described using serious and publicly available references. ChristianKl (talk) 11:50, 14 July 2017 (UTC)

Line-ups

How do we model line-ups in sport team competitions? E.g. if I want to add the players in Arminia Bielefeld versus FC Bayern Munich 2006–07 Fußball-Bundesliga (Q30224025), how is this supposed to be done? Steak (talk) 09:06, 14 July 2017 (UTC)

If we have information about individuals, it should be in participant (P710)
We have qualifier for "position": position played on team / speciality (P413)
I'm not sure which qualifier to pick for "played for team". d1g (talk) 10:27, 14 July 2017 (UTC)
Maybe member of (P463)? Steak (talk) 11:47, 14 July 2017 (UTC)
It should make sense here IMO (given other options now) d1g (talk) 12:16, 14 July 2017 (UTC)

Upper/lower script in string properties

Does someone know how to write in upper/lower script when entering text as value of a string property ? Snipre (talk) 13:42, 14 July 2017 (UTC)

It's unicode. https://en.wikipedia.org/wiki/Unicode_subscripts_and_superscripts ChristianKl (talk) 14:20, 14 July 2017 (UTC)

RDF

I just created Wikidata:RDF, in order to have a page to point to when asked about our RDF output. It could do with some content from people who work with Wikidata's RDF. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:10, 9 July 2017 (UTC)

We should point to Wikidata RDF Dump Format (Q32786132), not to "Wikidata" d1g (talk) 14:34, 9 July 2017 (UTC)
  • I added skos:prefLabel (which WD emits as rdfs:label: that was surprising to me) and skos:altLabel (which WD emits as skos:altLabel).

@D1gggg: I agree it's a bit confusing but how would you propose to reformat it? --Vladimir Alexiev (talk) 15:18, 9 July 2017 (UTC)

@Vladimir Alexiev: perhaps one table is not sufficient for Wikidata model, because there are many levels of abstractions; RDF is one of the lowest. Many RDF features are not used yet (no need to compare them). d1g (talk) 02:03, 10 July 2017 (UTC)

  Comment

Wikidata:Data access → "database dumps documentation" link → Wikidata:Database_download#RDF_dumps

Wikidata:RDF: we have different RDF dumps (I forgot about it too), maybe we should cover RDF dumps at one page? d1g (talk) 19:19, 14 July 2017 (UTC)

Use of genre on humans

People don't have genres, but it appears that artists do on Wikidata. These should all be moved to P101 field of work. Can someone help with this? Jane023 (talk) 08:27, 10 July 2017 (UTC)

It seems like the property constraints and descriptions says something different. However, I agree with you. This will also influence the entity suggester in a better way. Sjoerd de Bruin (talk) 10:13, 10 July 2017 (UTC)
Thanks. Like you I felt that the constraints should have prevented this, but I also noticed some strange subclasses of genres as well. I started looking at occupations in general for artists, which is how I discovered this, but looking around it seems there is a broader problem with describing people using properties related to occupations (Genre I only noticed because I saw that people and artworks were both linking to genres). So for example I noticed that someone might be best known as a writer and have (correctly) the occupation writer, but that they also have an additional occupation "Assistant lecturer", which I am now thinking is incorrect. I mean, the person might be an assistant lecturer, but this is not their "claim to fame" and therefore irrelevant. It's like saying a famous painter of dogs also was a dog groomer, though in some cases "dog groomer" might actually be a claim to fame for someone else known for winning dog show competitions, etc. Jane023 (talk) 11:51, 10 July 2017 (UTC)
We ned to be careful about applying this as an item-level statement vs. a qualifier. Consider amateur painters who do so in a particular genre but are known for something entirely different. "Adolf Hitler -> field of work -> landscape painting" would not be sensible. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:41, 10 July 2017 (UTC)
I agree (see my comments on guitar players who might be mayors) Jane023 (talk) 15:23, 10 July 2017 (UTC)

It is not necessary better to have "field of work" mixed with musical genres or we should review P101 for every art-related profession, not just for musicians or vocalists. I had plans to request "able to play instrument". Genres are about "how they play" such instruments d1g (talk) 12:03, 10 July 2017 (UTC)

Well I guess it is OK for a guitar player to have occupation "guitar player" but then "field of work" could be "pop" or "classical". It might be overkill to say this person has occupation "musician" with "field of work" "drums"and "guitar" if they really mostly play the guitar. It depends on the person and if they are actually a city mayor who plays the guitar, then they should at least also have the occupation "politician" and maybe drop the musical occupation altogether. I just deleted the occupation "house painter" from a bi-athlete. Then there is e.g. Near Eastern archaeologist (Q26424344) which I think is wrong - this should be archeologist with field of work "Middle-east" or something. Jane023 (talk) 12:16, 10 July 2017 (UTC)
"drums"and "guitar" are far more suitable for "instruments" property
We cannot change genre to occupation because we don't know how many of them play for money.
Correction: I am asking to switch the items listed for humans under genre to Field of work. Jane023 (talk) 15:26, 10 July 2017 (UTC)
It is difficult to say about individual performers d1g (talk) 13:32, 10 July 2017 (UTC)
  • It does not matter for occupation (P106) whether the occupation is professional (paid) or not. We don’t really know this in so many cases anyway.
  • If an occupation is not the “claim to fame” as Jane023 described it above, it is still valuable in many cases to add this claim. However, preferred rank could be used for the “claim to fame” in that case.
MisterSynergy (talk) 13:39, 10 July 2017 (UTC)
Yes that is an interesting idea that I hadn't thought of. Jane023 (talk) 15:23, 10 July 2017 (UTC)
We have a couple of generic properties that cause undesirable “everything-links-to-everything” situations because both their domain and range of values are fairly unconstrained. genre (P136) is one of them, sport (P641) another one with a very similar problem. However, their claims are conveniently accessed from connected sitelinks, and Wikipedias use them a lot to present data in infoboxes etc. At the same time, querying these properties is very difficult since the result can contain pretty much everything.
In many cases, a better (cleaner) data modelling would make the infobox coding a little more complex (e.g. accessing a claim of a value item, instead of using a value item directly), but Wikipedias nevertheless seem to prefer simple access. The extra load is on Wikidata side, so they don’t really have to deal with the problems arising from this strategy. However, there are in fact cases where a clean modelling would make data retrieval in a template/module pretty difficult: traversing the knowledge tree from Wikipedia is not efficiently possible yet to my knowledge. But this is not a relevant problem for generic properties. —MisterSynergy (talk) 13:57, 10 July 2017 (UTC)
I must admit I noticed an alarming similarity between occupations and (English) wikipedia categories for people. At first I thought this had to do with people using Petscan to update Wikidata, but now after your comment I am inclined to think this might have to do with infobox logic. I agree we need to be careful how to enable re-use of our data without messing up our model of occupations. Jane023 (talk) 15:30, 10 July 2017 (UTC)
  Comment

We don’t really know this in so many cases anyway

We have such problem with creative jobs, that's why I'm asking if it make sense to mix full-time senators with part-time street guitarist :) d1g (talk) 14:01, 10 July 2017 (UTC)
I don't think so. All I am asking in this thread is to take the time to look. I also want to switch the genres to field-of-work, but that is a different subject (use of wrong property for same thing). Jane023 (talk) 15:23, 10 July 2017 (UTC)
  Comment
Since occupation (Q12737077) is described as "any activity of a person (hobby, work, pastime, professional sport...)", it seems like pastimes and hobbies are perfectly appropriate values for the occupation (P106) of a person, and are in fact required for a complete profile. Perhaps we need different properties for a person's work or career and their other interests, but it would be really difficult to know where to draw the line, even for many famous and well-documented people. Also, I think it's important to understand where people have come from. Yes, Harrison Ford was a carpenter when he was a struggling young actor, and we should acknowledge that as part of his biography. - PKM (talk) 20:27, 14 July 2017 (UTC)

Source without item

Succu has reverted my source because "comment (DEPRECATED) (P2315) is depricated". But what solution should I used to add a source which has not item yet? --Infovarius (talk) 01:04, 14 July 2017 (UTC)

Add the detail as a reference, you have enough ... "American Journal of Science 30(5): 138—141 (English)"  — billinghurst sDrewth 01:13, 14 July 2017 (UTC)
@billinghurst: Hm, with what property? --Infovarius (talk) 04:39, 14 July 2017 (UTC)
Help:Sources#Scientific, newspaper or magazine article. —MisterSynergy (talk) 04:50, 14 July 2017 (UTC)
...Why not just create the item? --Yair rand (talk) 01:15, 14 July 2017 (UTC)
Sometimes yes. But I don't want to think about notability of such works. --Infovarius (talk) 04:39, 14 July 2017 (UTC)
If the item is used as a reference, it has structural need and is therefore notable. —MisterSynergy (talk) 04:50, 14 July 2017 (UTC)
Exactly, we can have an item per every edition of the book. 09:41, 14 July 2017 (UTC)
  • I agree that having such a property would be helpful. The last time we discussed the idea it didn't found a consensus for creation but feel free to make a new proposal. ChristianKl (talk) 15:41, 14 July 2017 (UTC)
Which is incomplete, because you omitted the fact that the article was published in 5th series of American Journal of Science (Q465355) in volume 30. Thats why issue (P433) and volume (P478) are wrongly given. --Succu (talk) 21:56, 14 July 2017 (UTC)

Happy Birthday!

Many happy returns (if a little belatedly) to Wikidata:Property proposal/head coach of, our oldest unresolved property proposal, which was opened on 13 July 2016. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:45, 15 July 2017 (UTC)

Which Wikimedia project for historical data (time series)?

Is there a wikimedia project that hosts notable time series data? I am considering to provide the 622– nile maximum and minimum water levels at Kairo published 1925 by Omar Tousson, but only the maximum values can be found online. This time series is still used a lot in current research. Any suggestions? --DeWikiMan (talk) 12:54, 15 July 2017 (UTC)

There are two ways:
  1. Use the tabular data type on WikiCommons.
  2. Create a property proposal for minimum and maximum water levels and add it to Nile (Q3392). This solution means that we have to first agree on the best way to model the water levels in general. ChristianKl (talk) 13:18, 15 July 2017 (UTC)
Thanks, Christian. Will give your first suggestion a try. --DeWikiMan (talk) 13:27, 15 July 2017 (UTC)

Q33000000

Now reached this one.--GZWDer (talk) 12:31, 13 July 2017 (UTC)

Why is this worthy of noting? With so many duplicates created and needing to be deleted/merged, we can 'pump up' an insignificant number as something of value when it isn't. I think that the number of live items is a better measure that this 'faux' number.  — billinghurst sDrewth 01:16, 14 July 2017 (UTC)
We reached 519000000 edits. --Pasleim (talk) 07:46, 14 July 2017 (UTC)
How about counting the number of redirects already within these latest 3 million items due to the many duplicates introduced by the indirect geonames data import. Ahoerstemeier (talk) 09:12, 14 July 2017 (UTC)
At the moment there are 21912 redirects in latest 3 millions of items[5]. XXN, 15:22, 14 July 2017 (UTC)
Now 26909. In less than 2 days +5000 recent items were merged. --XXN, 00:51, 16 July 2017 (UTC)

Maybe we should just use Special:Statistics "content pages" figure (29,020,672 right now). That variable excludes redirects. Emijrp (talk) 11:08, 14 July 2017 (UTC)

  • The userpage at User:Succu has an interesting graphic illustrating the missing 4 million items (12%).
    --- Jura 14:44, 14 July 2017 (UTC)
ːYes, but it's six months old. Can somebody update this graph? Steak (talk) 14:59, 14 July 2017 (UTC)
And it does not mention the Q's that are never created. -- Innocent bystander (talk) 15:04, 14 July 2017 (UTC)
Done, Steak. --Succu (talk) 14:07, 15 July 2017 (UTC)

confusing date of birth

Can someone help me figure out best known date of birth for Italian painter Orazio Borgianni (Q2265356)? Many reliable sources (some printed and some online) have different dates. --Jarekt (talk) 17:29, 13 July 2017 (UTC)

Are you asking for which is the best reference with which to set a preferred date?  — billinghurst sDrewth 05:12, 14 July 2017 (UTC)
I guess so. My problem is that I am not sure how to deal with conflicting data, especially if it looks like someone's typo. The best thing to do would be to check all the printed sources, which I do not have. It is a case where more research is needed by someone, and we should probably tag it somehow as conflicting data. Another similar example would be date of birth for Russian painter Nikanor Chernetsov (Q15080902). My guess is that the date coming from Russian Wiki has a typo, but since it is backed up by printed sources it is hard to verify. I guess We should leave a note on Wikipedia projects about cases like that and hope that someone will look up printed sources. --Jarekt (talk) 16:59, 14 July 2017 (UTC)
Listing several dates with the normal rank is the best way we handle conflicting data. If there's one value that you consider most likely to be true, give it the preferred rank. If there are values that you believe to be false, depreciate them. ChristianKl (talk) 11:14, 16 July 2017 (UTC)

Uncertainty of population (P1082)

At the Dutch wp we use the population from wikidata not only as a value but also to express the density of an area by {{#expr: {{{population}}}/{{{area}}}}}. Since the P1082 value also uses the uncertainty of ±0, the expression gives an error. Is there a way to not include the uncertainty to keep the expression working? Have a nice day! Q.Zanden questions? 20:53, 14 July 2017 (UTC)

Why not simply remove the uncertainty from all values? Steak (talk) 21:31, 14 July 2017 (UTC)
I guess they are imported by bot automatically or with QuickStatements, and maybe because there is an uncertainty... Q.Zanden questions? 22:01, 14 July 2017 (UTC)
@QZanden: How is the values transported to nlwiki? If you use things like {{#property:P1082}} you will get the uncertainty and maybe also will get adapted decimal separators and thousand delimiter, all things that makes it impossible to use the data in the way you want yourself. If you instead use LUA, you can yourself decide if and how you want to use the uncertainty and wait with adapting to local language until after all the calculations have been done. -- Innocent bystander (talk) 06:58, 15 July 2017 (UTC)
±0 is relict from days when we worked around ±1 by explicitly typing "±0". The process of removing explicit zero bounds to no bounds is ongoing but I don't know at the moment what's remaining to be done.
If you are also using area (P2046) as {{{area}}}, please note that the property is not guaranteed to always have the same unit. Matěj Suchánek (talk) 07:44, 15 July 2017 (UTC)
And dividing with "38 163 hectar" is neither easy with the #expr:-parser. But that is also solvable with LUA, as long as the unit is specified and convertible to the units you want to use. -- Innocent bystander (talk) 08:07, 15 July 2017 (UTC)
@Innocent bystander, at the moment I am testing it here. You can see the problem at Cameron (Louisiana) where the Infobox states: Inwoners 406±0 (0 inw./km²). BoH (talk) 09:00, 15 July 2017 (UTC)
@BoH, QZanden: I have made some changes to the template now. But I do not know what happens if you add this template to an article that has two or no values in P1082. I think you have to look into that too before you implement this on a large scale! You probably have to ask the developers of the local Wikidata-module to see to those cases! -- Innocent bystander (talk) 09:21, 15 July 2017 (UTC)
@BoH, QZanden: And note that the main problem with the density here was not that the number included "Uncertainty" but that the value from Wikidata was never transported into the "#expr:"-parser! It was still the bevolkning-parameter that it was based on, and that was "nil" and nil divided by almost anything becomes 0. -- Innocent bystander (talk) 09:35, 15 July 2017 (UTC)
Thanks a lot! Your explanation is a bit too technical for me, but it works now.
Regarding different population numbers, if I look at New York City (New York (stad)) where a range of population data exists, removing the population data results in the 2015 data being used. So that seems to work fine.
One thing I haven't been able to get working properly however, is get the date of the population count in correctly. This edit by me is not the solution. Do you have an idea? BoH (talk) 10:33, 15 July 2017 (UTC)
@BoH: If you look into Q60#P1082 you there see that only one of all these New York-numbers has "preferred rank". So no matter which system you use, you aren't bothered by importing several numbers. When I made some experiments in your nl:Wikipedia:Wikidata/zandbak with code like {{#invoke:Wikidata|claim|p1082}} I get 77. So it only gives me one single number and that is good! The problem is that I want the latest number. But the module does not give me that, at least not by default. It gives me the first number, no matter if it is the oldest or youngest. I have no clue about how advanced your module is, if it supports giving you the latest number. And I do neither know if it can give you the date. It definitely looks like it can give you a reference, and that is really good.
Very much can be done with modules like this. In sv:Tjäll you see an article that get both population with date and reference, area with unit and reference, pop-density, coordinates and codes with reference, all from Wikidata. The template in sv:Lanna, Värnamo kommun gives you information of how big parts of the village of Lanna is located in Värnamo and Gnosjö, all from Wikidata. But it is a little tricky to solve it by templates like these. It would be better if the whole infobox was LUA-based. -- Innocent bystander (talk) 20:02, 15 July 2017 (UTC)
I'm afraid that is a bit too complicated for me, I have no clue about LUA. Thanks anyway! BoH (talk) 08:23, 16 July 2017 (UTC)
BoH, do you know anybody at nl.wikipedia who have such skills? Or are there anybody here with such skills? I have expanded the functions of the local module at svwiki, but I am afraid I am an self learned amateur who tends to create spagetti-code. So I am not the right person to help you there. (And I am a little too busy IRL at the moment.) -- Innocent bystander (talk) 09:00, 16 July 2017 (UTC)
There should be a few, I will try. Thanks again! BoH (talk) 13:05, 16 July 2017 (UTC)

Strategy discussion, cycle 3. A new challenge

Hi! It's the third week of our Cycle 3 discussion, and there's a new challenge: As Wikimedia looks toward 2030, how can we counteract the increasing levels of misinformation? You can suggest solutions here. Earlier challenges can be discussed as well. SGrabarczuk (WMF) (talk) 12:11, 16 July 2017 (UTC)

Please enlighten me! Why do we need separate Q nrs for all items in a biographical dictionary?

I just added the subject and then undid my edit for Pauw, Adriaen (Q27585665) because I apparently already did that before. I was looking for the person and this biographical item came up. I missed the memo on why these are here as separate items and not just listed on the subject's item under identifiers. Is this something to do with interwikilinks for Wikisource? There are so many biographical dictionaries out there in so many languages, and it doesn't feel to me like this is the right way to do this. I am probably late to the party though and this must have been discussed before, so just point me to the conversation, thx. Jane023 (talk) 06:40, 11 July 2017 (UTC)

We don't, but we do include sitelinks to Wikisource pages. The advantage is precisely that one can identify the subject of such an entry.
--- Jura 06:55, 11 July 2017 (UTC)
We don't what exactly? Are you saying this should be merged or not? The sitelink to wikisource for the subject in this case, namely the person Adriaan Pauw (Q367636), doesn't exist (yet), though I suppose I could make one as an example. I would make a wikisource page called "Author:Adriaan Pauw" and then in the wikisource author page I could list all articles about him and his treaty, do the same on Wikisource in all other probable languages like nl/de/fr and then on German wikisource the subject of this item Pauw, Adriaen (Q27585665) could link both to his author page and the Munster page (for the treaty). I guess my question has to do with how we want to deal with Wikisource in the long term: will all distinct ws pages be getting a Q number or not? I should think not. Jane023 (talk) 09:07, 11 July 2017 (UTC)
If there was no sitelink to Wikisource, the answer to "Why do we need separate Q nrs for all items in a biographical dictionary?" would be "we don't". Q27585665 needs mainly main subject (P921) not a sitelink to a Wikisource author page. Wikidata:WikiProject DNB more or less spells it out for another dictionary.
--- Jura 09:19, 11 July 2017 (UTC)
See my opening comment. I added main subject to that page twice already. My point it that this specific person (as many like him) have multiple encyclopedia articles written about him that could potentially already be on various Wikisource projects. My question remains therefore, is it correct to assume that each page on any wikisource will get its own Q number in future? Jane023 (talk) 10:01, 11 July 2017 (UTC)
With some exceptions, yes! "Q27585665" represents an article about Adiaen Pauw. It does not represent the person him/herself. Some pages in main namespace can share item with Wikipedia-articles. Among them are the "Author-pages". Some projects have a separate namespace for those, but for example dews have them in main namespace. Some pages are regarded as "Wikisource-disambigs" and they can correspond to normal Wikipedia-articles. For example a page linking to all versions of "Raven by Poe", listing all versions and translations of that specific poem. Those can be linked to the item about the poem. But the pages with the text of Poe should not be linked in such a way. Do not mix these WS-disambigs with other ordinary disambigs who lists everything named "Raven", no matter if it has a relation to Poe or not.
The notability criteria tells that subpages in main namespace in Wikisource not should be included here. But I doubt that those discussions are based on real facts about how Wikisource is organized. It is true that many such pages are only chapters in books. But such "chapters" can also be found elsewhere in Wikisource. And many of these chapter can stand alone as a text of its own. "ADB:Pauw, Adriaen (niederländischer Staatsmann)" is more or less a good example of that! -- Innocent bystander (talk) 12:41, 11 July 2017 (UTC)

I run into this issue before, where the only item on some person was a "biographical article" on Wikisource. I was not sure if it was OK to treat it like regular biographical item or not. In the end I created a regular item for that person, but it was confusing. --Jarekt (talk) 12:27, 11 July 2017 (UTC)

Thanks. I also find it very confusing, but mostly because ws is still so sparsely populated. For example, most Dutch encyclopedic articles of the 19th-century and earlier are not on Dutch wikisource at all, though this German dictionary article looks translated from the Dutch one. So in terms of "the future" I suppose we will eventually have items for each encyclopedia article that exist in each language on Wikisource, but not in other external sources, which is weird. I guess once there is more ws links in general it will make more sense to me. Jane023 (talk) 14:32, 11 July 2017 (UTC)
I also found it very confusing, and occasionally I am still perplexed when trying to figure out which is which. I assumed it was a relic of automatic uploading. Now, I just ignore it like I do entries for disambiguation pages from Wikipedia and other odd entries. --Richard Arthur Norton (1958- ) (talk) 23:57, 11 July 2017 (UTC)
  • There used to be a huge mess with DNB before Wikidata:WikiProject DNB: items had partial labels, hardly any description and/or statements. People kept merging or confusing them with items for people.
    If the Wikibase installation for Commons works out, it might be easier to do a similar installation for Wikisource. This way all its metadata could be stored at one place.
    --- Jura 09:44, 12 July 2017 (UTC)

  Comment @Jane023: There is no requirement for every entry in every compilation to have an item in Wikidata. There is the ability for biographical works to have an item at Wikidata where there is a (sub)page at a Wikisource. The "why" on how and why this would happen relates to 1) how we can directly link and reference an article at a WP or somewhere else, [think about a having a reference at a WP with a simple template link that calls all the data from WD] 2) to how we can produce/interlink metadata within WD for an article within a compiled work, [think about how we find and express information on the articles written by Joseph Foster (Q6283171) in many publications] and 3) how we can compile and express the components of a work by a query [think how WD can store and have queryable all the articles written in Dictionary of National Biography (Q1210343) by volume, or by edition, or produce a simple complete alphabetical, etc.]. I think that it is better to think of a biographical dictionary as just one type of compiled work, and the process that we have for compiled work is universal within wikidata, be it a bio dictionary, an encyclopaedia, a newspaper, a journal.  — billinghurst sDrewth 10:43, 12 July 2017 (UTC)

Interesting link, thanks! I guess I have always thought of this project as the ODNB, but I guess the ODNB is the successor to the DNB. Theoretically it is of course important to have this data so we can query it. I would be very interested for example if all the articles were completed with their metadata information, not only about their subjects, but also the wikidata items for their authors. This is important information for provenance of information in general and for studies on (for example) systemic bias. There should probably be similar projects per country (Netherlands: Van der Aa, USA: Britannica 1911, etc). Jane023 (talk) 11:55, 12 July 2017 (UTC)
the existing Q number for bio dictionary are an historical oddity, reflecting confusion on bibliographic metadata. you are right to be confused. - yes, ultimately we could merge all the bio articles under article subject - "described by source". the DNB is the PD version of paywalled ODNB, but they are different sources. (or we could leave bio article item as supporting reference for article subject). do we want to include Q number for every academic paper at wikisource or with a doi ? the DNB is above average, in that it gives us article author information, so article items could be kept for queries about the work. other bio dictionaries not so much. Slowking4 (talk) 12:41, 12 July 2017 (UTC)

@Jane023: Jane, the complete situation is like this: main subject (P921) properly links a "metadata item" like Q1234567 to another item like Q7654321, which is the true item on its subject. This has been envisaged since May 2015, when discussion concluded that, yes, all Wikisource items can have metadata pages here on Wikidata. I spent much time last year putting in such links for the old DNB.

What happens in practice is this: Q1234567 may exist, and its "main subject" cannot be added, because there is no such item about the subject person. Then either the situation should be left as is; or a new item created. Data about the _person_ should anyway not be added to Q1234567, which is marked instance of "biographical article" or suchlike. Charles Matthews (talk) 08:36, 13 July 2017 (UTC)

Yes Charles I got all of that. I am just wondering whether separate Q numbers are really useful if they are not going to be modelled the way I suggested. Just linking ws pages seems so useless. Either you want the whole thing because you are going to actively use it somehow as a dataset (for studies on systemic bias - e.g. how many articles are based on people from London vs other places, etc), or you can better turn it into a property that links properly to various sections of ws. If I look at the growth of the English Wikisource for the Britannica 1911, it seems pretty sad that it's not completed when there is only 65,000 articles in total and this was supposedly what English Wikipedia was seeded with at its inception. That means that if you would even try to do what you have done with the DNB you would first have to finish the 65,000 articles on English ws source first. If I compare this work to what I am doing with the art catalogs of Hofstede de Groot then it seems similar datamodelling work is going on. The difference is that if I only add the subject items, which are "instance of painting" and not "instance of article about painting". If I can't get sources for the painting then I don't add that painting to Wikidata as an item. I will only include catalog numbers that have an image or collection metadata. Otherwise it could just be a double entry or a flat-out mistake. So to my mind, it would have been better to keep the whole dataset on ws and just link the ws article to subject items only. The problem is with Wikisource and the ability to link articles about a topic to each other. In the case that I started with, namely a notable Dutch "grand pensionary" who happens to be poorly documented on wiki projects (so far), linking one encyclopedia article to him from one ws instance (german ws) as a separate item just seems so random to me. If you had a "topic" namespace on Wikisource that was similar to the "author" namespace, then you could interlink such articles within the same ws instance, as well as link out to other language ws projects and the proper wikidata items. I guess I am in disagreement with the current vision for the future of ws and wikidata in this sense. Jane023 (talk) 09:29, 13 July 2017 (UTC)
well, the EB1911 biographical articles are "finished" (not proofed or validated) and only 2,849 wikipedia articles to reference. (incorporating a citation from the 1911 Encyclopaedia Britannica with no article parameter) "random" is the wiki way. tell me more about the vision. Slowking4 (talk) 23:34, 17 July 2017 (UTC)

TinyURL

I think it is strange that we encourage the use of tinyurl.com to link to Wikidata Query Services (it's the short URL produced on the Query Service page!), and then block it from entering it into Wikidata. I couldn't use it on my user page to bookmark a search I produced! While I understand that shorteners shouldn't be used in the main wikidata space, the question is: Is there a possibility to exclude user and talk pages from the block? --Anvilaquarius (talk) 08:53, 17 July 2017 (UTC)

See Wikidata:Contact the development team/Archive/2017/02#Query result shortlink is a tinyurl... and it's forbidden in Wikipedia.--Jklamo (talk) 09:30, 17 July 2017 (UTC)

Improving documentation during the Wikimania hackathon

Hello all,

As you may know, the international conference Wikimania will take place in August in Montreal. Before that, two days of hackathon will allow people to work on any Wikimedia-related project. Just like during the Wikimedia Hackathon, we will organize a documentation sprint, where everyone can come and help improving documentation pages on any topic.

We suggested a few tasks related to Wikidata:

If you have any other ideas, feel free to create a comment or a subtask of the main task.

The hackathon will take place on August 9th and 10th. Of course, you're welcome to participate remotely, and continue working on the projects after that! Lea Lacroix (WMDE) (talk) 08:54, 17 July 2017 (UTC)

@Lea Lacroix (WMDE): - Great to hear it! As one of the organizers of Wikiconference North America pre-conference (happening at the same time, Aug 10) we were thinking of doing Wikidata training with our North American constituents, including showcases of how to use Quickstatements, Google Spreadsheets, OpenRefine. Perhaps we could join forces here, even if it's just for 90 minutes or so in the day. -- Fuzheado (talk) 15:29, 17 July 2017 (UTC)

Please help fix

Wikidata as usual is not making it easy for me. I linked en:Elwro to pl:Elwro and cannot undue it, it's a mistake en:Elwro should link to pl:Wrocławskie Zakłady Elektroniczne Mera-Elwro instead. Thx --91.226.39.108 11:52, 17 July 2017 (UTC)

  Done I have moved the link from Elwro (Q9253030) to Elwro (Q3051868). --Sintakso (talk) 12:35, 17 July 2017 (UTC)

Wikidata weekly summary #269

Merged entities IRL with funky Wikidata results

Not sure what the proper venue for this is, so pointers for that would be appreciated.

I just ran across some funkyness that I don't know how to fix.

What seems to be the IRL background, is that the French commune w:Sacy, Yonne was merged into w:Vermenton (on 1. Jan. 2016). Through whatever process this change has been reflected here as the original Wikidata item for Vermenton (Q1422869) (and possibly Sacy too, I haven't checked) being changed to some kind of "former" status, and a new Wikidata item Vermenton (Q28732226) created for the new combined entity. Two issues with this:

First, it seems, to me, to be wrong. If Sacy was merged into Vermenton, as the enwp description suggests, then the Vermenton entity didn't really change its identity (it just expanded scope). Sacy did, so that probably does need some kind of "former" status.

Secondly, the new Wikidata item for Vemonton lacks a description (which was how I noticed; a Q-code was displayed in a Commons Creator template that pulls from Wikidata), iw links, and a bunch of the properties that are on the old Wikidata item.

In any case, for my purposes I could just copy and paste the description between items, but as best I can tell the above reflects an actual (not just cosmetic) problem with the data. And since I have no idea how to fix that (or indeed what the correct state is for this situation) I figured I'd drop a note here so someone with more Wikidata-fu can sort it out. --Xover (talk) 14:53, 17 July 2017 (UTC)

Item pages not loading fully

For the last half hour or so, I've been unable to load item pages properly; I get no "edit" links, with the occasional exception. I've tried re-starting my browser, and then my machine, to no effect. I have the latest Firefox (54.0.1), under Win 10, on a powerful Dell XPS. How can I fix this? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:40, 21 July 2017 (UTC)

I bet this has something to do with phab:T170668. Matěj Suchánek (talk) 14:12, 21 July 2017 (UTC)

One of the affected items is Andrea Leadsom (Q750091). How does it look to other editors? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:12, 21 July 2017 (UTC)

"no edit links" - got this bug yesterday with Firefox 54.0. But right now everything works fine with Q750091. Firefox seems to be 2-4 times slower in general with Wikidata than Chrome. d1g (talk) 18:22, 21 July 2017 (UTC)
Thanks. I get the same symptoms, including on that item, in Chrome. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:31, 21 July 2017 (UTC)
I can not edit d:Q30426860 (creator link has to be removed). --Jarekt (talk) 18:38, 21 July 2017 (UTC)
Same here. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:43, 21 July 2017 (UTC)

As noted in the Phabricator discussion, purging the item fixes it - but that is obviously not a sustainable solution. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:46, 21 July 2017 (UTC)

It's a workaround till the patch is reviewed, merged and deployed. Mbch331 (talk) 20:38, 21 July 2017 (UTC)
You clear the cache by adding "?action=purge" after the URL.. Thanks MisterEnergy for your support on IRC. Thanks, GerardM (talk) 08:07, 22 July 2017 (UTC)
Thanks for sharing hints - we're going to have a look as soon as possible. Lea Lacroix (WMDE) (talk) 13:49, 22 July 2017 (UTC)
This should be fixed now.
This section was archived on a request by: Matěj Suchánek (talk) 13:23, 24 July 2017 (UTC)

No add button

Does anyone know why there is no add button for the item Ana Cartianu? Thanks in advance. Mycomp (talk) 13:06, 22 July 2017 (UTC)

Try adding ?action=purge at the end of the link in the url bar of your browser and see if you can edit it then. If so, then it's a know bug and will be fixed soon. Mbch331 (talk) 14:18, 22 July 2017 (UTC)
Thank you, it worked! Have a nice day! Mycomp (talk) 14:20, 22 July 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 13:23, 24 July 2017 (UTC)

Cannot edit Karlie Kloss (Q292749)

But it doesn't seem protected. Can anyone explain why? —Justin (koavf)TCM 23:35, 22 July 2017 (UTC)

@Koavf: See the "Item pages not loading fully" section above. Try adding ?action=purge on the end of the URL, the edit links should show then. Thanks. Mike Peel (talk) 23:40, 22 July 2017 (UTC)
@Mike Peel: Good man. —Justin (koavf)TCM 00:35, 23 July 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 13:23, 24 July 2017 (UTC)

Move gadget is not working

When using "Move" gadget, after entering the Q ID and pressing the "Move" button, it just keeps loading without doing anything. I am using Firefox 54.0.1 (32 bit) in Windows 10. --Yiwucat (talk) 02:30, 23 July 2017 (UTC)

Try a refresh. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 02:46, 23 July 2017 (UTC)
That's not helping, see phab:T171401. You should have received a notification about that, due to stalking all Wikidata related tasks (and every task created by me!) with a Herald rule. Sjoerd de Bruin (talk) 17:32, 23 July 2017 (UTC)
I had this problem yesterday. I was able to delete the site link from the current Q item and add it to the new one in two steps without any issues. You might try that as a temporary solution. - PKM (talk) 23:56, 23 July 2017 (UTC)
  Fixed Matěj Suchánek (talk) 07:19, 24 July 2017 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 07:19, 24 July 2017 (UTC)

Missing "Add statement" button

No "Add statement" button appear at Q11394795. Nor at Maruyama school (Q11394770) nor Q11599058. All other items I have tried had the button.

I tried with Firefox and Chromium. Syced (talk) 10:03, 24 July 2017 (UTC)

Try it with ?action=purge in the address bar of your browser for these items. Mbch331 (talk) 10:28, 24 July 2017 (UTC)

Same problem, purge doesn't help. Tracking now. --Magnus Manske (talk) 12:46, 24 July 2017 (UTC)

This section was archived on a request by: Matěj Suchánek (talk) 13:23, 24 July 2017 (UTC)

Some updates about Recent Changes page

Hello!

Sorry to write in English. Please help translate to your language is needed and share it with your language community.

As you may already know, the Collaboration team has created a Beta feature. This feature is on your wiki since few months: "⧼eri-rcfilters-beta-label⧽". You can activate it in your Beta preferences.

What is this feature again?

This feature improves Special:RecentChanges and Special:RecentChangesLinked. It adds new features that ease vandalism tracking and support of newcomers:

  • Filtering - filter recent changes with easy-to-use and powerful filters combinations.
  • Highlighting - add a colored background to the different changes you are monitoring. It helps quick identification of changes that matter to you.
  • Quality and Intent Filters - those filters use ORES predictions. They identify real vandalism or good faith intent contributions that need help.

You can know more about this project by visiting the quick tour help page.

What's new?

Since the release, we have fixed small bugs and improved the interface. We have also released a way to bookmark your favorite configurations of filters.

We plan to add more new features! The full list is on this Phabricator page (in English) but here are the most important ones:

  • Filters for Namespaces and tagged edits and, later, filters for Categories and Usernames
  • Redesign navigation by using an improved time selector and better integrated navigation options
  • Add live updates
  • Have a more clear interface by putting community-defined 'related links' into a collapsible panel

This last change is about links displayed on top of the RecentChanges (see how they look like on your wiki). We do that change because we have discovered that those links are not that much used. Also, they sometimes take a lot of space on small screens. To help people to focus on recent changes patrolling, we will hide those links, with an option to show them. We have created some examples to show you how it will look like. If you like to see those links all the time, you will have to click on the link to show them and they will remain open. We welcome your feedback about this change.

You can ping me if you have questions. If this message is not at the right place, please move it and tell me. :)

Most of this information was already introduced in the Collaboration monthly newsletter. Please subscribe to get regular updates!

All the best, Trizek (WMF) (talk) 09:27, 6 July 2017 (UTC)

Recent changes slow

Hello. I don't know if it is related to the new filters, but I have been experiencing very a high loading time recently. Like 30 seconds or more when I click in Special:Recentchanges. It is the only slow page for me in Wikidata. Emijrp (talk) 10:14, 6 July 2017 (UTC)

not only has it been REALLY slow for me also, but just now the Recent changes page generated a "Database error": "[WV@gGApAIDcAACNsuSwAAABI] 2017-07-07 14:53:09: Fatal exception of type "Wikimedia\Rdbms\DBQueryError"" - I've never seen that in wikidata before. Maybe this "Recent changes" thing should be rolled back and reconsidered for wikidata? I think what we need here may be different in character from what other wikimedia sites need. ArthurPSmith (talk) 14:59, 7 July 2017 (UTC)
Hello
Please note those filters provide many more combinations than the default RC page does, and loading time for some combinations may be longer for certain combinations than others. It is especially the case when the prediction and intention filters are involved. Emijrp, can you send me the link(s) that give you the longest loading time (the full URL in your browser bar)?
ArthurPSmith, do you still have this issue? I don't have it. Do you have it when you click on this link? The link uses the safemode that removes additional scripts which may potentially be in conflict with the filters.
Thanks, Trizek (WMF) (talk) 15:45, 7 July 2017 (UTC)
Trizek (WMF) - the database error just happened the one time. But Recent changes is slow for me on wikidata all the time now - with or without your special safemode link it takes at least 30 seconds to display anything. Even if I select "Show new changes since ..." and there are only a handful of changes it takes that long. If I select any of your filter options (say "Unpatrolled"), I have to wait at least 30 seconds to see the results. It's pretty un-useable at the moment. ArthurPSmith (talk) 15:57, 7 July 2017 (UTC)
I'm sorry it is un-usable for you at the moment, ArthurPSmith. I'll report this problem sor that we can improve that Beta feature. I have tried with "Unpatrolled" and I have the same loading time as yours. Trizek (WMF) (talk) 16:03, 7 July 2017 (UTC)

@Trizek (WMF): Your link is slow for me too, it takes about 30 seconds, and I have got similar database errors to User:ArthurPSmith. I am in Spain, just in case it depends on region. I have had this issue for a week or two, I don't remember. I stopped checking recentchanges because it. Emijrp (talk) 16:42, 7 July 2017 (UTC)

Thanks, Emijrp. We are investigating. If you have the problem with some other sets of filters, please telle me. Trizek (WMF) (talk) 14:13, 11 July 2017 (UTC)
I haven't tried it for a few days, but today Recent changes is quite speedy for me, at most a 1 or 2 second wait. ArthurPSmith (talk) 17:54, 13 July 2017 (UTC)

@Trizek (WMF), ArthurPSmith: Today it is fast for me too. Emijrp (talk) 10:40, 15 July 2017 (UTC)

Emijrp , a patch has been released on July 12. :) Trizek (WMF) (talk) 13:54, 18 July 2017 (UTC)

Fetching QIDs through SPARQL?

Hi all. I have a very daft question: how can I fetch QIDs through SPARQL? Doing a query like "SELECT ?item WHERE { VALUES ?val {wd:Q33093130} ?item wdt:P31 ?val }" returns a wikilinked value for ?item, rather than the raw QID. I can't spot an example that does this at Wikidata:SPARQL query service/queries or through google (but this is a difficult thing to search for!). Thanks. Mike Peel (talk) 23:19, 16 July 2017 (UTC)

I'm not sure if there's a simpler approach, but one way would be cast the value to a string to get the full URI:
SELECT (STR(?item) AS ?qid) WHERE { 
  VALUES ?val {wd:Q33093130} 
  ?item wdt:P31 ?val 
}
Try it!
If you just need the final part, then you could also use REPLACE (again, not sure if there's a more suitable URI-processing function):
SELECT (REPLACE(STR(?item), 'http://www.wikidata.org/entity/', '')  AS ?qid) WHERE { 
  VALUES ?val {wd:Q33093130} 
  ?item wdt:P31 ?val 
}
Try it!
--Oravrattas (talk) 06:08, 17 July 2017 (UTC)
I routinely use
SELECT (STRAFTER(STR(?item), 'http://www.wikidata.org/entity/')  AS ?qid) WHERE { 
  VALUES ?val {wd:Q33093130} 
  ?item wdt:P31 ?val 
}
Try it!
for that Jneubert (talk) 06:42, 17 July 2017 (UTC)
I use this approach but save some typing through STRAFTER( STR( ?item ), STR( wd: ) ). Matěj Suchánek (talk) 07:48, 17 July 2017 (UTC)
And I typically use replace(str(?item), ".*Q", "Q"). - Nikki (talk) 06:53, 17 July 2017 (UTC)
I've created phab:T170788 asking for a more direct method. It seems like enough people would use it. - Nikki (talk) 07:35, 17 July 2017 (UTC)
  • Some also use substr .. I'm curious which function is the most efficient one.
    --- Jura 11:20, 17 July 2017 (UTC)
Thanks for the suggestions! It would be good if this could be documented somewhere so it doesn't get lost in the archives (since it doesn't look like the phabricator ticket will lead to a change with this any time soon). Thanks. Mike Peel (talk) 12:21, 18 July 2017 (UTC)

A@Wikipedia:Talk page guidelines (Q5641607) and Help:Talk pages (Q4592157)

ro:Wikipedia:Pagină de discuție is listed in the "Help:Using talk pages", but the Romanian page is marked as a guideline. Shouldn't it belong to the list of "Wikipedia:Talk page guidelines"? Also, some others listed in the "Help:" query page should belong to the "Wikipedia:" query page. Must I be bold to change the lists, i.e. move some items to the other query? --George Ho (talk) 06:44, 11 July 2017 (UTC)

sv:Wikipedia:Diskussionssidor is a guideline, but only for the talk_pages of the main namespace. Talk_pages for user_talks and places like the Village pump is not directly involved. I doubt there are so much strong opinions here at Wikidata. These items do not look very useful for the database we have main interest in. -- Innocent bystander (talk) 12:25, 11 July 2017 (UTC)
Sorry for belated reply. Anyway, can changes to the queries be done right away? If not, what else shall be done with these queries? --George Ho (talk) 10:52, 19 July 2017 (UTC)

Difference between P1943 and P242

It seems that location map (P1943) and locator map image (P242) are quite the same, and are being used elsewhere to state the same. See their talk pages and, for example, Gansu (Q42392) -Theklan (talk) 14:06, 19 July 2017 (UTC)

location map (P1943) is for inside the boundaries of the place, while locator map image (P242) should show the location of the place within the higher administrative unit. Sjoerd de Bruin (talk) 15:42, 19 July 2017 (UTC)
Commons has a good description of the difference. Mahir256 (talk) 15:48, 19 July 2017 (UTC)

Restore please

Q33082442 and Q20669522 were deleted as not notable. Can they be restored, and can we see if the person responsible, User:ValterVB deleted other records under the same premise. --Richard Arthur Norton (1958- ) (talk) 17:03, 15 July 2017 (UTC)

I don't see any references or identifiers on the deleted item, though. Sjoerd de Bruin (talk) 17:25, 15 July 2017 (UTC)
Restore it so that I can see it please. It should have a link to Wikimedia Commons. It may have been deleted during the filling in process. --Richard Arthur Norton (1958- ) (talk) 17:56, 15 July 2017 (UTC)
The history shows that there were never any sitelinks added. Mbch331 (talk) 18:06, 15 July 2017 (UTC)
That is awesome, can I still have it restored so I can see it. Wikimedia sitelinks is only one of multiple reasons for having an entry. Lack of sitelinks is not a valid reason for deletion. Am I at the wrong venue for undeletion requests? --Richard Arthur Norton (1958- ) (talk) 19:24, 15 July 2017 (UTC)
I think Mbch331 misinterpreted you! You said that it was deleted before you had the time to add a sitelink, and Mbch331 replied by telling it had no sitelinks. The correct place to ask for a restore of a deleted item is WD:AN, but if it solved here, that is no big deal. Personall,y I do not see what makes it so difficult to restore deleted versions of a simple item. If it wasn't what you thought it was, it can easily be deleted again. -- Innocent bystander (talk) 19:34, 15 July 2017 (UTC)
Per WD:N, a single sitelink to Commons does not make an item notable. The item never had one and it's final version didn't have any references and any external identifiers. Matěj Suchánek (talk) 19:37, 15 July 2017 (UTC)
  • Actually WD:N reads the opposite: "It contains at least one valid sitelink to a page on Wikipedia, Wikivoyage, Wikisource, Wikiquote, Wikinews, Wikibooks, Wikidata, Wikispecies, Wikiversity, or Wikimedia Commons."  – The preceding unsigned comment was added by Richard Arthur Norton (1958- ) (talk • contribs).
    Read further: In addition, an item with only a sitelink to a category page in Wikimedia Commons is not allowed on main article items. However, it is allowed to link Wikimedia Commons categories with categories in other Wikimedia sites in items. Matěj Suchánek (talk) 20:06, 15 July 2017 (UTC)
    What are "main article items"? Wikidata doesn't have prose articles, what does that even mean? What does "categories with categories in other Wikimedia sites" mean? I read 10 times, and still have no clue what it means. --Richard Arthur Norton (1958- ) (talk) 21:00, 15 July 2017 (UTC)
Conflicted 2 time :) The item was without references, have you a source that indicates that it is the person you say? So I can restore and add reference. --ValterVB (talk) 19:39, 15 July 2017 (UTC)
If we must delete unreferenced entries, I will write a script to search for them. If you show me the rule, I will start work on the script and we can begin massive deletion within the hour. Also show me the rule that requires an external identifier. --Richard Arthur Norton (1958- ) (talk) 19:59, 15 July 2017 (UTC)
Already exist: is Wikidata Query Service but we can't delete automatically, it's necessary check every item before delete them. --ValterVB (talk) 20:01, 15 July 2017 (UTC)
@Richard Arthur Norton (1958- ): One modify suggestion: It contains at least one valid sitelink to a page on any of our supported Wikimedia Projects. --Liuxinyu970226 (talk) 03:44, 20 July 2017 (UTC)
  • Check what? You just said we have to delete unreferenced records and the search finds unreferenced records, what is the holdup? Get deleting. --Richard Arthur Norton (1958- ) (talk) 20:04, 15 July 2017 (UTC)
Every item I have created have been empty when I start to edit them. What makes it so difficult to restore it so it can be filled with more claims and sitelinks? -- Innocent bystander (talk) 20:06, 15 July 2017 (UTC)
  • Then delete all unreferenced entries greater than 12 hours old. A simple fix!
(After multiple edit conflicts) And the rule is WD:N. It didn't have sitelinks, so it fails #1. No links to the item (otherwise it wouldn't have been deleted), so fails #3. So remains #2. There weren't any references, so how can an admin know if he can be described using serious and publicly available references? With the number of DR's it's not doable to check every item if there might be publicly available references. So that information has to be present somehow in the item and that wasn't the case. Mbch331 (talk) 20:10, 15 July 2017 (UTC)
WQS can't show if someone removed valid site links or valid references, that's what needs to be checked. Mbch331 (talk) 20:10, 15 July 2017 (UTC)
Innocent bystander: I can't see Richard saying he wants to add information, just to see if the deletion was valid. Mbch331 (talk) 20:10, 15 July 2017 (UTC)
Re on "Check what?": Before deleted an item is necessary check if someone deleted valid sitelink, or someone moved sitelink without gadget, so we must search destination item and do merge, in some case we must check backlink (sparql can check only ns0). --ValterVB (talk) 20:16, 15 July 2017 (UTC)
Re on "Then delete all unreferenced entries greater than 12 hours old. A simple fix!" Do you understand "We must check?" It isn't a problem of time. --ValterVB (talk) 20:40, 15 July 2017 (UTC)
  • OK, I gave up and recreated them as Q33082442 for Q20669522(I cannot find the original number, it gets deleted from my contributions) and Q33082485 for Q20669522, someone can merge the data fields later for me. Twice the work, but it will make contributors to this thread happier. Thanks! --Richard Arthur Norton (1958- ) (talk) 20:44, 15 July 2017 (UTC)
Do you have read " have you a source that indicates that it is the person you say? So I can restore and add reference"? --ValterVB (talk) 20:48, 15 July 2017 (UTC)
Please can you read all this page and avoid to create not notable item or item with wrong claim? How is possible that value of spouse (P26) is a category item? --ValterVB (talk) 20:54, 15 July 2017 (UTC)
Wikidata notability: "can be described using serious and publicly available reference" How does it fail this? If someone had Wikidata notability they can be concatenated to other family members that we have fields for. That is why Wikipedia does not allow articles on presidential children if not independently notable, yet Wikidata does because they "can be described using serious and publicly available reference", a census entry is a "serious and publicly available reference" as is a marriage certificate or a death certificate. --Richard Arthur Norton (1958- ) (talk) 20:56, 15 July 2017 (UTC)
  • Again, is this a strict rule, that we delete unreferenced entries, or is it just used ad hoc to delete entries that people with admin rights do not like? --Richard Arthur Norton (1958- ) (talk) 21:07, 15 July 2017 (UTC)
I repeat my self: Do you have read " have you a source that indicates that it is the person you say? So I can restore and add reference"? --ValterVB (talk) 21:09, 15 July 2017 (UTC)
  • Answer my question first: Is this a strict rule, that we delete unreferenced entries, or is it just used ad hoc to delete entries that people with admin rights do not like? --Richard Arthur Norton (1958- ) (talk) 21:54, 15 July 2017 (UTC)
Can someone do a quick count on how many articles would be deleted if we followed the rule of deleting unreferenced Wikidata entries? --Richard Arthur Norton (1958- ) (talk) 21:54, 15 July 2017 (UTC)
15,000±5,000 (my estimation). Matěj Suchánek (talk) 07:56, 16 July 2017 (UTC)
7174 items about people would be deleted [6], that are 0.2% of all our items about people. --Pasleim (talk) 20:17, 16 July 2017 (UTC)
I restored the item. The marriage certificate hosted on WikiCommons is clearly a serious and public source. ChristianKl (talk) 22:17, 15 July 2017 (UTC)
Thanks! Can you restore Q33082442 I mistyped it last time. Can you also respond to the wording below. We have few eyes on here as compared to Wikipedia so rules creep in without any challenge. --Richard Arthur Norton (1958- ) (talk) 22:29, 15 July 2017 (UTC)
Currently, the Q33082442 reads "Category:Lena Elaine Olson" and is instance of (P31) Wikimedia category. While I consider an item with instance of (P31) human (Q5) for Lena Elaine Olson to be notable, an item for the category isn't. ChristianKl (talk) 22:55, 15 July 2017 (UTC)
He just makes the situation even more confusing by editing messages hours later while other people responded to them already. Highly unwanted. Sjoerd de Bruin (talk) 23:05, 15 July 2017 (UTC)
Of course restoring it like I asked would have solved all the problems and amounted to 1/10 of the work. --Richard Arthur Norton (1958- ) (talk) 02:25, 16 July 2017 (UTC)
  • Can someone please merge the new with the old, all outside links that are pointing in to Wikidata now use the new Q numbers. Thanks. I changed the Q numbers pointing in. --Richard Arthur Norton (1958- ) (talk) 23:46, 15 July 2017 (UTC)
Was it so hard? If when I asked to you "have you a source that indicates that it is the person you say? So I can restore and add reference" (my first answer) you answered: "Yes, here: https://www.geni.com/people/Andrew-Jensen/6000000006341725132" You saved a lot of words. --ValterVB (talk) 07:29, 16 July 2017 (UTC)
I can cut and paste too. Is it still impossible to answer? (third time asking) Is this a strict rule, that we delete unreferenced entries, or is it just used ad hoc to delete entries that people with admin rights do not like? --Richard Arthur Norton (1958- ) (talk) 15:17, 16 July 2017 (UTC)
If point 1 and point 3 are not applicable and I can't check if point 2 isn't respected I delete item, so: yes, for me is a rule. --ValterVB (talk) 15:49, 16 July 2017 (UTC)
i think we need a consensus to establish a rule. Slowking4 (talk) 23:43, 17 July 2017 (UTC)

Notability

Can someone explain what this means, it is currently being used as a rationale for deletion but it makes no sense as written: "In addition, an item with only a sitelink to a category page in Wikimedia Commons is not allowed on main article items. However, it is allowed to link Wikimedia Commons categories with categories in other Wikimedia sites in items." Can someone find the discussion that took place to add this to WD:Notability? Wikipedia has "articles", Commons hosts files, Wikidata hosts Q entries that contain other property (P) fields. I think it was added at some point to disallow Q entries for Wikicommon entries that do not have Wikipedia articles or WikiQuote or Commons gallery entries, so the wording about "articles" makes no sense. If correct it should read: "Wikidata entries are not allowed for Wikimedia Commons categories if the subject does not contain an entry in Wikipedia or WikiQuote or WikiSource." Someone else will have to figure out the second part. Anyway, it contradicts the very first rule of notability, if my interpretation is correct. Many category entries at Commons do not have a corresponding "Commons gallery" page, because no one bothered to make one. I never bother to, even when adding a dozen files on a notable subject. --Richard Arthur Norton (1958- ) (talk) 22:01, 15 July 2017 (UTC)

Wikipedia not only has articles but also Category pages and Disambiguation pages. Both of those are notable enough to warrant their own Wikidata items. Category:Germany (Q1410828) is for example no article but a category. Given that the category has sitelinks to Wikipedia it's notable, if it would only have sitelinks to WikiCommons that wouldn't be enough for notability. ChristianKl (talk) 22:37, 15 July 2017 (UTC)
Where is the discussion for including this as a notability rule, and the discussion that determined the wording? --Richard Arthur Norton (1958- ) (talk) 04:07, 16 July 2017 (UTC)
The problem with the notability of Commons sitelinks in my opinion is that you there can find categories and galleries like Files uploaded by User:X, who looks community-centric, instead of contributing to the knowledge base we want to build here. (This person maybe is notable by other reasons, but that is another question.) I therefor do not think that every category or gallery on Commons is a good base for notability. But in many cases I think they are, like this random piece of musical instrument. That does not mean that it is a good idea to mass-import by bot such items. Instead they can be imported when the need is found. -- Innocent bystander (talk) 06:55, 16 July 2017 (UTC)
Wikidata:Requests for comment/Commons links, Wikidata talk:Notability/Archive 3#Are Commons categories notable by themselves? Matěj Suchánek (talk) 07:41, 16 July 2017 (UTC)
That is a good reason to surgically exclude those specific self-referential categories about contributors. Currently the rule is so poorly written it can used as a bludgeon to delete anyone with only a biographical category at Commons. Lets close the loophole by rewording that sentence, and lets have a genuine discussion, before we have a major rule change. --Richard Arthur Norton (1958- ) (talk) 14:39, 16 July 2017 (UTC)
Currently c:Category:Solidago canadensis is not linked as a sitelink to any Wikidata item, despite it being a category for Solidago canadensis (Q254436). That's because c:Solidago canadensis is linked to that item instead. Am I to understand the notability rules correctly, that I am not allowed to create a new Wikidata item for that Commons category, since there is no other sitelink for the same category? (tJosve05a (c) 15:21, 16 July 2017 (UTC)
Yes. Matěj Suchánek (talk) 16:23, 16 July 2017 (UTC)
+1 --Succu (talk) 20:42, 16 July 2017 (UTC)
And for that use we also have Commons category (P373) where you can add the category instead of putting it in the sitelink-section. Q.Zanden questions? 23:26, 16 July 2017 (UTC)
... which means that sitelinks aren't included at the commons category. :-( Mike Peel (talk) 23:29, 16 July 2017 (UTC)
Maybe I am misreading the statement above, however, I dispute QZanden's suggestion that the means to progress is to exclude categories as sitelinks for Commons, or that this is the consensus that has been reached. I will accept that where a gallery exists, it precedes a category as the interwiki, however, where no gallery exists, I firmly believe that a category link should be used as the interwiki. 1) The interwikis existing are valuable and easily usable; 2) There will be so many categories that should never have galleries at Commons, and as such the category interwiki serves a strong function. I would always have the category added using P373 whether I have it interwiki'd or not.  — billinghurst sDrewth 01:52, 17 July 2017 (UTC)
@billinghurst, no I meant that if there are both a page and a category at commmons, then you put the page as sitelink and the category in P373. If there is no page, then put the category in the sitelinks. Q.Zanden questions? 12:45, 17 July 2017 (UTC)
The rule is footnoted to a discussion, but that discussion was only about creating a field to hold a commons category. I do not see consensus in the discussion for disallowing the creation of a Wikidata entry, or the deletion of a Wikidata entry, because it only links to a commons category. Many biographical entries only have a link to a commons category and not a commons gallery. This is because it is a lot of work just to create a commons gallery, just so you can sort the images chronologically instead of alphabetically. The discussion here that is cited above is also nebulous. I think the wording has to be changed, since it is being used as a basis for deletion or at least for not restoring entries, that have previously been deleted. It is also being arbitrarily enforced, which is never good. We either have to change the poorly worded rule, or purge all the entries that only have a link to Commons category. I think what the rule was meant to say was to not automatically create entries for every Commons category because they contain a few category types not useful for Wikidata. --Richard Arthur Norton (1958- ) (talk) 17:00, 17 July 2017 (UTC)
If item is notable, we delete sitelink, not the item. --ValterVB (talk) 18:51, 17 July 2017 (UTC)
Can you show me where this discussion took place about deleting Commons Category links. Is this occurring now? --Richard Arthur Norton (1958- ) (talk) 00:01, 18 July 2017 (UTC)
  •   Comment well I am confused about what was the point of the conversation, and it seems that there is a whole lot of cross-purpose conversation. All it points to me is that we do need to better explain Commons linking. 1) The requirements for main ns item, and what items it includes, and then what interwiki links, and in preferential order of linking; 2) the requirements for Category: creation, the body items, and the interwiki links, again in preferential order. This help should be based on the outcomes of discussions, as having users have to pull up each page to know the protocol is antediluvian. Is this a call for [[Wikidata:For Commoners]] or some such name ?  — billinghurst sDrewth 01:37, 20 July 2017 (UTC)

Units with Quickstatements

I asked a user at sv.wikipedia to do some edits with Quickstatements. He reports that he has problems adding units to area (P2046) with the new version. Using the old one is not much of an options since it has other flaws, even if it looks like it works to add units with it. How could this still a problem? -- Innocent bystander (talk) 08:52, 18 July 2017 (UTC)

Seems that Magnus haven't found time to fix that yet. Matěj Suchánek (talk) 09:07, 18 July 2017 (UTC)
Darn, I was thinking of using Quickstatements to add some area information for towns in Vermont from the US Census. Jc3s5h (talk) 10:20, 18 July 2017 (UTC)
I hope Magnus will find his way to review and merge this. Matěj Suchánek (talk) 10:28, 18 July 2017 (UTC)
@Jc3s5h: I skipped waiting för the bugfix, and created this listeria-page who list all (for me) relevant pages with missing units. All 800 pages will not be fixed today, but in due time... -- Innocent bystander (talk) 14:05, 19 July 2017 (UTC)
Thanks for the comment, but my main interest is being able to add areas, together with the appropriate units, as stated in a reliable source. I have reviewed the majority of the towns, and so far, either the area is absent or it is stated in appropriate units. Finding and identifying quantities that lack an appropriate unit of measure is not my area of interest. Jc3s5h (talk) 20:43, 19 July 2017 (UTC)

Wikimedia Commons URL considered invalid

I get an 'invalid character' error trying to add https://commons.wikimedia.org/wiki/File:Libro_fotografico_digitale_"Wikimania_2016".pdf to Wikimania 2016. The World Gathering of Wikipedia in Esino Lario (Q32979517) using full work available at (P953). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:54, 19 July 2017 (UTC)

Escaped double quotes with "%22" and it worked. Matěj Suchánek (talk) 12:21, 19 July 2017 (UTC)
Yes, but why is a working URL, from our own sister project, rejected? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:28, 19 July 2017 (UTC)
Working ≠ valid. It doesn't work in wikitext either: "Wikimania_2016".pdf. Matěj Suchánek (talk) 13:59, 19 July 2017 (UTC)
A bug in Commons (or MediaWiki), then? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:20, 19 July 2017 (UTC)
Not really either. Valid URLs may contain a pre-defined set of characters, and the quote character (") is not allowed and needs to be encoded as Matěj indicated. However, to make URLs human-readable, one can decode the valid URL, and all modern Browsers and most content management systems (such as Mediawiki) support this functionality to make non-valid URLs working. One could argue now that there should be some URL encoding happening as well if you add a non-valid URL to a claim, but I am not sure which unexpected side effects could arise then… —MisterSynergy (talk) 20:27, 19 July 2017 (UTC)

precision of birth and death dates

I was working a lot lately with importing date of birth (P569) and date of death (P570) from Commons Creator pages. In case of no birth/death dates on wikidata I just imported them with "Imported from Commons" reference. Now I am working on cases where we have date with day-precision on Commons and with year-precision on Wikidata. At the moment I was adding and merging them by hand, looking up the references and setting ranks to "preferred" for day precision dates. That takes a lot of time and since there are hundreds of possibly thousands such items it is not practical to do by hand. There are several options here:

  • I import high precision dates so we would have high and low precision dates with the same rank in large number of items. Unfortunately multiple dates with the same rank is also how we model cases where sources claim two possible dates. This solution also leaves the dates with poor "Imported from .." references.
  • A better solution would be if someone did a bot run and added day-precision dates from RKDartists ID (P650) or other databases (most of the time affected item also have RKDartists ID (P650) which I use as a reference). Those dates should be either added with preferred rank if year-precision dates are present or they should replace year-precision dates.

Any chance someone could do such bot run? If not than unless there are objections I will ask at Wikidata:Bot_requests. --Jarekt (talk) 18:02, 19 July 2017 (UTC)

Please explain how this database identifies whether dates are Julian or Gregorian calendar, and your plan to take this into account. Also please identify whether dates in this database are local or universal time, and if local, your plan to deal with the fact that Wikidata uses Universal Time. Jc3s5h (talk) 18:14, 19 July 2017 (UTC)
Dates in Creator templates can specify if unusual dates are in Julian calendar. This relates mostly to 18 and 19 century Russians for whom dates are usually recorded in Julian calendar and I take that into account. I do not deal with birth and death dates with precision higher than a day so local vs. universal time is not much of an issue. We report dates in literature and those rarely come with time zone info. In my opinion any "adjustments" to the recognized published dates would be very confusing, as project using the data are unlikely to be able to "unadjust" before displaying. --Jarekt (talk) 18:57, 19 July 2017 (UTC)
It may be best to modify the creator template so that it compares dates with the equivalent in Wikidata, and adds any templates where the date(s) do not match to tracking categories. Solutions can then be crowd--sourced. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:18, 19 July 2017 (UTC)
I am already adding dozens of tracking categories, see c:Category:Creator template maintenance. Creator template track when:
  1. data on Commons matches Wikidata (such metadata is then removed from commons and pulled from Wikidata) -> "Creator templates with Wikidata link: redundant ..." categories
  2. Commons has metadata that is missing on Wikidata ("Creator templates with Wikidata link: item missing ..." categories)
  3. Commons and wikidata have the same information but it does not match -> "Creator templates with Wikidata link: mismatching ..." categories
  4. I am also tracking few of the fields to see if they use local or wikidata metadata -> "Creator templates with Wikidata link: local ..."
  5. pages in c:Category:Creator templates with Wikidata link: quick statements‎ have   icon which will launch preloaded QuickStatements tool which will upload metadata missing on Wikidata (at the moment it is mostly day precision dates maching year precision dates on Wikidata)
If some people would like to dive in and help with figuring out the missing or mismatching categories that would be great. let me know if there are any questions. --Jarekt (talk) 03:24, 20 July 2017 (UTC)

Not HTML not exact match search results?

Special:Search/Jakob Esau Rubens returns The reconciliation of Jacob and Esau (Q6102747) and Q27979801, wbsearchentities returns nothing. Is there a way to find and get the items in a machine readable format without scraping the HTML? – preferably language independent, but not necessarily. Thanks in advance, --Marsupium (talk) 13:30, 27 July 2017 (UTC)

Use action=query&list=search instead of wbsearchentities [7] --Pasleim (talk) 14:51, 27 July 2017 (UTC)
Cool! Thanks a lot! --Marsupium (talk) 19:06, 27 July 2017 (UTC)
This section was archived on a request by: --Marsupium (talk) 19:06, 27 July 2017 (UTC)

Migration of constraint definitions to property statements

Hello all,

As we previously announced, we are about to migrate the constraint definitions to statements. On (next Wednesday), we will enable constraint statements on Wikidata so Ivan (owner of KrBot) can start migrating existing constraints from templates into statements on property pages.

To have a smooth transition and allow you to make some tests, we deployed a test wiki where you can create and test constraint statements. You can find there a few examples, or create your own.

If you have any question or comment on this topic, feel free to ping Lucas Werkmeister (WMDE). Thanks, Lea Lacroix (WMDE) (talk) 15:48, 4 July 2017 (UTC)

Lucas! Looks nice, but how do I add exceptions to the constraints? -- Innocent bystander (talk) 17:17, 4 July 2017 (UTC)
@Innocent bystander: once we have constraint statements: add a exception to constraint (P2303) qualifier to the constraint statement. --Lucas Werkmeister (WMDE) (talk) 17:44, 4 July 2017 (UTC)
@Jura1: There is a semi-related ticket (phab:T169374, triaged low priority), but IMHO… a constraint that has over two thousand exceptions should simply not exist. --Lucas Werkmeister (WMDE) (talk) 18:17, 4 July 2017 (UTC)
This property is used more than 2M times, thus the exception rate is lower than 0.1%. Which is not that bad in my opinion. —MisterSynergy (talk) 18:26, 4 July 2017 (UTC)
But they do exist, Lucas Werkmeister (WMDE). Homonyms accross different nomenclature code (Q2673092) are normal. --Succu (talk) 18:32, 4 July 2017 (UTC) PS: See The problem of hemihomonyms and the on-line hemihomonyms database (HHDB)
But there are maybe other ways to solve that?! -- Innocent bystander (talk) 18:35, 4 July 2017 (UTC)
@Innocent bystander: If you want to replace a straightforward solution with a complex one, yes. But I fear splitting taxon name (P225) ínto Code dependent subproperties will not solve this issue and would raise a lot of confusion. --Succu (talk)
A splitting could help to simplify Property:P225#regulärer Ausdruck, but not will omit the NCBI and other cases. --Succu (talk) 20:39, 4 July 2017 (UTC)
A possibility could be to keep this constraint to the bot generated report and not use the gadget for them.
An other possibility could be to load exceptions from dedicated items. Including them on properties might lead to the chessplayer syndrome anyways.
--- Jura 19:50, 4 July 2017 (UTC)
@Succu: The present solution does not looks very "straightforward" to me. We have today bot owners who help us updating the constraint reports. I think it will not be very difficult to find those who can continue help us with this single one. There are many more very complex relation out there. So I doubt this specific property will be the only who needs such help. I currently read a sci-fi novel with time-travel problems and parallel universes. People die before they are born, die twice etc etc. Try to make birth-and death-constraints to that! The number of biographies in StarTrek is huge, we only have a fraction of them here yet. The time-travel problem by some reason reminds me of the "simple located in the administrative territorial entity (P131)-hierarchy" in Q34. -- Innocent bystander (talk) 07:53, 5 July 2017 (UTC)
Innocent bystander, I'm not sure how characters of Star Trek (Q1092) and time travel (Q182154) are helpful here. taxon name (P225) is not about a hierarchy. I'm not sure about you refering to bot owners. What could they do? I'm one of them. --Succu (talk) 20:51, 5 July 2017 (UTC)
Some bots like PLbot and KrBot analyses dumps and based on that updates constraints reports today. I think, if we ask carefully, we can find bots who can handle such things in the future too. If we import all the data from the StarTrek-wiki you will find many contradicting claims. People dies and dies again, have multiple fathers, and have sons they are not parents of. Fiction is fiction, it fits our constraints very poorly. We therefor in the future probably have to tighten up more of our constraints or else our exceptions will grow in numbers. -- Innocent bystander (talk) 06:16, 6 July 2017 (UTC)
  Support definetly a right move: to keep metadata (description of values) close to data (property entity) - the only reason not to do it is performance reasons
Maybe we should use a more suitable language for this task later? d1g (talk) 20:32, 4 July 2017 (UTC)
Update: Due to build problems, there was no software deployment this week. The new planned date for enabling constraint statements from our end is (Thursday next week).
(Ivan has already migrated the constraint statements – thanks! – but they’re not currently used by Special:ConstraintReport and the gadget.) --Lucas Werkmeister (WMDE) (talk) 13:38, 13 July 2017 (UTC)
Does this mean there will be not contraint violations reports for the next seven days? This looks so.... Steak (talk) 09:17, 14 July 2017 (UTC)
No, the database reports by KrBot are independent from the WikibaseQualityConstraints extension. From our end, this just means that constraint checks will still use the old parameters imported two months ago for one more week. I don’t know what’s up with KrBot, sorry. --Lucas Werkmeister (WMDE) (talk) 11:19, 14 July 2017 (UTC)
I saw this problem yesterday at another quantity property and I guess the problem was an incomplete range constraint statement on the property, which has been fixed meanwhile. The other property received an update yesterday after I applied a smiliar fix manually. —MisterSynergy (talk) 11:29, 14 July 2017 (UTC)
@Lucas Werkmeister (WMDE): et al. I work some with single value and unique value constraints for Swedish urban area code (P775) and Swedish minor urban area code (P776). In these cases the exceptions are fine as long as they are associated with qualifiers describing in what time they were valid, for example Skärgårdsstad (Q2720871) and Q31899698 for a unique-problem and Spillersboda (Q2006732) for a single value-problem. -- Innocent bystander (talk) 09:52, 15 July 2017 (UTC)
@Innocent bystander: It sounds like phab:T170403 (and more generally, phab:T167653) would also be useful here – Skärgårdsstad (Q2720871) could have a preferred-rank Swedish urban area code (P775) no value Help statement, and in Spillersboda (Q2006732) the current statement could be changed to preferred rank. --Lucas Werkmeister (WMDE) (talk) 13:19, 15 July 2017 (UTC)
@Lucas Werkmeister (WMDE): All these "old data" has always been visible in the infoboxes at Wikipedia, so I am not sure if that is the way we prefer it in these cases. They have never been treated like population-numbers that are replaced every time new numbers come. I can easily modify the WP-templates so they show both preferred and normal rank values, but I still doubt this is the way we prefer it. Especially is this true for example Stockholm City and Stockholm Municipality who have the same Municipality code. These two never coexisted. The latter replaced the first in the 1970's. So adding "novalue" as municipality code för Stockholms City would only be true before the codes came into use in the 1950's. And adding "novalue" to Stockholm Municipality would never be true, since it got its code directly when it was founded and still has the same code. (Stockholm City changed it codes when it changed county in the 1960's, but that is another story.) Take this into contrast to Skärgårdsstad who is a "populated place" and has been so for decades, both before and after Statistics Sweden classified it as an urban area of its own. -- Innocent bystander (talk) 19:15, 15 July 2017 (UTC)
  Done see #Constraint statements update. --Lucas Werkmeister (WMDE) (talk) 21:38, 21 July 2017 (UTC)

Editwar at Desmopachria barackobamai (Q30434384)

Mr. Mabbett insists in keeping this version which has multiple flaws as I tried to explain in my edit comments. Please have a look. --Succu (talk) 15:07, 7 July 2017 (UTC)

I wondered when Succu would finally deem to discuss their disputed edits. We've discussed previously why named after (P138) belongs as a qualifier on taxon name (P225) and not as a top-level statement (consider a taxon with vernacular and scientific names derived from different subjects); and Succu is repeatedly removing the reliably cited statement that the taxon was described as a species nova (Q27652812). Note also that Succu has falsely accused me - twice, in edit summaries - of committing vandalism. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:16, 7 July 2017 (UTC)
Please give a link to that discussion. --Succu (talk) 16:29, 7 July 2017 (UTC) PS: Mr. Mabbett, you reverted with the following comments: „as before“, „as previously“, „per cited source“, „per source“, „obvious trolling“. There was no indication to an outcome from an earlier discussion. --Succu (talk) 19:00, 7 July 2017 (UTC)
User:Josve05a, could you please comment your changes. Thanks in advance. --Succu (talk) 18:47, 7 July 2017 (UTC)

species nova

Mr. Mabbett is now reverting at species nova (Q27652812) without any comments. --Succu (talk) 19:32, 7 July 2017 (UTC)
Please argue here, Mr. Mabbett, and stop reverting. "sp. nov." did not mean „provisionally, named“ and so on. --Succu (talk) 19:57, 7 July 2017 (UTC)
In an attempt to justify their bad editing on Q30434384, Succu is removing a statement with three reliable sources form Q27652812, and is changing it from "subclass of species" (how can a "new species" be anything but?) to instance of Latin phrase (Q3062294) and abbreviation (Q102786). That's no more correct than making mountain (Q8502), water (Q283) or Douglas Adams (Q42) into instances of words or phrases, because the strings "mountain", "water", and "Douglas Adams" are the names of entities or concepts. This pointed, disruptive behaviour really needs to stop (or be stopped) for good. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:02, 7 July 2017 (UTC)
Reverted as previously by Mr. Mabbett. --Succu (talk) 11:29, 8 July 2017 (UTC)
You are mixing up different thinks, Mr. Mabbett. So lets start with the description of the item. User:Brya corrected your description „term used in biological taxonomy to denote a provisional name for a newly identified species“ to „phrase used when publishing the name of a new species“. Later you changed it to „species which is newly, and provisionally, named“, stating again that a species name marked with sp. nov. is only a provisional name. I restored Bryas description, what was reverted multiple times by you. None of your „reliable sources“ supports that a species name marked with sp. nov. is only a provisional name. --Succu (talk) 14:45, 8 July 2017 (UTC)
I'm not clear on what basis you imagine that your disapproval of the wording of the description in one language justifies your repeated and complete removal from the item "species nova" (meaning and aliased "new species") of the statement "subclass of species", which is supported by three reliable, quoted, sources. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:10, 8 July 2017 (UTC)
We should clarify this issue first. Then I will adress another point. --Succu (talk) 15:17, 8 July 2017 (UTC)
This "spec. nov." is used in the publication where the species is described (in what is called the protologue in the ICNafp) and never thereafter. It is not linked to the species, but to the spot where the name is published. It is not a provisional name (in fact in the ICNafp provisional names are not validly published, that is they are not names). It is in no sense provisional, except as all names are provisional for the first few decades. - Brya (talk) 16:12, 8 July 2017 (UTC)
If Mr. Mabbett agrees to this version we could do the next step. BTW: Please keep in mind instance of (P31)=Latin phrase (Q3062294) was proposed by you. --Succu (talk) 22:00, 8 July 2017 (UTC)
Species novae is not an "instance of" it is part of the publication of the species and as such should not be included on a species level. At best it is a qualifier. Thanks, GerardM (talk) 05:57, 9 July 2017 (UTC)
sp. nov. (or spec. nov.) indicates the role of the cited publication as a first description (of a taxon) (Q1361864) and should be part of a reference, as done in Bartramia (Q2789944) and a lot more taxa. It's not perfect and should be reviewed. --Succu (talk) 18:12, 9 July 2017 (UTC) PS: Maybe type of reference (P3865) should used for the role of the given reference. --Succu (talk) 14:50, 11 July 2017 (UTC)
  Info I've made the substitution. --Succu (talk) 19:38, 11 July 2017 (UTC)

named after

The taxon name is obviously not "named after" as it contains multiple parts and only the least significant part of the name is relevant here. When a name is preserved, it may happen that it becomes part of a subspecies containing three parts to the name as per Wikidata doctrine. In this case the "named after" part remains the least significant part but it does change at the end of the name. In conclusion, can we please have the arguments in place why we want a rationale for the choice for the positions taken. We may even ask an expert for an opinion. Thanks, GerardM (talk) 21:15, 7 July 2017 (UTC)
Thanks, GerardM. Do you have any suggestion how we could the epithet (Q207869) of a taxon better relate to named after (P138) in general? --Succu (talk) 21:27, 7 July 2017 (UTC)
I would include the "named after" as a statement, not as a qualifier. Thanks, GerardM (talk) 08:07, 8 July 2017 (UTC)
@GerardM: Please explain how you would apply that model to Kentish Plover (Q18855), with reference to both Kent (Q23298) and Alexandria (Q87); and to the vernacular names Hvidbrystet Præstekrave and Txirritxo hankabeltz, each of which is currently used as a label. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:49, 8 July 2017 (UTC)
The taxon name would have a statement, the vernacular name has to have a qualifier as it is language specific (as mentioned above) Thanks, GerardM (talk) 13:35, 8 July 2017 (UTC)
@GerardM: "The taxon name would have a statement" I asked you to make reference "to both Kent (Q23298) and Alexandria (Q87); and to the vernacular names Hvidbrystet Præstekrave and Txirritxo hankabeltz". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:01, 8 July 2017 (UTC)
@GerardM: If you can't demonstrate that your model works for this use case, then clearly it is not the correct one. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:25, 9 July 2017 (UTC)
@Pigsonthewing: Andy, please read my responses. I said it before and repeat it just for you; the taxon including what it refers to is on a statement level. When a name in a language refers to whatever, it has to be a qualified because it is language specific. Thanks, GerardM (talk) 15:38, 9 July 2017 (UTC)
I have read each and every one of your responses. In none of them do you address the issue I raised. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:43, 9 July 2017 (UTC)
If "named after Barack Obama" doesn't apply to "Desmopachria barackobamai" as a taxon name (which of course it does), for that reason, then it doesn't apply to "Desmopachria barackobamai" as the name in the label, for the same reason. (Also, "barackobamai" is literally the (most) specific part of the name.) Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:02, 7 July 2017 (UTC)
Do you know User:Daniel Mietchen/Wikidata lists/Named after started at 27 July 2015‎, which is based on having named after (P138) as a statement? --Succu (talk) 20:49, 20 July 2017 (UTC)

Margaret Elizabeth Barr-Bigelow

The English article on Mrs Barr-Bigelow has quite a number of taxons named after her. The intention is to include information like this in Wikidata. With the current edit war being waged, I want quidance how this is to be done and how the author indication is to be included for a taxon. for instance "E.W.A.Boehm, A.N.Mill., Mugambi, Huhndorf & C.L.Schoch" as an example. Thanks, GerardM (talk) 08:04, 8 July 2017 (UTC)

The current practice is to use named after (P138) as a statement. --Succu (talk) 11:41, 8 July 2017 (UTC)
Obviously but how.. I would use it on the item level. Now how about the author indication for the taxon? Thanks, GerardM (talk) 11:50, 8 July 2017 (UTC)
If you want to refer to epithet (Q207869) you could use applies to part (P518). Maybe we should create a new item species epithet. --Succu (talk) 12:10, 8 July 2017 (UTC)
Items can have different names in different languages and also multiple names in the same language. "Named after" is a property of a specific name of an item and therefore I agree that it makes more sense as a qualifier. I remember the previous discussion for named by (P3938) which we created as a qualifier but I don't remember one to transform named after (P138). ChristianKl (talk) 12:06, 8 July 2017 (UTC)
Last week named after (P138) as a qualifier to taxon name (P225) had a single usage. --Succu (talk) 12:17, 8 July 2017 (UTC)
Because some editors (well, at least one) disruptively remove it. Wikidata does not work by Fait accompli (as, for example, explained on Wikipedia). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:27, 8 July 2017 (UTC)
Please note my edit comment: „if you want to set this only usage as qualifier as standard over a lot of usages as property then please discuss this first“ (dated 14 April 2017‎). --Succu (talk) 13:00, 8 July 2017 (UTC)
QED Note that Succu again removes a reliable citation, with a quote. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:33, 8 July 2017 (UTC)
Unfortunately your „reliable citation, with a quote“ gives no statement about the origin of the genus name. --Succu (talk) 15:24, 8 July 2017 (UTC)
Poppycock. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:50, 8 July 2017 (UTC)
„Upland Sandpiper, Bartramia longicauda, was known... as Bartram’s Sandpiper. William Bartram (1739-1823) was not only long-lived, but he was considered the grandfather of American ornithology“ only mentions Upland Sandpiper (Q530194) and William Bartram (Q18243), but not that the genus Bartramia (Q2789944) was named after him. --Succu (talk) 16:23, 8 July 2017 (UTC)
The third word in the quote is "Bartramia". QED. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:50, 8 July 2017 (UTC)
Q.E.D. (Q188722)? No, you demonstrated the presence of the word „Bartramia“ and not that the genus was named after William Bartram. And please do not remove the author citation with the given source at Bartramia (Q2789944) again and again. --Succu (talk) 17:30, 8 July 2017 (UTC)
If you keep removing good content, then adding something else to "poison the well" against being reverted, you can't complain when your addition is lost during a revert. I've done enough cleaning up after you from that kind of edit. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:30, 8 July 2017 (UTC)
The current best practice is to use named after (P138) as a qualifier on the taxon and/or vernacular name(s), as appropriate. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:27, 8 July 2017 (UTC)
Why is it the best practice according to you? After all on the taxon it is ambiguous. Vernacular names are different; they are language specific so there it has to be a qualifier. Thanks, GerardM (talk) 12:45, 8 July 2017 (UTC)
I did not say "best practice according to me"; I said ""best practice". Please address my Kentish Plover (Q18855) example, above. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:50, 8 July 2017 (UTC)
I indicated that you think it is a best practice. I disagree on that. I already addressed how I would do vernacular names. When you really want to know how I really address these issues, check out OmegaWiki. I consider it a flaw in the system. Thanks, GerardM (talk) 13:26, 8 July 2017 (UTC)
@GerardM: You have yet to respond to my Kentish Plover (Q18855) example; please do so. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:28, 8 July 2017 (UTC)
Is there some support that Carl Linnaeus (Q1043) named the species after Pope Alexander of Alexandria (Q44794)? --Succu (talk) 21:57, 13 July 2017 (UTC)
The placing of "named after" is a matter of formatting, and as such debatable. Personally, I would prefer to place it as a statement, but I would also prefer to place "taxon author" and the "date of taxon name publication" as a statement. - Brya (talk) 16:18, 8 July 2017 (UTC)
The "date of taxon name publication" is ambiguous. There are names that have been published on several dates in distinct publications. Thanks, GerardM (talk) 06:09, 9 July 2017 (UTC)
Then change the label, or description, of the property to "date of taxon name's first publication", which is what is meant, and what is significant in taxonomy. Or give two dates and two sources. Either way, that's irrelevant to the issues raised here. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:25, 9 July 2017 (UTC)

Items included in this conflict

Besides Desmopachria barackobamai (Q30434384) and species nova (Q27652812) there are now: Bartramia (Q2789944), Pinkfloydia (Q20675082) and Synalpheus pinkfloydi (Q29367343). --Succu (talk) 17:46, 8 July 2017 (UTC)

Yes you're stalking my edits, too. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:12, 8 July 2017 (UTC)
Still stalking my edits. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:25, 9 July 2017 (UTC)
I think they should be both qualifiers and main properties. 1 2
This is useful duplication, we probably cannot avoid this using only one approach: property is too inexact, white qualifier is less used for mentioned properties (named after - named after (P138))
Secondly, this particular duplication wouldn't change in future (unnamed after) d1g (talk) 08:57, 9 July 2017 (UTC)
Your duplication proposal breaks in cases like Kentish Plover (Q18855), which has different labels in different languages. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:31, 9 July 2017 (UTC)
Can you provide full example what should we do with Kentish Plover (Q18855)? d1g (talk) 11:47, 9 July 2017 (UTC)

I think we can add named after (P138) as allowed qualifier in taxon common name (P1843) and similar properties until we have a better option. d1g (talk) 12:11, 9 July 2017 (UTC)

Generally about P138

named after (P138) generally has one problem, it is language-specific. The month of March (Q110) is according to the item named after A roman god with the same name. But in "country: Turkmenistan" it is named after the day of new year in the Persian calendar. (I doubt this, since the Swedish name is the same in Turkmenistan as it is in any other part of the world.) In two named languages it is named after a genus of plants. None of these claims have any source, but that is maybe not the largest problem here. It is rather that there is no standard of how this should be described. I would prefer to see "language:Whatever" as a standard qualifier here, but that works poorly for scientific names, since it isn't a language. Maybe "applies to part" would work? -- Innocent bystander (talk) 11:11, 9 July 2017 (UTC)

That's not a problem with the property, but with its usage. It is why it should be used as a qualifier on specific names (whether taxon names, or "name in native language", or whatever) not on the item generally; hence my giving Kentish Plover (Q18855) as an example - still unaddressed - above. Using it as a qualifier is both more precise, and addresses the issue you raise. Note also that no counter argument has been advanced. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:27, 9 July 2017 (UTC)
@Pigsonthewing: The months do not have "name(s) in native language"! Neither has elements like mercury (Q925) "native names". We cannot put qualifiers to labels today, I doubt that we ever will! -- Innocent bystander (talk) 11:34, 9 July 2017 (UTC)
Hence "or whatever". And that we cannot add qualifiers to labels is my point. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:48, 9 July 2017 (UTC)
On reflection, for the names of things such as mercury (Q925), which are not proper nouns, it may be better to apply the etymology to the relevant Wiktionary item; another option is to use the ISO code mul for multi-lingual content, like this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:58, 9 July 2017 (UTC)
If you take Patient Protection and Affordable Care Act (Q1414593) you could say that it's named in English after Obama but that wouldn't be very useful. It would make more sense to say that the item has as name (P2561) Obamacare and qualify that with named after (P138). I think we should have a discussion on the talk page of named after (P138) to transform it into a qualifier and set the appropriate constraints. ChristianKl (talk) 16:41, 9 July 2017 (UTC)
named after (P138) is not directly relevant here, but:
country specific or ethnic or religious, but language-specific is only when everything else is unknown. d1g (talk) 11:39, 9 July 2017 (UTC)

Summary?

It would be good if someone could give a short summary about this topic, including some conclusions how to proceed. Thank you. --Succu (talk) 18:14, 18 July 2017 (UTC)

With pleasure: named after (P138) belongs as a qualifier on taxon name (P225) and/or taxon common name (P1843) and not as a top-level statement. This is the only viable model discussed, for a taxon with vernacular and scientific names derived from different subjects; for example Kentish Plover (Q18855), which not only has different labels referring to Kent (Q23298) and to Alexandria (Q87); but also to the vernacular names Hvidbrystet Præstekrave and Txirritxo hankabeltz. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:28, 19 July 2017 (UTC)
This is your POV and does not reflect all arguments given. --Succu (talk) 14:06, 19 July 2017 (UTC)
I repeat: "the only viable model discussed". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:23, 19 July 2017 (UTC)
It is not. Thanks, GerardM (talk) 04:32, 20 July 2017 (UTC)
If you believe so, then please point out where another viable model was discussed, and please explain how you would apply that model to Kentish Plover (Q18855), with reference to both Kent (Q23298) and Alexandria (Q87); and to the vernacular names Hvidbrystet Præstekrave and Txirritxo hankabeltz, each of which is currently used as a label. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:48, 20 July 2017 (UTC)
I know so and made my point before. A fact you failed to acknowledge. The taxon is relevant in any language and consequently what it is named for is universal. The labels for a specific language are on a different level, they are relevant for one language only and consequently they have to be qualified. Thanks, GerardM (talk) 13:40, 20 July 2017 (UTC)

"Unreferenced" Items

As a general question, for those of you concerned with unreferenced items, what exactly constitutes a "reference"? A sitelink? A reference/citation on any statement? An identifer? A described by source (P1343) statement? A described at URL (P973) statement? - PKM (talk) 21:14, 15 July 2017 (UTC)

Wikidata:Glossary#Reference. Matěj Suchánek (talk) 07:35, 16 July 2017 (UTC)
I’d like to expand Matěj’s answer a bit. Written down as bullet points, since several loosely related points are addressed.
  • Almost all items with sitelinks and backlinks (according to WD:N) are pretty safe anyway.
  • Difficult are cases that are notable due to point 2 of WD:N. The phrase “… can be described …” seems a little outdated meanwhile as Wikidata has matured a lot, and there is also uncertainty about the matter what a “serious” source is.
    • Although I don’t know what the exact reason for these uncertainties is, I suspect that this is intentional flexibility that allows us to deal with all kinds of special cases in an apropriate manner. The uncertainty typically isn’t used to aggressively delete items, it rather enables us to allow very special situations as well.
    • Regarding seriousness of sources: according to my impression this means coverage by an external source that is maintained by some kind of authority. Thus, many external databases (external identifiers) are serious sources, but not all are. Facebook and the like are not serious. User-generated content (without any supervising authority) in general is tenuous, yet not generally non-serious.
    • As a rule of thumb it would thus help a lot if items which are notable according to #2 of WD:N are equipped with an external identifier, which is not user-generated content. There are certainly exceptions to this “rule”. Anything that helps to unambiguously identify this entity is extremely useful.
  • Besides that, we also work for individual references for each and every claim, but this isn’t something we will achieve that soon.
  • I’d also like to address why this is important. We do have notability criteria, since we are not able to cover each and every entity in this world—neither technically, nor socially with this community. In order not to drown in promotional content and fake content (there is indeed a lot of both) as well as other non-notable items, we do need to check precarious items for their notability. This is unfortunately done by just a few admins, and the backlog of problematic cases reaches back until at least 2014 (!). A typical such item does not provide much more content than “John Doe, male American person, born 1940”. How would one check whether this is a notable “John Doe”? It is in fact not really possible for admins to do that. The typical workflow is thus to identify a list of problematic items, check for references (in each claim, not “imported from: some Wikipedia”), check for identifiers and backlinks, check history for (deleted) sitelinks, and if it is not clear then what or whom this item is about, it is deleted. Since there are myriads of cases, admins could do this for hours without significantly reducing the backlog, and this is really the most boring Wikidata activity one can imagine. As long as only a few admins/editors work in this field, the number of cases does not permit to ask involved editors individually, and neither does it allow to perform Google searches.
  • If such an item is deleted, kindly ask the admin or WD:AN what it was about, and whether it could be restored. I would be happy if this could be done without much drama.
MisterSynergy (talk) 19:45, 16 July 2017 (UTC)
Thanks, that answers my question. - PKM (talk) 19:38, 21 July 2017 (UTC)

Too many redirects

Is the quality of items going down? Because on this graph I have noticed a ~300000 increase in redirects in one day! This trend of mass redirecting seems to have started on July 8, probably from GZWDer (flood) (talkcontribslogs). PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:57, 19 July 2017 (UTC)

I can see +30,000, not +300,000. Moreover, I don't think number of redirects created corelates with item quality. Matěj Suchánek (talk) 14:18, 19 July 2017 (UTC)
I think redirects can be a good marker of improved quality, in that redundant items in different languages or scripts are being identified and merged. We still have a lot of those to find. - PKM (talk) 19:41, 21 July 2017 (UTC)

Lag times

High dispatch lag

There has been a higher dispatch lag since the end of last month. The English Wikipedia is now three days behind Wikidata, according to Special:DispatchStats. I would like to ask everyone to keep mass edits to items with sitelinks to a minimum low until the dispatch lag is on a reasonable level.

I would also like to ask Magnus Manske to build proper editing limits into his tools. It should not be possible to run more like one instance of QuickStatements for example. Some tools also don't seem to have limits at all. Sjoerd de Bruin (talk) 12:27, 20 July 2017 (UTC)

Server lag and bot edit rates

In the last weeks Wikidata has experienced some cases of "server lag" (dispatch lag, job queue). Currently, according to admins the dispatch lag and job queue is pretty high. I have been requested to stop my bot tasks for a week. I have about 10 tasks approved, I was just running 1 of them (at 130 edits/min, needing several days to complete) and I am just 1 of dozens of bot operators here in Wikidata, and that was enough to cause trouble.

I think that the current lag was provoked by the creation of 3 million new items for categories in the last weeks. Anyway, I don't want to point to anybody. I think we should discuss about this situation, as now bots "can't edit" until the dispatch log goes down. I have some proposals to mitigate this in the future:

  • Any bot task must be approved (I have seen bots running several tasks, and their bot templates just show 1 approved task) and it has to state an upper edit rate limit
  • Develop a pywikibot throttle function that counts not only maxlag but dispatchlag figures
  • Ask WMF/WM-DE to allocate more resources for Wikidata (job servers?). Some bot tasks are so big that running at 50 edits/min would need months to complete (and after that time, they would need to be restarted again). Anyway, running just 6 different bots at that rate, would sum up 300 edit/min, which is probably a rate high enough to cause trouble to servers.

Wikidata needs bots to keep items updated, tons of bots. I wonder what would happen when Commons structured data and Wiktionary lexemes (dozens of million data snipets) are added to Wikidata, will this turn unsustainable?

Comments? Thank you. Emijrp (talk) 12:33, 20 July 2017 (UTC)

Me changing my signature template on 1000+ pages didn't help, but that should take a minute. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:12, 20 July 2017 (UTC)

You should not be using a template for your sig. Please disable it, and then it can be substituted and deleted. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:37, 20 July 2017 (UTC)
And at least make the font size smaller first. Sjoerd de Bruin (talk) 15:47, 20 July 2017 (UTC)
@PokestarFan: Did you see this? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:28, 21 July 2017 (UTC)

lagtime

Hi, sometime I wanted to run multi jobs more than 4 jobs, but then my bot will do more then 350 edit in minute via API.?

def New(data, summary):
    r4 = session.post(api_url, data={
        "action": "wbeditentity",
        "format": "json",
        "maxlag": "3",
        "new": "item",
        "summary": summary,
        "bot": 1,
        'utf8': 1,
        "data": str(data) , 
        'token': r3.json()['query']['tokens']['csrftoken'],
    })
So is there a way to make the bot do 60 edit per minute at most in multi jobs? --Mr. Ibrahem (talk) 13:11, 20 July 2017 (UTC)
If you are doing multiple jobs from the same application, do not execute requests directly but make a separate variable with callbacks and use threading.Timer like:
from threading import Timer

queue = []
stop = False
timer = None

def next_callback():
    if queue:
        callback = queue.pop(0)
        callback()
        if stop:
            return
    timer = Timer(1, next_callback).start()

timer = Timer(1, next_callback).start()

# ...

def New(data, summary):
    queue.append(lambda: session.post(api_url, data={
        "action": "wbeditentity",
        "format": "json",
        "maxlag": "3",
        "new": "item",
        "summary": summary,
        "bot": 1,
        'utf8': 1,
        "data": str(data) , 
        'token': r3.json()['query']['tokens']['csrftoken'],
    }))
If you are running bot from multiple applications, just wait (time.sleep) before each request 1 * (number of jobs). Matěj Suchánek (talk) 14:30, 20 July 2017 (UTC)

The dispatch, job, and lag status

The problem is so great that I think a discussion on Project Chat isn't going to cut it. I think we need to spread the message of how to help the problem to the entire site, probably with those banner things that can be placed on the top of a page. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:18, 20 July 2017 (UTC)

But just pointing bots as is the main source of problem is not enough: what's about QuickStatement and other tools allowing mass edits or fast edits ? And what's about the efficiency of all those edits when we hear that more than 30'000 items where recently merged ? To much freedom is going to kill WD: lonely contributors shouldn't be able to decide mass edits without support of some groups of contributors like wikiprojects and with an appropriate curation of data BEFORE the import in WD. Snipre (talk) 16:11, 20 July 2017 (UTC)
Exactly. I like the interface of QuickStatements 2, maybe some option that another user (with some protection to avoid sockpuppets) needs to approve a bath task first? Sjoerd de Bruin (talk) 16:14, 20 July 2017 (UTC)
We have to put more constraints in items creation: at least one label with a description and at least a instance of/subclass of statement. And for tools like QuickStatements, a number of edits per batch as to be defined with a consistent lag time. To overcome these limitations, an evaluation like the one used for bot ask should be required with link to the discussion defining the need of the mass edition. ~Just some ideas. Snipre (talk) 16:20, 20 July 2017 (UTC)
QuickStatements is like running a bot. I think that any QS task should be approved, like any bot must be approved first (Though we know that here in Wikidata many bots managers post the first task to get the flag, and then they don't look for approval anymore). Sometimes categories aren't linked between Wikipedias, so I wouldn't blame the bot operator if that is what happened here. Importing unconnected categories and merging them here, helps to improve the connectivity, reducing Wikipedia maintenance backlog, specially for small languages. Emijrp (talk) 16:27, 20 July 2017 (UTC)
Is it fair to say that recent changes is not production ready? Thanks, GerardM (talk) 17:57, 20 July 2017 (UTC)
I use QuickStatements all the time to add multiple statements to single-item new creations, often for books to be used as references, or for missing "steps" in a class hierarchy. Requiring approval for "any QS task" is not going to solve our problem here and is going to make it harder for individual contributors to do quality work. - PKM (talk) 19:56, 20 July 2017 (UTC)
It now looks like the GUI has trouble from time to time, making manual editing fail. The statement-parts of the page looks read only while the label/description looks editable, but when you "edit" the interface fails, only providing editing in one single language. -- Innocent bystander (talk) 08:35, 21 July 2017 (UTC)
Currently QuickStatements2 have security flaw that even blocked users can use this tool to edit.--GZWDer (talk) 10:57, 21 July 2017 (UTC)

Strategy discussion, cycle 3. Challenge 4

Hi! The movement strategy discussion is still underway, and there are four challenges that you may discuss:

  1. How do our communities and content stay relevant in a changing world?
  2. How could we capture the sum of all knowledge when much of it cannot be verified in traditional ways?
  3. As Wikimedia looks toward 2030, how can we counteract the increasing levels of misinformation?
  4. and the newest one: How does Wikimedia continue to be as useful as possible to the world as the creation, presentation, and distribution of knowledge change?

The last, fifth challenge will be released on July, 25.

If you want to know what other communities think about the challenges, there's the latest weekly summary (July 10 to 16), and there's the previous one (July 1 to 9).

SGrabarczuk (WMF) (talk) 13:36, 20 July 2017 (UTC)

Query Service not Working

So I decided to fnd rivers in the USA with a query, so I used the valid query:

SELECT ?item ?itemLabel ?itemDescription WHERE {
  ?item wdt:P31 wd:Q4022.
  ?item wdt:P17 wd:Q30.
}
#This is to make the template work

Try it!

I got some results (23k to be exact), but none of them is showing up with a label or description, even though the items have labels and descriptions. What is wrong with the Query Service? PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:45, 20 July 2017 (UTC)

You should add the line SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }, because otherwise WDQS doesn't know in which language it should show the descriptions and labels. Q.Zanden questions? 13:59, 20 July 2017 (UTC)
Oh, I thought that was always optional. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 15:12, 20 July 2017 (UTC)

Scores in ORES will be more accurate

Hello,

If you use ORES to watch vandalism, you may be interested by the next evolutions of the tool.

Currently, when patrollers hit the ORES API to check if an edit is vandalism or not, they get a score between 0 and 1 (example). From July 25th, these scores will change for edits, will get more signal out of edit summaries, and will be more accurate. We will for example use existing lists of bad words to catch this kind of vandalism more easily. You can see the details on the task.

If the patrollers set the threshold in their tools to highlight an edit if it was, for example, more than 90%, they need to revisit the number once the new changes got deployed. On this page you can find full statistical parts and find out what is the new threshold when it gets deployed.

If you need any help to set up your tool, feel free to contact Ladsgroup. Thanks, Lea Lacroix (WMDE) (talk) 15:15, 20 July 2017 (UTC)

How to access Wikidata with Lua

I am trying to figure out how to access list of all Wikisource or wikiquote sitelinks from Lua. Current mw.wikibase.entity:getSitelink function allow me to access then if I know which language I want, but so far I could not figure out how to get a list of all choices (other than test every supported language). Any Ideas? I am also trying to figure out how to access item aliases from lua. --Jarekt (talk) 16:57, 20 July 2017 (UTC)

@Jarekt: Here follows a list of all Wikisource sites and its corresponding pages in Project:Village pump (Q16503).
iswikisource:Wikiheimild:Potturinn, bnwikisource:উইকিসংকলন:লিপিশালা, ptwikisource:Wikisource:Esplanada, itwikisource:Wikisource:Bar, etwikisource:Vikitekstid:Üldine arutelu, cswikisource:Wikizdroje:U pramene, bswikisource:Wikizvor:Čaršija, aswikisource:ৱিকিউৎস:কেন্দ্ৰীয় আলোচনা, srwikisource:Викизворник:Писарница, trwikisource:Vikikaynak:Köy çeşmesi, mrwikisource:विकिस्रोत:चावडी, orwikisource:ଉଇକିପାଠାଗାର:ଆଲୋଚନା ସଭା, frwikisource:Wikisource:Scriptorium, hrwikisource:Wikizvor:Pisarnica, arwikisource:ويكي مصدر:الميدان, brwikisource:Wikimammenn:An davarn, guwikisource:વિકિસ્રોત:સમાજ મુખપૃષ્ઠ, huwikisource:Wikiforrás:Kocsmafal, svwikisource:Wikisource:Mötesplatsen, euwikisource:Wikiteka:Txokoa, zh_min_nanwikisource:Wiki Tô·-su-kóan:Chheh-pâng, hywikisource:Վիքիդարան:Խորհրդարան, cywikisource:Wicidestun:Y Sgriptoriwm, azwikisource:VikiMənbə:Kənd meydanı, idwikisource:Wikisource:Warung kopi, sawikisource:विकिस्रोतः:समुदायद्वारम्, pawikisource:ਵਿਕੀਸਰੋਤ:ਸੱਥ, jawikisource:Wikisource:井戸端, lawikisource:Vicifons:Scriptorium, fawikisource:ویکی‌نبشته:دفترخانه, dawikisource:Wikisource:Skriptoriet, cawikisource:Viquitexts:La taverna, vecwikisource:Wikisource:Ciacole, bgwikisource:Уикиизточник:Разговори, zhwikisource:Wikisource:写字间, tewikisource:వికీసోర్స్:రచ్చబండ, bewikisource:Вікікрыніцы:Супольнасць, dewikisource:Wikisource:Skriptorium, hewikisource:ויקיטקסט:מזנון, ukwikisource:Вікіджерела:Скрипторій, skwikisource:Wikizdroje:Portál komunity, elwikisource:Βικιθήκη:Γραμματεία, mkwikisource:Викиизвор:Селска чешма, pmswikisource:Wikisource:Piòla, enwikisource:Wikisource:Scriptorium, slwikisource:Wikivir:Pod lipo, glwikisource:Wikisource:A Taberna, mlwikisource:വിക്കിഗ്രന്ഥശാല:വിക്കി പഞ്ചായത്ത് (സാങ്കേതികം), nlwikisource:Wikisource:De kroeg, plwikisource:Wikiźródła:Skryptorium, fiwikisource:Wikiaineisto:Kahvihuone, liwikisource:Wikibrónne:Gebroekersportaol, ruwikisource:Викитека:Форум, fowikisource:Wikiheimild:Undirhúsið, eowikisource:Vikifontaro:Diskutejo, kowikisource:위키문헌:사랑방, eswikisource:Wikisource:Café, nowikisource:Wikikilden:Kontoret, rowikisource:Wikisource:Scriptoriu, knwikisource:ವಿಕಿಸೋರ್ಸ್:ಅರಳಿ ಕಟ್ಟೆ, tawikisource:விக்கிமூலம்:ஆலமரத்தடி, viwikisource:Wikisource:Thảo luận, yiwikisource:װיקיביבליאָטעק:געמײנדע
You see the code to access this at Module:Sandbox/Innocent bystander
It is also possible to access any alias in any language. -- Innocent bystander (talk) 18:11, 20 July 2017 (UTC)
Here follows all English aliases in Richard Feynman (Q39246)
Richard Phillips Feynman, Richard P. Feynman
The code can be seen in the same page. -- Innocent bystander (talk) 18:28, 20 July 2017 (UTC)
Innocent bystander, thank you. So no fancy lua calls just dive into the entity data structure. I will have to remember this approach when running into other issues like that. --Jarekt (talk) 19:10, 20 July 2017 (UTC)
It is possible that there are shortcuts to access this information, but I think this method is straightforward, since all data is there, aliases, sitelinks, labels, descriptions and statements. You do not have to download twice, once for label and once for the sitelink. Sometimes the fallback to English is not the best of options, like when looking up transcriptions from a language with Cyrillic script, then labels in German is often much more useful. Other times the fallback to Bokmål or Danish is much better, for example names of occupations. The ordinary way to access the label is then not the best option. -- Innocent bystander (talk) 19:25, 20 July 2017 (UTC)

Permission for 12000 categories creation from wiktonary

Permission granted? PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 23:04, 20 July 2017 (UTC)

@PokestarFan: thanks a lot if you have a bot able to do this. Do you plan to import categories from all the Wiktionary projects or only from few of them (English one and maybe few others)? Pamputt (talk) 23:24, 20 July 2017 (UTC)
@Pamputt: I need to request permission first. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 02:50, 21 July 2017 (UTC)
No. You don't seem to understand why you have to ask for permission, otherwise you would not make such vague requests. You should also reread Wikidata:Project_chat#Lag_times. We are trying to avoid mass edits right now, particularly things like mass creating items for categories. - Nikki (talk) 03:08, 21 July 2017 (UTC)
Which categories? Almost all categoies with interwikiki are already imported except some of those which are on fr.wikt and end.wikt only and except categories with some interwiki conflicts from top 35. JAn Dudík (talk) 05:10, 21 July 2017 (UTC)
To receive the permission you have to explain what you will do and how you do it. And the best is to discuss first with people from Wikisource Wiktionary to see what they want to see in WD. Just creating items is no sense without a purpose. Snipre (talk) 11:29, 21 July 2017 (UTC)
@Snipre: I guess you wanted to write Wiktionary (and not Wikisource). Pamputt (talk) 13:12, 21 July 2017 (UTC)

search items by altlabel

Hi, I try to search item by alternative label and it does not work for some reason.

when searching according to labels it does work:

SELECT DISTINCT ?s ?sLabel ?sAltLabel WHERE {
  ?s wdt:P19 wd:Q844930.
  ?s rdfs:label ?sLabel .
  FILTER(?sLabel="Plato"@en).
 
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
} limit 100

Try it!

But when searching according to altLabel I get no results:

SELECT DISTINCT ?s ?sLabel ?sAltLabel WHERE {
  ?s wdt:P19 wd:Q844930.
  ?s skos:altLabel ?sAltLabel. 
  FILTER(CONTAINS(?sAltLabel, "Bolatu"@en)). 
 
 
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
} limit 100

Try it!

why is that and how to fix it? thanks — Preceding unsigned comment added by 31.168.180.89 (talk) 08:33, 21 July 2017 (UTC)

If you wrap the part with skos in "OPTIONAL" you will get a hit.
SELECT DISTINCT ?s ?sLabel ?sAltLabel WHERE {
  ?s wdt:P19 wd:Q844930.
 OPTIONAL { ?s skos:altLabel ?sAltLabel. }
 FILTER(CONTAINS(?sAltLabel, "Bolatu"@en)) 
 
 
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
} limit 100

Try it!

I can't explain why. --Larske (talk) 09:31, 21 July 2017 (UTC)
Wikidata:Request a query is a better place to ask.--Jklamo (talk) 10:21, 21 July 2017 (UTC)


SELECT DISTINCT ?s ?sLabel ?sAltLab WHERE {
  ?s wdt:P19 wd:Q844930.
 OPTIONAL { ?s skos:altLabel ?sAltLab. }
 FILTER(CONTAINS(?sAltLab, "Bolatu"@en)) 
 
 
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
} limit 100

Try it!

I think ?sAltLabel has something with "SERVICE wikibase:label" if you change it to any thing, you will get results. --Mr. Ibrahem (talk) 13:21, 21 July 2017 (UTC)
  • Interesting find with "?sAltLabel"!
    --- Jura 23:45, 21 July 2017 (UTC)

Make the merge gadget a default option?

This would help to cut down on the amount of requests at WD:RFD. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 01:54, 22 July 2017 (UTC)

Previous discussion at Wikidata:Project_chat/Archive/2017/06#Integrate_the_merge_function_into_the_default_UI and see also phab:T140124--GZWDer (talk) 05:50, 22 July 2017 (UTC)

Job queue has dramatically lowered

I think that bot jobs can be restarted, according to this graph, there are only 283k jobs, and the number is getting lower and lower. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 15:20, 22 July 2017 (UTC)

English Wikipedia is still 2 days and 17 hours behind, according to Special:DispatchStats. Sjoerd de Bruin (talk) 15:47, 22 July 2017 (UTC)
I don’t think job queue is the same thing as dispatch queue, and there are still about 700k dispatches pending for the stalest wiki (enwiki, apparently) according to the wikidata-dispatch board. Historically, the normal size of the stalest dispatch queue seems to be somewhere around 100-200 items, so 700,000 is still way too much. --Galaktos (talk) 16:59, 22 July 2017 (UTC)
I did not mean sitelink jobs, I meant lael/description/alias/property jobs where sitelinks are not involved. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 18:21, 22 July 2017 (UTC)

Lavoisier Medal

The articles on several Wikipedias is not about one but about three distinct Lavoisier medals. Obviously all three need their own item. But how are we going to link them with Interwiki links. They are not disambiguation pages.. Thanks, GerardM (talk) 15:55, 22 July 2017 (UTC)

instance of (P31) Q17519152 with the several medals as has part (P527). Sjoerd de Bruin (talk) 16:05, 22 July 2017 (UTC)
Thanks, GerardM (talk) 16:24, 22 July 2017 (UTC)

Module:Cycling race

Hello, can someone help me to install Module:Cycling race on Wikinews Dutch? It's a good system and it would add value to the articles about cycling. I tried to contact the maker but it is temporarily inactive. It would be great if anyone would like to do this! Thanks in advance! --Livenws (talk) 20:49, 22 July 2017 (UTC)

@LIVE NIEUWS: Have you tried copying and pasting from w:nl:Module:Cycling_race? —Justin (koavf)TCM 22:19, 22 July 2017 (UTC)
@Koavf: I'm not so technical. Do you only have to copy that template? Nothing more extra? --LIVE NIEUWS (talk) 22:22, 22 July 2017 (UTC)
I have copied that template here. But it doesn't work at the moment. I have tried it with {{Cycling race/stageinfobox|Q27437605}} on an article, but without success --Livenws (talk) 22:32, 22 July 2017 (UTC)
@LIVE NIEUWS: It will probably work but of course, it's better to actually import it. If you're an admin, you should be able to import. If the module doesn't work, that is probably because it depends on other modules or templates, so if you import those, then it should work. If you can't import, copying, pasting, and providing attribution is all you need. —Justin (koavf)TCM 22:33, 22 July 2017 (UTC)
Thank you, I will continue working on it tomorrow and report it here how it goes. --LIVE NIEUWS (talk) 22:36, 22 July 2017 (UTC)

Userboxes in the user namespace

Hi! Some userbox templates on frwiki have interwiki links to pages in the user namespace on other wikis (especially on the English Wikipedia). Examples:

Can I add those userboxes to the corresponding Wikidata items? Orlodrim (talk) 17:29, 23 July 2017 (UTC)

I do not think that is a big problem, as soon as you can find a helpful sysop. Because only sysops can add sitelinks in user namespace to items. -- Innocent bystander (talk) 17:39, 23 July 2017 (UTC)

Notabilty rules limits in footnotes

Was it proper for User:ValterVB here to revert my emendation of a footnote? The actual discussion about the rule change occurs at the link I provided in addition to the previous link. The previous discussion cited is about the creation of the Commons Category link system, not about the limits of its use. --Richard Arthur Norton (1958- ) (talk) 23:57, 17 July 2017 (UTC)

Yes, it was, because the reason he reverted you was need consensus. I read quickly through the page you added as reference, and to me it did not look like there was any consensus. The most of our guidelines/policies and especially new additions or removals to existing guidelines/policies should always be based on censensus (not majority). Q.Zanden questions? 09:50, 18 July 2017 (UTC)
I do not see any consensus at the first link either. I do not see any place where the wording gained consensus. The first link discusses the creation of the field to hold "commons category". --Richard Arthur Norton (1958- ) (talk) 14:05, 20 July 2017 (UTC)
As you can see, the first link has been "completed" by marking it with a blue square around the whole text, a sign that the outcome of the RfC was succesful and as written under Addendum 2 there was a succesful outcome of option V as well. Your link still does not end with any conclusion and as it is in the archive-section there will not be any ending conclusion in favor of your extra link. Beside that, there is already one footnote for the text, so why would you wnat to add a second one? Q.Zanden questions? 14:22, 20 July 2017 (UTC)
I still do not see what you are referring to, can you show me where it discusses: "an item with only a sitelink to a category page in Wikimedia Commons is not allowed on main article items. However, it is allowed to link Wikimedia Commons categories with categories in other Wikimedia sites in items." I do not see consensus for deleting items if they only have a link to a category in WikiCommons. --Richard Arthur Norton (1958- ) (talk) 19:01, 23 July 2017 (UTC)

Request for a gadget to collapse properties

I would like a gadget (and others would too probably) that lets me collapse properties & their values like wikitables. It would make navigating certain items easier. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 23:28, 22 July 2017 (UTC)

@PokestarFan: This isn't exactly like wikitables, but it might work for you anyway: User:Yair rand/PropCollapser.js. --Yair rand (talk) 22:35, 23 July 2017 (UTC)

Native names

We have native label (P1705) and name in native language (P1559). This seems a bit redundant to me - do we need both? (I'm asking here rather than nominating one for deletion as I'm not sure of the background/thinking here). Thanks. Mike Peel (talk) 23:38, 22 July 2017 (UTC)

I think that P1559 is for "full name in native language", while P1705 is not limited for names.
Yes, we need both because personal names are statistically different from names of other things.
It is common to use only one part of names, not both, so both properties are somewhat needed d1g (talk) 01:56, 23 July 2017 (UTC)
So, P1559 is the same as P1705 + P31? I'm still not seeing the structural difference here. Thanks. Mike Peel (talk) 21:52, 23 July 2017 (UTC)

wbEntity config variable to be deprecated

Hello,

This is an important message for people who develop gadgets or user scripts for Wikidata’s frontend.

We plan to deprecate the wbEntity javascript config variable in the next months. The variable is currently used on every page load, even if when it is not useful (for example on mobile). With that change, we will increase the load speed of the entity pages. After that the expression mw.config.get( 'wbEntity' ); will not work any more. All code that uses this config variable should be migrated towards using the Mediawiki hook wikibase.entityPage.entityLoaded.

When a user opens an entity page (item or property), as soon as the JSON representation of the entity stored on the current entity page is loaded, the Mediawiki hook wikibase.entityPage.entityLoaded is triggered. Listener callbacks should expect the entity as a native JavaScript object (the parsed JSON serialization) passed as the first (and only) argument.

Note: The entity object is completely frozen (read-only) to avoid the case when one of the clients accidentally changes it and breaks other clients.

Here are a few examples on how you can use it:

 // Basic usage
mw.hook( 'wikibase.entityPage.entityLoaded' ).add( function ( entity ) {
  'use strict';
  // Your code goes here
  console.log( entity );
} );

// Convert to jQuery promise
var entityPromise = $.Deferred( function ( deferred ) {
  mw.hook( 'wikibase.entityPage.entityLoaded' ).add( function ( entity ) {
     deferred.resolve( entity );
  } );
} ).promise();

// Convert to native promise
var entityPromise = new Promise( function ( resolve ) {
  mw.hook( 'wikibase.entityPage.entityLoaded' ).add( function ( entity ) {
     resolve( entity );
  } );
} );

See also: the related ticket, the code of wikibase.entityPage.entityLoaded.

If you have any question or need support with migrating specific gadgets, feel free to write to me. Thanks, Lea Lacroix (WMDE) (talk) 09:29, 18 July 2017 (UTC)

Updated DuplicateReferences, Move, AuthorityControl, Preview, and my user scripts. Matěj Suchánek (talk) 07:27, 24 July 2017 (UTC)

Improvements for the Query Helper

Hello all,

Following our prototype, the Query Service will soon include a new version of the Query Helper. This area will be located on the left side of the editor, allowing beginners to build queries from scratch, using context-aware filters.

If you don't want to use it, you can remove it (with the X button) or display it again (with the i icon in the menu). Lea Lacroix (WMDE) (talk) 07:55, 24 July 2017 (UTC)

Last days to submit a project for the WikidataCon

Hello all,

Since the WikidataCon is an event designed for and by the Wikidata community, the program of the conference is open and all the attendees can suggest ideas, talks and workshops to be included.

The deadline for proposing projects is July 31st, next Monday. After that, the program committee will review, select and organize the projects to fit in the schedule, and announce the final program around September 1st.

If you want to participate to the program and make sure that the WikidataCon contains the topics you want, it's time to submit one or several proposals! A lot of different formats are possible: talk, workshop, demo, lightening talk, round table, discussion, request for comment, meetup, hackathon, sprint...

You can find a list of ideas here. You're welcome to choose one of these topics and make it yours. We have for example:

  • demo all the useful tools for Wikidata,
  • an overview on how Wikidata is used in Wikipedia,
  • some thoughts about the community (how do you deal with vandalism?),
  • discussions about ontologies and data model,
  • showcase of a successful project,
  • meetups for thematic projects... and many more.

Thanks in advance for making the program of the WikidataCon amazing :) Lea Lacroix (WMDE) (talk) 08:15, 24 July 2017 (UTC)

New PD images of insects - superb quality

 
Long-horned bee, female (Apidae, Svastra petulca (Cresson))

The Insects Unlocked (Q33236104) project at the University of Texas at Austin are making superb, high res images of insects, like the one above, available in the public domain - almost a thousand, so far. I'm uploading them to Wikimedia commons, in Commons:Category:Photographs by Insects Unlocked.

Please help, by adding categories on Commons, attaching them to Wikidata items. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:20, 24 July 2017 (UTC)

Wikidata weekly summary #270

wbsearchentities with ElasticSearch test

Hi!

In the course of work on improving search in Wikidata, we now have an implementation of entity prefix search (wbsearchentities API) using ElasticSearch via CirrusSearch extension.

This search mode can be used via useCirrus=1 parameter on wbsearchentities API (without the parameter, the API works as before). Example:

  https://www.wikidata.org/w/api.php?action=wbsearchentities&search=Green&format=json&language=de&uselang=de&type=item&useCirrus=1

The results format is the same as before, but the search is driven not by SQL but by ElasticSearch.

We've created a test page to test newly developed search of Wikidata entities via wbsearchentities here: http://elastic-wikidata.wmflabs.org/wb.html

This page compares the results of old (SQL) and new (ElasticSearch) searches. Please use this page to test the search and if there's something broken please tell us.

Please note the following:

  • ElasticSearch parameters are not fully tuned yet, so there might be weird results, wrong ordering, etc. The purpose of this test phase is to discover and weed out such occurrences and figure out the best tuning parameters. As such, discussions of ideas and things that are not working right are welcome. If you plan to make a lot of tests - which is great! - please summarize. You can post feedback on the list, or on the Wikidata search talk page or anywhere on the wiki, please send me a link :)
  • The test page UI is kind of rough, so it relies on language and type parameters to be accurate and does not have any error control. It's just a test :) Suggestion on improving it are welcome though.

--Smalyshev (WMF) (talk) 20:59, 25 July 2017 (UTC)

@Smalyshev (WMF): Thanks for letting us know about this. It doesn't suffer from this problem, which is quite great if this new search implementation can be deployed soon. Mahir256 (talk) 21:51, 25 July 2017 (UTC)

Qualifiers on qualifiers?

Petrus Scriverius (Q2063138) was a 16th-century painter writer that was active in Leiden (Q43631) in circa (Q5727902) 1593. As best I can tell, this is usually registered on Wikidata with work location (P937) set to Leiden (Q43631) and the qualifier point in time (P585). But how do you specify that the point in time is approximate (circa)? Adding sourcing circumstances (P1480) set to circa (Q5727902) would apply to the location, not the time; and there is no way I can find to add a qualifier to the qualifier. Help? --Xover (talk) 07:30, 26 July 2017 (UTC)

Feel free addind also for time. Ex. Pythagoras (Q10261): date of death (P570) 500 BCE sourcing circumstances (P1480) circa (Q5727902). - Kareyac (talk) 09:40, 26 July 2017 (UTC)
@Kareyac: I'm not sure if that was the question. The problem seems to be that point in time (P585) is a qualifier yet in Xover's example that can't take a qualifier on itself unlike the non-qualifier property date of death (P570) in your example.
@Xover: Yes, I ran into that problem, too, haven't a good solution though, sorry. We could create value items for sourcing circumstances (P1480) that refer to a qualifier or a "sourcing circumstances of qualifier" property, both approaches would only shift the problem to the case when there is more than one qualifier. --Marsupium (talk) 16:49, 26 July 2017 (UTC)
Kareyac, work location (P937) should be accompanied by work period (start) (P2031) and work period (end) (P2032) properties, not by point in time (P585) qualifier. --Jarekt (talk) 17:09, 26 July 2017 (UTC)
@Jarekt: Not sure work period (start) (P2031) in circa (Q5727902) 1593, work period (end) (P2032) in circa (Q5727902) 1593 can be a good desition for Petrus Scriverius (Q2063138) from Leiden (Q43631). Сan floruit (P1317) be used in this case? - Kareyac (talk) 17:38, 26 July 2017 (UTC)
@Jarekt: I don't think so. Is that stated somewhere, a consensus somewhere? I've used point in time (P585) and for ranges start time (P580) and end time (P582) so far. I think work period (start) (P2031) and work period (end) (P2032) for expressing a floruit range, not limited to some location? --Marsupium (talk) 18:03, 26 July 2017 (UTC)
I did not know about floruit (P1317) and have never seen floruit (P1317) used and it seems like a better fit than work period (start) (P2031) and work period (end) (P2032) properties. Floruit is defined as a "date or period during which a person was known to have been alive or active". It seems to me that work period (start) (P2031) and work period (end) (P2032) define Floruit period and floruit (P1317) Floruit date, but they are not restricted to any location. I guess I now understand better Xover original question. We have a location Leiden (Q43631) and want to express that the person was active there around some time, we can add point in time (P585) qualifier for precise time or start time (P580), start period (P3415), end time (P582), end period (P3416) for time brackets, but it is hard to add "circa" to those dates. One solution would be to do work location (P937) set to 1590s which is a bit more broad than circa 1593. --Jarekt (talk) 18:25, 26 July 2017 (UTC)
Right. I'm not sure why we need both floruit (P1317) and work period (start) (P2031) + work period (end) (P2032) (which latter have aliases "floruit (start/end)"), but floruit in general is normally used to indicate the total span of years an artist was active (flourished), and if birth and death dates are not known then floruit dates are used as a substitute. Often, floruit dates come from picking the start of the first, and end of the last, of a list of times when the artist was known to be active. Thus there is a need to list Petrus Scriverius as being active in Leiden in circa 1593, in Amsterdam around (near) 1597, and in Oudewater from 1656 to 1660. His floruit (or "work period") dates would then be c. 1593 to 1660. Tying it to a place (rather than just a series of dates) is important to distinguish multiple artists with similar names, working in similar mediums, on similar motifs, in the same general area (which, you might be surprised to learn, is not actually uncommon). For instance, in a different case we're trying to sort out right now, there may be as many as three painters of the same name, with overlapping lifespans, painting peasants and rural scenes, in the Spanish Netherlands. The sources use things like time active in a particular city to separate them.
In any case, would it be completely off base to say that this is in fact a fundamental limitation of Wikidata's current information model? It seems to me that to handle this requires either properties or values to be multi-axial: i.e. either a property needs to be able to take multiple values (a place and a time, each of which can have attached qualifiers), or the value assigned to a property needs to carry associated properties directly. This seems to be half-true for time values (unless the UI is fooling me there), where the value carries a precision with it (that really should have been a qualifier for consistency), but not for other values and not for expressing things like level of certainty (as distinct from precision).
How about, just as a strawman proposal, if dates could have inherent certainty as well as precision? I mean like when you edit a property that takes a date value, the UI drops down a little box or card that tells you what calendar the date is in, and what level of precision (i.e. here often "year"-level precision), and could also let you choose level of certainty (circa, near, disputed, etc.). If the "date" data type inherently carries that information, then every single property that takes a date value will be able to express this, and, crucially, every qualifier that takes a date will also be able to express this.
I'm sure the servers would melt and the WMF would go broke and it'd probably hasten the heat death of the universe and so on... But wouldn't that be a much more elegant and general solution for this problem than arbitrarily nested qualifiers on qualifiers? What are the other ways this could be handled? --Xover (talk) 19:37, 26 July 2017 (UTC)

Rising Dispatch Lag

This is a warning. The dispatch lag has been steadily rising. It is now at 9.68 hours (9 hours and 44 min). The dispatch lag will go high again. Again, this is just a warning. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 03:45, 27 July 2017 (UTC)

@Muhammad Abul-Futooh: Mahir256 (talk) 04:50, 27 July 2017 (UTC)
Editing a project that has a serious lack of data has a high priority. When the dispatch lag cannot be kept under control, it is better not to have all the changes goto recent changes and have the needed edits made in stead. We can do without warnings. Thanks, GerardM (talk) 05:50, 27 July 2017 (UTC)

Automatically convert Article to Items

If i have a table with two columns, and the first one is:
Japan
Israel
Germany
Japan

How can I convert this lisr to this automatically:
Q17
Q801
Q183
Q17

Is there a tool for that? I tried doing this in Petscan, but it will only output: Q17
Q801
Q183

Without the last line, which I need. Thanks--Mikey641 (talk) 12:43, 27 July 2017 (UTC)

The Wikipedia plugin for Google Sheets can do this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:35, 27 July 2017 (UTC)
@Pigsonthewing: Thanks! I've been looking for something like this for a long time.--Mikey641 (talk) 13:37, 27 July 2017 (UTC)
@Mikey641: [ec] Actually called: "Wikipedia and Wikidata Tools". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:39, 27 July 2017 (UTC)

HELP: Associate article en:GRB 160625B with Item Q33520039?

HELP: First time user re WikiData - How do I associate article en:GRB_160625B with Item Q33520039? - TIA - Enjoy! :) Drbogdan (talk) 16:13, 27 July 2017 (UTC)

@Drbogdan: Look up here. To add a sitelink go to the right top box and press edit.--Bigbossfarin (talk) 16:49, 27 July 2017 (UTC)
@Bigbossfarin: Thank you for your reply - and help with this - all now seems *entirly* ok - Thanks again - and - Enjoy! :) Drbogdan (talk) 17:23, 27 July 2017 (UTC)

Accessible editing buttons

--Whatamidoing (WMF) (talk) 16:56, 27 July 2017 (UTC)

What country are Israeli Jewish settlements in?

moved from my talk page. Ijon (talk) 05:42, 18 July 2017 (UTC)

Hi. Regarding these your edits from late 2016, removing country (P17) isn't the best choice, even if it's values might be considered controversial by one or another part of people. We need a country (P17) claim for each item about localities, and usually in case of disputed territories there are added and kept both values. What do you think about this? --XXN, 19:35, 9 July 2017 (UTC)

Hi, XXN. They are outside the international borders of the State of Israel, even according to the government of Israel. I think the most accurate value for country (P17), if any, would be State of Palestine (Q219060). It may feel incongruous to those Jewish settlers who certainly want to believe they are in Israel, and the government of Israel is (criminally, according to international law) giving them benefits of Israeli citizens and applying Israeli law to them, but it is still a matter of fact that those places are outside the borders of the State of Israel. Ijon (talk) 07:01, 11 July 2017 (UTC)
I think the right thing to do is adding both State of Palestine (Q219060) Israel (Q801) because it's disputed.--Mikey641 (talk) 22:01, 17 July 2017 (UTC)
Mikey641: What, exactly, is disputed, though? Can you provide any reliable source for Israel's borders extending to include those settlements? Ijon (talk) 00:01, 18 July 2017 (UTC)
The Israeli government itself considers the territory to be disputed, but I don't think that's the primary issue here. I think there might be different views about what the country (P17) property is for. The government over the areas in question is clearly Israel, and the residents are Israelis. IIUC, the government doesn't claim that the area is in Israel, but it does recognize the national affiliation of the settlements to be Israel. Foreign powers don't consider the settlements themselves to be non-Israeli, although they do consider the territory they're located on to be so. As such, I'd recommend listing them as being exclusively Israeli.
Note that P17 is also used for non-territorial entities, such as government entities, armies, legal entities, and groups of people. --Yair rand (talk) 02:32, 18 July 2017 (UTC)
Let's move to WD:Project chat to get more opinions. I'm open to any solution that respects facts. It is IMHO incorrect to state as fact that the Jewish settlements in the West Bank are in Israel, even according to the Israeli government, however gladly it applies its laws there and supports the war crime that these settlements are (the fact of their being a war crime, unlike the fact about Israel's borders, is disputed by the Israeli government). It is true and accurate Jewish Israeli citizens live there, under Israeli military protection, enjoying Israeli infrastructure (such as electricity) etc., but none of these things change the internationally recognized borders of the State of Israel.
Thoughts, anyone? Ijon (talk) 05:42, 18 July 2017 (UTC)
Well, it is to some extent disputed if Israel is a recognised nation at all. Same thing could be said about Palestine. My opinion is "add both" and use qualifiers to describe the problems with those claims. Internationally recognised law and borders does not always agree with how things de facto works. Some here still claims that the way Norway separated from Sweden was not according to international law. I know some claim the same thing about Texas being a part of US. But international law does not always change how things de facto works. -- Innocent bystander (talk) 09:02, 18 July 2017 (UTC)
@Ijon: Indeed, but why not say those statements as disputed territory (Q15239622) or state with limited recognition (Q15634554)? --Liuxinyu970226 (talk) 11:21, 20 July 2017 (UTC)
As a general rule, don't add statements if you're not sure you can find a source to support them. This is especially true of statements that could be disputed.--Melderick (talk) 16:08, 18 July 2017 (UTC)
It is not unusual for towns or settlements in one one country to be populated mostly by people from some other country, such places still should have country (P17) property matching current internationally recognized borders. Almost any territory might be disputed by someone, but that is not enough to give those views equal footing and use "add both" solution. --Jarekt (talk) 19:29, 19 July 2017 (UTC)
There are plenty of reliable sources calling them Israeli settlements, and presumably none calling them Palestinian settlements. There are relatively few sources saying that they're "located in Israel", compared to saying that they're in the Palestinian territories. Which attribute is associated with P17? Location, or national/administrative affiliation? --Yair rand (talk) 20:30, 19 July 2017 (UTC)
Comparing those with an embassy, the answer is P17 = Location (in which country is the embassy, not which country is administrating it).--Melderick (talk) 21:55, 23 July 2017 (UTC)
Here's an essay I wrote on Commons about a similar issue: commons:Commons:Disputed_territories. I think most of it applies here too. --99of9 (talk) 23:41, 19 July 2017 (UTC)
Removing the values seems to be clearly wrong. On the one hand, even if you think that the claim is wrong it would be likely better to deprecate it then to remove it. On the other hand, it seems to me like the Israeli government does consider those settlements to belong to Israel. The fact that according to your beliefs they don't belong isn't ground for removing the claim. You can add "Disputed by". ChristianKl (talk) 07:28, 20 July 2017 (UTC)
Excuse me, ChristianKl but it is unfair to call facts "my beliefs". These are geographical facts: look up these settlements, and look up Israel's international borders. The former are outside the latter, and even the government of Israel does not dispute that they are. That government nonetheless extends its power over them, in a way that seems illegal, but in terms of international borders, those settlements (of Israeli citizens) cannot be said to be "in Israel". It is my understanding that country (P17) should refer to the country in whose territory (whether contiguous or not) the place lies. This is not the case for Israeli settlements in the occupied territories. I remain confused about what should be their "country"; perhaps none? Ijon (talk) 10:05, 21 July 2017 (UTC)
Or perhaps West Bank (Q36678), which appropriately is an "occupied territory", thus permissible as a country (P17), and isn't "the Palestinian Authority" (which would be misleading and factually incorrect, as it wields absolutely no power over the Israeli settlements). Would that be a good solution? Ijon (talk) 10:08, 21 July 2017 (UTC)
Or Palestinian territories (Q407199). XXN, 10:46, 21 July 2017 (UTC)
That could work too, XXN, although there are no longer any Jewish settlements in the Gaza strip (which this term includes), so West Bank (Q36678) would be more specific and accurate. Does this sound like a good solution, folks? Ijon (talk) 15:25, 24 July 2017 (UTC)
I support the use of West Bank (Q36678) rather than the Gaza-inclusive Palestinian territories (Q407199); "West Bank" is also used in running text of lead paragraphs on the pages of several settlements I inspected. Note, also, that the definitions in West Bank (Q36678), and assorted aliases across the various languages, are both imprecise and inconsistent. It appears that the designation "West Bank" is synonymous with the Israeli geographic designation "Judea and Samaria", though its definition in the item Judea and Samaria Area (Q513200) is of an "administrative district" of the State of Israel" and therefore unsuitable for use as a country (= geographic) designation. Agreed? -- Deborahjay (talk) 07:58, 28 July 2017 (UTC)

Is there a way to easily add a page linked through identifier as a reference?

I am working on c:Category:Creator templates with Wikidata link: mismatching birthdate and other categories listing pages where local dates do not match Wikidata dates (or software can not compare them easily). I often have to check references to verify if dates were copied correctly. Many references are missing and if I find matching data in one of the databases linked through identifier I add them as a reference. Is there some gadget or tool to easily add an identifier as a references? Perhaps without multiple cutting and pasting. That would be very helpful. --Jarekt (talk) 12:36, 28 July 2017 (UTC)

+1 - I was thinking references in Wikidata are nice in being structured, but the UI could be much better, this would be one help ArthurPSmith (talk) 03:52, 29 July 2017 (UTC)

Strategy discussion, cycle 3. Challenge 5

There are only three days left (plus today) to take part in Cycle 3 of the Wikimedia strategy discussion. Insights to the last challenge our movement is facing has just been published. The challenge is: How does Wikimedia meet our current and future readers’ needs as the world undergoes significant population shifts in the next 15 years?

The previous challenges are:

  1. How do our communities and content stay relevant in a changing world?
  2. How could we capture the sum of all knowledge when much of it cannot be verified in traditional ways?
  3. As Wikimedia looks toward 2030, how can we counteract the increasing levels of misinformation?
  4. How does Wikimedia continue to be as useful as possible to the world as the creation, presentation, and distribution of knowledge change?

On this page, you may read more, and suggest solutions to the challenges. Also, if you're interested in related discussions that are taking place on other wikis, please have a look at the weekly summaries: #1 (July 1 to 9), #2 (July 10 to 16), #3 (July 17 to 23).

In August, a broad consultation will take place, but it'll differ from what we've been conducting since March. This is your last chance to take part in such a discussion! SGrabarczuk (WMF) (talk) 13:33, 28 July 2017 (UTC)

Official portraits of Members of Parliament of the United Kingdom

Official portraits of 620 of the United Kingdom's 650 MPs are now on Wikimedia Commons, in Commons:Category:Official United Kingdom Parliamentary photographs 2017. As can be seen here:

there are four versions of each.

Please help to add these to the relevant Wikidata items (or suggest a Bot that might do so), along with the categories that are currently being created, and to Wikipedias in your own language(s) which don't pull images from Wikidata. I suggest that the "crop 2" format is most suitable; and please use those with names beginning "Official portrait of " as other, duplicate, versions are liable to be deleted. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:17, 27 July 2017 (UTC)

Andy, are you aware of WDFIST? --Edgars2007 (talk) 14:42, 27 July 2017 (UTC)
Alas, no. How would it help here? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:53, 27 July 2017 (UTC)
S*, right, not this time. Sorry, Thursday was a hard day :) --Edgars2007 (talk) 16:40, 29 July 2017 (UTC)

Benefactor

How do you indicate that Mrs Alexander was the benefactor of the University of California Museum of Paleontology? Thanks, GerardM (talk) 11:27, 29 July 2017 (UTC)

occupation = benefactor? Slowking4 (talk) 12:22, 29 July 2017 (UTC)
affiliation (P1416) with subject has role (P2868) benefactor (Q4887411) as qualifier. Sjoerd de Bruin (talk) 12:31, 29 July 2017 (UTC)
P2868 as qualifier would be more precise d1g (talk) 12:43, 29 July 2017 (UTC)

Constraint statements update

Hello all,

following the announcement two weeks ago, constraint statements have been enabled on Wikidata and all constraints have been migrated from templates on property talk pages to statements on the property itself (thanks Ivan for doing the actual migration!). You can now edit the constraint statements on a property, and the change will be reflected almost immediately when you check the constraints with the gadget or special page (see usage instructions). You can also see if there are any problems with the constraint statement on that statement itself: if the constraint parameters are invalid, the gadget adds a report similar to the normal constraint report, but with a bright orange icon (I’ve added an example on Sandbox-String (P370)).

However, there is one unanticipated problem: a few constraints (specifically, on taxon name (P225), GNIS ID (P590), IMA status and/or rank (P579), Commons category (P373), category's main topic (P301), and INSEE municipality code (P374)) have so many exception to constraint (P2303) qualifiers that they don’t fit in the constraint database, so that a constraint check on any item with a statement for one of these properties crashes with a fatal error (if you use the gadget, you won’t see anything, the constraint reports will just be missing). We’ll deploy a fix for this as soon as possible, but seeing as it’s Friday evening, that might still take a few days. In the meantime, you can also fix this problem on Wikidata by reducing the exception lists of those constraints, or by removing the constraints completely. (The fix we’ll deploy will not actually add support for such long exception lists, it will just reduce the impact of the error and report it properly. We have no immediate plans to fully support extremely long exception lists.)

If you have any questions, feel free to ping me. Thanks! --Lucas Werkmeister (WMDE) (talk) 21:37, 21 July 2017 (UTC)

What about usability, Lucas Werkmeister (WMDE)? taxon name (P225) started with 20.508 Bytes (20 June 2017) and has now (16 July 2017) - after migration - 631,637 bytes. It is - in my opinion - not maintainability anymore. --Succu (talk) 22:13, 21 July 2017 (UTC)
IMHO these constraints should just be deleted, maintaining such a long list of exceptions is against the spirit of the system anyway. − Pintoch (talk) 22:32, 21 July 2017 (UTC)
That’s my personal opinion as well, and finding out that the long exception list broke constraint checks on most major items (anything with a Commons category) didn’t exactly help ;) --Lucas Werkmeister (WMDE) (talk) 13:00, 22 July 2017 (UTC)
Convert it to complex SPARQL constraint as for now.
  Comment We should rework UI to display "best" values first and "other values" should be collapsed.
Something similar should be done with long list of qualifiers (show first 10, provide an option to show all) d1g (talk) 03:54, 22 July 2017 (UTC)
  •   Comment Of the ones you mention, the first constraints with the most exceptions seem to be the most appropriate (P225, P590). It's just that they have tons of uses and so even 1/1000 of exceptions start adding up. P374 should probably be solved with ranks instead. P373 can eventually be replaced with some Wikibase function, so we probably don't want to waste more time with it.
    Given the big progress you made developing Wikibase for constraints, I think we should be able to bear with the fact that for two properties another custom solution needs to be used and that eventually we will have to find a way to scale the system to properties with more uses. --
    --- Jura 23:34, 21 July 2017 (UTC)
Personally, I would welcome custom solutions for these constraints with many exceptions. If they’re “distinct values” constraints, those are implemented in SPARQL anyways (you can see the query templace here), so it should be possible to do the same thing in an external service.
I should also mention that many constraint checks take statement ranks into account. Specifically, a “single value” constraint will not count deprecated statements. Would that help with the GNIS ID (P590) constraint? (I’m not familiar with that ID, so I don’t know if it would make sense to make one of the multiple values deprecated.) --Lucas Werkmeister (WMDE) (talk) 13:00, 22 July 2017 (UTC)
The single value constraint on P590: I'm not sure: the same item just has multiple identifiers. It would probably be somewhat random if we tried to determine which one is the preferred one. One could question if it's really worth listing all exceptions on the property itself. Maybe it would be sufficient to list these elsewhere or not at all (we don't do it for P214, but P214 has some automated identifier merging). A difference to the distinct value constraint on P374 is that there, the same identifier applies in a different period of time.
--- Jura 14:34, 22 July 2017 (UTC)
  • For the taxa is it possible to make another type of constraint that requires (Distinct values OR different from (P1889))? --99of9 (talk) 01:14, 22 July 2017 (UTC)
  • For P225 a work-around would be most welcome. My guess would be that this should take the form of 1) migrating the exceptions to separate pages (plural) or 2) like 99of9 suggests, some extra property/ies to be taken into account (that is, an item that holds such a claim will be taken to be an exception, even though not listed as such). In the case of separate pages of exceptions, these almost certainly have to be plural as the list of exceptions had been reduced quite substantially by leaving out many items (those including a P2743 claim) that should really be listed. Nevertheless, in spite of that reduction, the list of exceptions on the Talk page skirted the limits of what the editor could load, and now that it has been migrated to the property page cannot be loaded by the editor at all. - Brya (talk) 05:21, 22 July 2017 (UTC)
  • Update: A quick fix for the exceptions issue has been deployed. This does not add support for extremely long exception lists, it just makes sure they don’t crash the constraint check. For example, you’ll still see a violation on every IMA status and/or rank (P579) statement because the parameters of the “item requires claim” constraint can’t be read, so the required property is missing (and you can also see that on P:P579#P2302). --Lucas Werkmeister (WMDE) (talk) 14:16, 24 July 2017 (UTC)

Gadget for changing property?

Is there some gadget or user script for changing properties with all content? Example situation P123 = Q1, Q2, Q3, Q4, Q5, Q7 and I want to change P123 to P456. Now I must add P456, then all values from old property and then delete all from P123. But better would be if I change only property and all values moves to new property. JAn Dudík (talk) 06:54, 26 July 2017 (UTC)

I would find this very useful as well. - PKM (talk) 20:06, 26 July 2017 (UTC)
There is no gadget to my knowledge. However, bots can do this. —MisterSynergy (talk) 20:20, 26 July 2017 (UTC)
See phab:T149905.--GZWDer (talk) 15:20, 30 July 2017 (UTC)

Q34000000

We now have this.--GZWDer (talk) 12:39, 30 July 2017 (UTC)

It's just another scientific article. We have them in spades. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:04, 30 July 2017 (UTC)

Current status of a power station

How do I add the "Status" field of Infobox power station on Wikidata? I'm working on making this infobox to work like the amazing Infobox telescope, but I don't know how to migrate the field from Wikipedia articles to the corresponding Wikidata item page. Random example of that field in use: Lakvijaya Power Station. Rehman 13:44, 30 July 2017 (UTC)

Almost the same question as #Proposed_building_or_structure d1g (talk) 14:04, 30 July 2017 (UTC)
@Rehman: I support your edits at Lakvijaya Power Station (Q6479997). Such statements could be used in SPARQL relatively easy.
We don't have specifically designed LUA module to re-use data back in Wikipedias. d1g (talk) 14:10, 30 July 2017 (UTC)
Oh okay. So you mean, at this moment, I can't get the "status" parameter to work exactly like how it is shown in the non-wikidata supported state of the power station infobox? Rehman 14:42, 30 July 2017 (UTC)
It is very possible for one item Module:Event - but I'm not used to work with LUA. d1g (talk) 15:11, 30 July 2017 (UTC)

Cities

Hello, I need a DB of all the cities in the world, an alternative to geonames org and since wikipedia seems to have a page of all the cities, I guess I could use this site. So im trying to learn the query, and made this query for all the cities of italy but the result is only 150 rows instead of 7000, so am i doing anything wrong ? The query:

SELECT DISTINCT ?cityLabel ?city ?official_name ?located_in_the_administrative_territorial_entity ?country WHERE {
  ?city (wdt:P31/wdt:P279*) wd:Q515.
  FILTER (?country = wd:Q38)
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
  OPTIONAL { ?city wdt:P1448 ?official_name. }
  OPTIONAL { ?city wdt:P131 ?located_in_the_administrative_territorial_entity. }
  OPTIONAL { ?city wdt:P17 ?country. }
}

Try it!

Thanks.  – The preceding unsigned comment was added by Widauser (talk • contribs).

@Widauser: proper place to ask it here: Wikidata:Request a query d1g (talk) 23:25, 20 July 2017 (UTC)

The following query uses these:

  • Properties: instance of (P31)    , subclass of (P279)    , country (P17)    
    1 SELECT DISTINCT ?c ?cLabel
    2 WHERE
    3 {
    4   ?c wdt:P31/wdt:P279* wd:Q486972. # 138 Results with Q515; 63553 Results with Q486972
    5   ?c wdt:P17 wd:Q38.
    6   
    7   SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
    8 }
    
We really need a good definition of City here. -- Innocent bystander (talk) 07:02, 21 July 2017 (UTC)

Thanks, it works! Now how Can I select the cities bigger than 10000 people ? I tryed "?c wdt:P1082 ?population > 10000." but no... About definition of City, When I asked "all the cities in the world" I ment also the small towns. "human settlement" is what I needed but its not very user-friendly... I'd call it 'cities and towns', and a mall in the desert is another thing I think...  – The preceding unsigned comment was added by Widauser (talk • contribs).

We have several definitions in Swedish only, some of them not fiting with subclasses of "human settlement". -- Innocent bystander (talk) 17:36, 24 July 2017 (UTC)

Finally I used this query:

SELECT DISTINCT ?c ?cLabel ?official_name ?country ?population WHERE {
  ?c (wdt:P31/wdt:P279*) wd:Q486972.
  ?c wdt:P17 wd:Q38.  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
  OPTIONAL { ?c wdt:P1448 ?official_name. }
  OPTIONAL { ?c wdt:P17 ?country. }
  ?c wdt:P1082 ?population.
  FILTER ( ?population > 14999)
}

Try it!

Than I made this rude comparison between wikimedia and geoname; "limit" is the min population, the first number is the number of cities found on wikimedia and on geonames the second; In brackets there is the clause used for the country.

ITALY: limit 15000: 395 575 | limit 20000: 324 420 | limit 25000: 288 311

UK (?c wdt:P17 wd:Q145.): limit 15000: 108 755 | limit 20000: 91 588 | limit 25000: 84 468

GERMANY (?c wdt:P17 wd:Q183.): limit 15000: 804 1058 | limit 20000: 593 775 | limit 25000: 463 585

SPAIN (?c wdt:P17 wd:Q29.): limit 15000: 583 619 | limit 20000: 448 478 | limit 25000: 341 374

CANADA (?c wdt:P17 wd:Q16.): limit 15000: 147 224 | limit 20000: 122 176 | limit 25000: 106 152

USA (?c wdt:P17 wd:Q30.): limit 15000: 3334 3145 | limit 20000: 2564 2424 | limit 25000: 2062 1934

WORLD: limit 15000: 25185 23754 | limit 20000: 20148 19222 | limit 25000: 16742 16050

Now I'm extreamly confused, however considering the cities 25000+ the difference is smaller. Now I dont know if the reason is missing cities or wrong or outdated or missing population value. And of course I wonder which one is more reliable!

P.S. In the uk the difference is huge, there must be some error.  – The preceding unsigned comment was added by Widauser (talk • contribs). 22:46, 25 July 2017 (UTC)

  CommentIn some languages there is not difference between town (Q3957) and city (Q515). JAn Dudík (talk) 06:17, 31 July 2017 (UTC)

Another problem. I was happy to see that the cities has the geonames id! Unluckily many are wrong / outdated :(  – The preceding unsigned comment was added by Widauser (talk • contribs).

A Big Task: Adding language descriptions to scientific article items

As you may know, Research Bot (talkcontribslogs) has been importing mass scientific articles, but they are in English only. I want to try and add descriptions in Spanish, French, Italian, German, Chinese, and a few other languages. The descriptions will be 'scientific article' translated into each language.

  • There are millions of items on scientific items.
  • There is no way this will be possible without a bot.

Since my bot requests are usually denied, I wanted to discuss it here first. I already have a script for the first 1 million items. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 04:11, 28 July 2017 (UTC)

@PokestarFan: You should put it on GitHub sowe can see it first before discussing it. Mahir256 (talk) 04:50, 28 July 2017 (UTC)
Is there any benefit in adding "scientific" to the description? Are we trying to sort between "scientific" articles and other articles in journals?
--- Jura 05:18, 28 July 2017 (UTC)
Only "article" is not enough. "Scientific article" is fine to me. BTW, in all these articles, one has instance of (P31) = scholarly article (Q13442814) that is pretty clear. Pamputt (talk) 05:38, 28 July 2017 (UTC)
Also note that a scholarly article (Q13442814) is different from (P1889) an academic journal article (Q18918145). Mahir256 (talk) 23:49, 28 July 2017 (UTC)
This is partly done by User:Edoderoobot. --Succu (talk) 07:06, 28 July 2017 (UTC)

I am OK with adding missing descriptions with bot, but...:

1) There are several styles for academic articles: some include the year (published in YYYY; Q21030645), others include the year-month-day (Q21032450). Perhaps there are more styles. We should set a standard.

2) User:Edoderoo adds one description per edit, 7 edits for 7 descriptions. That is very inefficient. We should add all descriptions in the same edit.

    • The main reason for this several edits: I started with Dutch descriptions, as I mainly work on Dutch descriptions. I've added a few more languages, partly by request, but got comments from several users that they like/must have/need to/aggreed/personally prefer/etc for their language a different format in the date. Therefor it happened that I needed 12 edits for 7 descriptions. I only do not see any necessity for "efficiency", the only one seeing the edits are the people that look to the history of the item. If it's really such a big deal, every item should get all 250 language descriptions upon creating the item, but I don't see the need for that either. Edoderoo (talk) 07:50, 28 July 2017 (UTC)

3) Bot permissions are granted in Wikidata:Requests for permissions/Bot, not here.

Regards. Emijrp (talk) 07:24, 28 July 2017 (UTC)

  • Year first published would be helpful.
    --- Jura 07:47, 28 July 2017 (UTC)
@PokestarFan: Why are your "bot requests ... usually denied"? -- Innocent bystander (talk) 08:17, 28 July 2017 (UTC)
@Innocent bystander: Feel free to read up here. Mahir256 (talk) 23:49, 28 July 2017 (UTC)

I suggest "Scientific article published in [publication] in [year or full date]". That is generally sufficient to disambiguate two articles with the same title. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:07, 28 July 2017 (UTC)

Same, we should prefer more specific dates, but exact information (day) could be missing. d1g (talk) 10:34, 28 July 2017 (UTC)
In general, I do not think it is useful to run bots which only add descriptions to items. Auto-generated descriptions should be handled directly by the software, so that when the generating algorithm is improved or fixed, all auto-descriptions are updated. − Pintoch (talk) 11:45, 28 July 2017 (UTC)
I've heard this said before; but it's still just vapourware. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:15, 28 July 2017 (UTC)
And probably awkwardly difficult in some languages. -- Innocent bystander (talk) 13:30, 28 July 2017 (UTC)
Automated descriptions have been in use in Reasonator for a long long time. Their quality is much better than what manual descriptions provide. I often edit items in order to trigger new descriptions so that I can disambiguate between items. Thanks, GerardM (talk) 13:49, 28 July 2017 (UTC)
  • AFAIK Emijrpbot already works on this task (from time to time) adding by ~20 descriptions. --XXN, 19:34, 28 July 2017 (UTC)
Though my bot code includes some translations for this kind of items, I have never launched it, because I saw that there are several styles (without date, with year, with year-month-day) and I was waiting for a consensus about it. Emijrp (talk) 20:52, 28 July 2017 (UTC)
A good decision. --Succu (talk) 21:15, 28 July 2017 (UTC)

Why mention both the journal and publication date/year but not the author(s)?

"Are cats fluffy?": "Scientific article by A Benjamin, C David, E Franklin et al, published February 31, 2013"
makes more sense to me than naming the journal. -- Innocent bystander (talk) 05:59, 29 July 2017 (UTC)
I would only include the year in the description, and maybe the first author sumarizing the rest with "et al.". Why only the year? Because not all papers have a YYYY-MM-DD date available, so we will end again with 2 description styles (YYYY and YYYY-MM-DD). Also, when papers are cited, usually only the (year) between brackets is added. Why not to include the author? Because you can't easily translate their names to non-ascii languages like Hebrew, Japanese, Chinese, Russian, etc, and we would end again with 2 description styles (with author and without authors). I found this issue when adding descriptions to films (see Hebrew). Also, we should have in mind that the description purpose is to be shown in searchbox dropdown, so long descriptions (including full date, authors or journal name) are truncated. Emijrp (talk) 06:39, 29 July 2017 (UTC)
@Emijrp: If I search for a single article, it is much more likely that I know who the main author is than that I know which journal it was published in. And that it works poorly in Hebrew and Russian looks like a poor reason to not add a useful description in languages with latin script. Since when do we have to have one single "description style" for all languages? From what I have understood, the policies/guidelines for descriptions and labels are only ratified in a handful of languages. My language (Swedish) have no guideline/policy at all in this case. The purpose of the description is to help when you search, but you do not have to have exactly the same wording in all items about disambigs, templates or categories to accomplish that. -- Innocent bystander (talk) 19:47, 29 July 2017 (UTC)
I agree that the exact publication date in descriptions is not really necessary, and only the year is sufficient. XXN, 17:31, 29 July 2017 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── The code is too buggy and I am shutting it down. PokestarFan • Drink some tea and talk with me • Stalk my edits • I'm not shouting, I just like this font! 13:34, 29 July 2017 (UTC)

Using tools for mass-editing such a huge number of items to add one by one descriptions is highly unrecommended. Leave this task for a bot, as there is no urgency to edit these items right now. --XXN, 16:34, 29 July 2017 (UTC)

My bot just got blocked, and I got told that I am an asocial person, because one Raspberry Pi was thought to block all Wikimedia servers. So my bot will not add nor fix nor mess up any descriptions, neither in Dutch, neither in another language. I think it's about time that others will pick up that task.... Edoderoo (talk) 14:52, 29 July 2017 (UTC)

That's not what I said. I think it's asocial to run a bot on 155 edits per minute while a project has been dealing with dispatch issues for over a month, while also still having issues in the bot code that have been mentioned months ago. Sjoerd de Bruin (talk) 15:08, 29 July 2017 (UTC)
Leaving aside the bot speed, your bot was adding English labels as Dutch ones, and it was also adding a description which we are trying to consensuate in this very section. Emijrp (talk) 16:42, 29 July 2017 (UTC)
There will never be another 'title' then the English one, and translating it would be more wrong then using the English one, therefor it's decided before to use the English label for the Dutch label too. For books, that often do get translated, this would not be acceptable. About the description: there ain't no consensus yet, so I can not (yet) add a consensuate description, so why complain? Edoderoo (talk) 18:37, 29 July 2017 (UTC)
In that case, Research Bot should add the same English label for all languages when it creates the item, and we will save million edits. If we are trying to reach a consensus, it is a polite thing to join the discussion and wait before doing more edits. Or do you want to launch your bot again later, doing the double of edits? Emijrp (talk) 19:26, 29 July 2017 (UTC)
How about just filling in the "title" field and display that?
--- Jura 09:34, 30 July 2017 (UTC)
That would be great. - Brya (talk) 10:35, 30 July 2017 (UTC)
  • Obviously, the more details are added to the description, the more maintenance the bots will have to do: e.g. if the publication date is changed from May 30 to June 30, all descriptions will need to be updated. If the same is found to have been published in some other place before, all description need to be updated, etc.
    --- Jura 09:34, 30 July 2017 (UTC)
  • For your information, PokestarFan submitted three batches to QuickStatements (interestingly, he stated he would use Descriptioner). I was asked to create an AbuseFilter to block these batches because at least one of those descriptions was grammatically incorrect. (We weren't able to stop the batches via the new interface, seems there are still bugs.) I hope you all understand now that if an administrator blocks a malfunctioning bot, it's only supposed not to let the data quality degradate. Matěj Suchánek (talk) 09:58, 30 July 2017 (UTC)
  • I think we have two intertwined but separate issues here:
  1. adding bibliographic metadata harvested from external databases to items about papers
  2. adding multilingual information to these items
The information we get in under issue 1 from places like Crossref, PubMed, PubMed Central or arXiv is almost exclusively English, and all the tools we have in this space (including Wikidata Integrator and Fatameh as well as Source MD) are geared towards English-only.
On that basis, I think it's best to have the multilingual part (i.e. issue 2) covered by separate tools (and yes, ideally bots), and I am grateful for Edoderoobot having started to do this systematically a long time ago and having supported ever more languages and evolving formats. I agree it would be good if we could compress all those description edits into just one, but there is nobody to blame that we do not yet have a tool to do that, especially since the detailed specs in terms of the description format haven't been worked out yet. Perhaps that could be something to work on during upcoming hackathons, e.g. at Wikimania?
As for the description format, I'm not sure what the best approach here is. When I am doing paper item descriptions in languages other than English, I usually only care for items missing a description in the respective language, and trying to add the equivalent of "scientific article" first. If that does not work because an item with the same label and description already exists, I go for things like "scientific article published in 2005". If that is already "taken", I sometimes vary the format, e.g. "scientific article (published 2005)" or "scientific article (published in 2005)" or so, and in the few cases where the options (whose numbers differ between languages) have been exhausted this way, I am adding either the month of publication or the journal. I agree that this is messy, but so are our descriptions in general (e.g. we do not have a standard way to describe humans or highways or anything other). I would very much welcome more systematic approaches here, e.g. multilingual property-based auto-descriptions in the Wikidata UI when no description exists but I don't see how I could contribute to make that happen.
--Daniel Mietchen (talk) 20:23, 30 July 2017 (UTC)
"vary the format" - we shouldn't vary "()" to "" to "[]". This is very unintuitive. I would argue even for ...1 ...2 prefixes based on Wikidata entity id but not this. d1g (talk) 23:21, 30 July 2017 (UTC)
@Daniel Mietchen: On the issue of multilingual articles, it is important to be mindful of non-English scientific articles, whose title (P1476)s and native-language labels, if DOI (P356)s for those articles are available, should be adjusted promptly from their Crossref data, with some way to detect the language of the title since Crossref doesn't store this (and doesn't even address it in its API documentation). Mahir256 (talk) 00:34, 31 July 2017 (UTC)

SearchAll gadget

Hey, I just wrote a gadget that helps you search in all namespaces in Wikidata (right now, if you search "Wikidata:Project chat" at the top of the page, it doesn't suggest you this page. If you enable "SearchAll" gadget in your preferences, when you search, it shows suggestions from all other namespaces (not including files namespace, because it will completely filled with unrelated results). See phab:T48251#3485164 for pictures and more details :) Best. Amir (talk) 09:48, 31 July 2017 (UTC)

Finally, thanks! It's just a bit slower than the normal search, but better than nothing. Stryn (talk) 11:05, 31 July 2017 (UTC)

Wikidata weekly summary #271

Bug with P279 property suggestion fixed

Hello,

Some of you may have noticed a problem occurring with items having subclass of (P279) statements. The property suggester was making inaccurate suggestions. This should now be fixed. If you encounter any problem on this topic, let me know. Lea Lacroix (WMDE) (talk) 15:09, 31 July 2017 (UTC)

Complex constraint: help needed

Please can someone add a complex constraint to ORCID iD (P496), that reports if the subject has a date of death before 2012? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:56, 31 July 2017 (UTC)

I take it you are interested in a constraint when the subject died after 2012 .. ? Thanks, GerardM (talk) 13:30, 31 July 2017 (UTC)
Could you elaborate why this is to be a "constraint"? Matěj Suchánek (