Wikidata

Wikidata:Project chat/Archive/2015/09

< Wikidata:Project chat‎ | Archive‎ | 2015
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

New gadget: DuplicateReferences.js

 
Screenshot of DuplicateReferences on Earth (Q2)

I've created a script to copy and paste references within an item some months ago and after fixing all bugs that occured, I made it available as a gadget. You can enable DuplicateReferences in your preferences. The source code can be found here.

A link will be placed next to every reference which allows you to copy that reference and insert it to another statement. This is very useful if one imports several statements from the same source and some users have been complaining about the waste of typing the same reference again and again. I hope this gadget helps you all adding more and more references to our statements.

Issues and feature requests can be made on the talk page of the gadget or even better, directly on phabricator on our lovely Wikidata-Gadgets project.

Best regards -- Bene* talk 09:57, 30 August 2015 (UTC)

Thx for this very useful tool. Is it possible to add a reference to the toolbar or to wikidata useful? Pyb (talk) 17:53, 30 August 2015 (UTC)
Hey Pyb, can you explain your idea a bit further? Do you want the gadget to remember a reference across items? -- Bene* talk 18:32, 31 August 2015 (UTC)
I often use the same reference on a lot of items. Right now we couldn't quickly add a reference from Wikidata Useful or from the toolbar. It would be great if we could customize our javascript with our favorite references. A gadget to remember a reference across items would also be a good idea. Pyb (talk) 10:54, 1 September 2015 (UTC)

Wikidata:Database reports/Constraint violations/Mandatory constraints/Violations

Could some people help with cleaning up Wikidata:Database reports/Constraint violations/Mandatory constraints/Violations? The page has been around 60k bytes for quite some time now. Sjoerd de Bruin (talk) 07:01, 1 September 2015 (UTC)

CEO

Hoi, CEO only makes sense in relation to an organisation a person is CEO of. My question is, how do I add this on a person? Thanks, GerardM (talk) 19:39, 31 August 2015 (UTC)

Use chief executive officer (P169) on the item that the person is CEO of. --Yair rand (talk) 20:48, 31 August 2015 (UTC)
Item position held (P39) chief executive officer (Q484876). While you're at it, possibly also occupation (P106) businessperson (Q43845). --Izno (talk) 01:44, 1 September 2015 (UTC)
The current practice has ONLY CEO on people. CEO is by its very name related to a company. Businessman is not. Thanks, GerardM (talk) 06:19, 1 September 2015 (UTC)
Use <position held (P39) chief executive officer (Q484876)> with qualifier (of (P642):'company item'). P642 is widely used as a qualifier to P39. Joe Filceolaire (talk) 21:21, 1 September 2015 (UTC)
@Filceolaire, Izno, GerardM, Yair rand: Eh guys, please add this usecase to WikiProject Reasoning. It seems especially suitable to handle these representation mismatch of the same things.
Use <employer (P108): 'company item'> with qualifier <position held (P39): chief executive officer (Q484876)>. @Filceolaire: of (P642) can be widely used, but it would be better to keep it as a last option, specially if other (much more precise) alternatives are available. Andreasm háblame / just talk to me 09:57, 2 September 2015 (UTC)
+1. Yeah that is better. Joe Filceolaire (talk) 10:01, 2 September 2015 (UTC)
P39 is meant to be a first class statement given that it has required qualifiers (though I appreciate the use). --Izno (talk) 11:53, 2 September 2015 (UTC)

BLP

I don't like that we don't have a policy about biographies of living persons here. See for example Della Rose Joel (Q20888791). Describing very young childeren of people who are still alive... I don't think that is a good idea. Sjoerd de Bruin (talk) 08:04, 1 September 2015 (UTC)

We already have Princess Charlotte of Wales (Q18002970), so why not. But statements should be sourced.--GZWDer (talk) 09:23, 1 September 2015 (UTC)
Practically it is "Kian" who can do the best job by harvesting information continuously from Wikipedias. It is a matter of training the statements that are of a particular concern. As it is, many items for people are insufficient. BLP is quality and it is imho best when we concentrate on specific attributes one at a time and improve quality in this way.
Consequently as we get better in BLP related information, we will be able to signal BLP issues to Wikipedias. Thanks, GerardM (talk) 09:24, 1 September 2015 (UTC)
There is more than one aspect to BLP but the most fundamental aspects is that we avoid libelling (or is it slandering) living persons because they can sue. We don't want <instance of:war criminal> or <occupation:thief> or similar added to an item about a living person without solid references to an actual conviction in court. I guess this is what we should be training bots to look out for.
As well as statements there is also the possibility of libel in Descriptions; especially as we have hundreds of different Descriptions in different languages. This will, I think, be much harder for bots to detect - another argument for automatically generating Descriptions from statements instead. Joe Filceolaire (talk) 19:30, 1 September 2015 (UTC)
As it is there is no way for us to have something functional without tools like Kian. Quality is quality and BLP is nothing different. When a Wikipedia makes a BLP booboo, we will capture those booboos in Wikidata. Having solid tools like Kian, we can be the best because we capture changes in Wikipedias as well. The notion that Wikidata will have sources for everything that may be problematic in BLP is not really realistic. What we need is being responsive to quality issues notified to us by tools. That will for a long time be the best we can do. Thanks, GerardM (talk) 20:48, 1 September 2015 (UTC)

Request for vandalism fighters in Wikidata

Hello, As you may know I'm building AI tools to fight vandalism and triage edits for human review. I want to know what kind of vandalism is common in Wikidata, what kind of items are common targets and things like that. Specially some examples can help me a lot. Your input is valuable to me, and if you don't want to share your ideas publicly (in case you might be thinking that vandals will see) please feel free to send me an email (ladsgroup at gmail). There is a paper published in this case and we can discuss about that too :) Best Amir (talk) 15:28, 2 September 2015 (UTC)

You can start with looking into the abuse filters (Special:Tags could also be relevant). Matěj Suchánek (talk) 16:03, 2 September 2015 (UTC)
Do you have any code up to look at so far? :) / where would it be? :) ·addshore· talk to me! 16:14, 2 September 2015 (UTC)
Pasleim's PLbot collects possible vandalism at User:Pasleim/Vandalism, Ivan A. Krestnin's KrBot collects removed sitelinks at User:KrBot/Lost links, Emaus' LinkRecoveryBot used to revert these removals but sometimes had false positives. Matěj Suchánek (talk) 16:34, 2 September 2015 (UTC)

Thank you. I will look into them and I probably ask some questions here. @Addshore: the source code is here it's based on revision scoring (m:ORES), the most important part is building features for the AI system and then it will train and builds the classifier. If you need help in running the code, tell :) Thanks Amir (talk) 16:58, 2 September 2015 (UTC)

See analysis and studies summarized at Help_talk:Description#Vandalism_problem. --Atlasowa (talk) 18:36, 2 September 2015 (UTC)

Introducing the Wikimedia public policy site

Hi all,

We are excited to introduce a new Wikimedia Public Policy site. The site includes resources and position statements on access, copyright, censorship, intermediary liability, and privacy. The site explains how good public policy supports the Wikimedia projects, editors, and mission.

Visit the public policy portal: https://policy.wikimedia.org/

Please help translate the statements on Meta Wiki. You can read more on the Wikimedia blog.

Thanks,

Yana and Stephen (Talk) 18:12, 2 September 2015 (UTC)

(Sent with the Global message delivery system)

First version for units is ready for testing!

Hi everyone :)

We've finally done all the groundwork for unit support. I'd love for you to give the first version a try on the test system here: http://wikidata.beta.wmflabs.org/wiki/Q23950

There are a few known issues still but since this is one of the things holding back Wikidata I made the call to release now and work on these remaining things after that. What I know is still missing:

If you find any bugs or if you are missing other absolutely critical things please let me know here or file a ticket on phabricator.wikimedia.org. If everything goes well we can get this on Wikidata next Wednesday.

Cheers --Lydia Pintscher (WMDE) (talk) 15:54, 31 August 2015 (UTC)

Yay! Looking forward to playing with this :) —Galaktos (talk) 16:06, 31 August 2015 (UTC)
The link in the unit selection menu goes to /entity/Q1234 instead of /wiki/Q1234; is that a bug or just a configuration problem of the beta instance? —Galaktos (talk) 16:13, 31 August 2015 (UTC)
That's intended as that is the URI of the concept and that is semantically correct there. --Lydia Pintscher (WMDE) (talk) 16:20, 31 August 2015 (UTC)
Really? ’Cause following the link gives me a 404 error :D —Galaktos (talk) 16:22, 31 August 2015 (UTC)
Ohhhh ok. Yeah that redirecting is probably not set up for beta. It works on Wikidata itself. Thanks for checking! --Lydia Pintscher (WMDE) (talk) 16:24, 31 August 2015 (UTC)
Oh I see, over here /entity links work. Thanks! —Galaktos (talk) 16:25, 31 August 2015 (UTC)
  • Does not recognize aliases for units. "Meter" must be spelled in the American style; "metre" doesn't work.
  • No ability to enter an exact quantity. To enter an exact quantity one must enter, for example, 299792458±0, which the average user is unlikely to guess.
Is there a list of allowed units somewhere? Or can any entity on wikidata be a unit? Should there maybe be a class ("unit" - standard of measurement) with units instances of that or of subclasses? Also, will the rollout include creating all the properties listed on Wikidata:Property_proposal/Pending/2? ArthurPSmith (talk) 17:35, 31 August 2015 (UTC)
Yes any item on Wikidata can be used as a unit. There should later probably be constraints checks to find items that use items that likely should not be a unit. Once the code is rolled out the people with property creator rights can create the properties that have been waiting for unit support. That is however not a thing the dev team will do. I don't expect it to take long for that to get done. --Lydia Pintscher (WMDE) (talk) 17:37, 31 August 2015 (UTC)
ok, I guess that's fine. I didn't find it before but we do have unit of measurement that every unit should be an instance of (directly or indirectly) so doing the constraint check should be relatively straightforward. Probably needs some experience in use though. ArthurPSmith (talk) 17:48, 31 August 2015 (UTC)


If units have to be entities, it would sometimes be necessary to create an entity just to support a particular quantity, such as the gravitational constant 6·67428 × 10⁻¹¹ m³ kg⁻¹ s⁻². Giving that unit a label and finding it may be a challenge. Jc3s5h (talk) 18:34, 31 August 2015 (UTC)
can you think of a good alternative? Allowing the unit to be a combination of entities in some way perhaps? That seems overly complicated for most situations though. ArthurPSmith (talk) 19:11, 31 August 2015 (UTC)
aliases : there is aliases to help the search :) the gravitational constant unit can very well be aliased (or even labeled) "gravitational constant unit". author  TomT0m / talk page 20:25, 31 August 2015 (UTC)
The unit(s) (m³ kg⁻¹ s⁻²) is not the value of the constant (6·67428 × 10⁻¹¹ m³ kg⁻¹ s⁻²). How do you, on the page of the gravitational constant, describe its value? --Izno (talk) 01:29, 1 September 2015 (UTC)
dimensional analysis
@Izno: its a different question, retrieving easily an item and describing it. Any item can be a unit. Considering this specific unit is expressed wrt. other more elemental unit, we have to find a way to express (m³ kg⁻¹ s⁻²) in Wikidata. We already have items for "m", "kg" and "s", we could create item for m³, kg⁻¹, and s⁻² which mean we will have to introduce a property to express the power of units. We have to introduce a (n-ary) property to express multiplication of units, something like
⟨ m³ kg⁻¹ s⁻² ⟩ multiplication of units Search ⟨ m³ ⟩
together with (P1706)   ⟨ kg⁻¹ ⟩
s⁻² Search ⟨ {{{7}}} ⟩
. For the power items, we would have
⟨ m³ ⟩ unit power of Search ⟨ m ⟩
power value Search ⟨ 3 ⟩
. For example. author  TomT0m / talk page 08:13, 1 September 2015 (UTC)
I would expect clicking "X" on the unit box to remove the unit. (That is, change it from "1 meter" to "1".) Instead, it only hides the box. --Yair rand (talk) 20:51, 31 August 2015 (UTC)
Hmm yeah that is a tricky one with this user interface. I actually forgot to add this to the list of known issues. It'll have to be split into two proper input fields instead of this overlay. This is just the quickest we could technically do. Once we have two proper input fields the issue you are having should be solved. We have phabricator:T109459 for that. --Lydia Pintscher (WMDE) (talk) 08:40, 1 September 2015 (UTC)
Impossible to enter °F and °C by their aliases. When I use the Q number instead, the label shown is incomprehensible ("GUbpRaNn" and "zCBkZFkr"). --Casper Tinan (talk) 21:16, 31 August 2015 (UTC)
I just tried it on the test system and it worked fine. I had to create the item for °C first though. --Lydia Pintscher (WMDE) (talk) 08:40, 1 September 2015 (UTC)
Excelent. Thanks a lot. It might be possible to limit the suggested unit values to (1) often used units or (2) subject of the property (P1629) and from here measured by (P1880, reverse of measured physical quantity P111). Poul G (talk) 09:26, 1 September 2015 (UTC)
@Lydia Pintscher (WMDE): So, Wednesday's come and gone, and units wasn't deployed. Did something go wrong? Is there a new schedule? --Yair rand (talk) 04:03, 3 September 2015 (UTC)
Sorry I realize now that my wording was ambiguous. I meant Wednesday September 9th. Sorry for the confusion. --Lydia Pintscher (WMDE) (talk) 06:21, 3 September 2015 (UTC)

mobile improvements going live soon

Hey folks :)

As part of his internship with us Bene has been working on making Wikidata more usable on mobile devices. We'll be redirecting users on mobile devices there soon just like it is done on Wikipedia. You can go and have a look at it here: http://m.wikidata.beta.wmflabs.org/wiki/Q15905

Tracking bug: https://phabricator.wikimedia.org/T78430

Remaining issues:

Cheers --Lydia Pintscher (WMDE) (talk) 16:35, 31 August 2015 (UTC)

The style changes are now live here. Redirecting people on mobile devices isn't happening yet. --Lydia Pintscher (WMDE) (talk) 06:23, 3 September 2015 (UTC)

Flow

Please see the announcement at Wikidata talk:Flow#Priorities for the Collaboration (Flow) team. --- Jura 11:51, 2 September 2015 (UTC)

This is extremely disappointing and a slap in the face of everyone who invested time in getting this enabled. The code is just going to rot, we should stop all conversions immediately and revert all pages that already have been changed. Deploying an unfinished product for which the support will crumble over the next couple of months will just be a source of more WMF frustration. And this is coming from someone who was starting to like it. Multichill (talk) 20:34, 2 September 2015 (UTC)

Editing Wikidata:Glossary

I'm ready to start rewriting Wikidata:Glossary as my comment on the talk page but the translation tags are messing editing up. I am going to copy it to User:Filceolaire/Draft:Glossary, strip out all the translation tags, rearrange and rewrite then copy it back over Wikidata:Glossary and call for retranslation and tagging - unless anyone has a better idea? Joe Filceolaire (talk) 21:28, 31 August 2015 (UTC)

Done. Well ready for review. I have completely rewritten the Glossary at User:Filceolaire/Draft:Glossary and kept less than 10% of the old definitions. Anyone who wants to comment please do before I overwrite the existing. Joe Filceolaire (talk) 21:13, 1 September 2015 (UTC)
Sure. But don't think for a second that it is remotely ok to paste all of that into Wikidata:Glossary without the translation tags.--Snaevar (talk) 23:13, 1 September 2015 (UTC)
Thanks Snaevar for adding tags. I've incorporated all the comments and this has added a few sections since then I'm afraid and I am not familiar with Translation tags to fix it. I would like to get it finished so I can overwrite the existing page. Can you help or suggest where I can find out how to fix this myself? Joe Filceolaire (talk) 21:29, 3 September 2015 (UTC)

Redundancy?

Is Q1627243 and Q209896 the same? If yes: please merge into one object. --2A02:810D:27C0:5CC:5C05:B3B8:B224:E4D5 16:24, 3 September 2015 (UTC)

  Done by Vlsergey apparently. —Galaktos (talk) 21:40, 3 September 2015 (UTC)

Why are links to wiki items displayed in a foreign script?

Tracked in Phabricator
Task T111346

I'm finding that on pages such as: Wikidata:Database reports/Constraint violations/Mandatory constraints/Violations, some entries are displayed in a script I don't recognise, possibly Cyrillic. For example: Q501 is displayed as Шарль Бодлер (Q501), but if I go onto the item, Charles Baudelaire is displayed at the top, as that is the label in English. Is there a way to stop this? I have the language set to English in my settings, but is there some setting I'm missing? Silverfish (talk) 20:18, 2 September 2015 (UTC)

Add babel boxes to your user page. See my page for example. Languages are based on your current location without them. Sjoerd de Bruin (talk) 20:44, 2 September 2015 (UTC)
I've added a babel box, and now nothing displays the name, just the Q identifier. And after a little while I've gone back to the same page, and it's the same as before. Note it's only on pages like that that this happens, on the Autolist 2 tool it just displays the English label, for example, and that seems to the usual way things work. And on that page only some of the items have that sort of link with the text in Cyrillic, some have the text in English, and some just the Q identifier (those within an English label, I assume). Silverfish (talk) 21:14, 2 September 2015 (UTC)
This seems to be a very strange issue. On special pages we automatically add the label when an item is linked but in Wikitext that shouldn't happen. Apparently, something like that happens nevertheless. After purging the page, the labels are gone. You see the labels in Russian because the HTML gets cached after saving in the language the last editor of the page has set in their preferences. -- Bene* talk 11:49, 3 September 2015 (UTC)
Check which namespaces/pages are actually cached. Jeblad (talk) 09:44, 4 September 2015 (UTC)

Is Mens College Basketball a 'sport'?

I was editing Labette Cardinals (Q17015538) - a bunch of sports teams attached to a college in Kansas - and I came across this problem:

We have sport (P641) to indicate what sport a team or individual plays but most competitions are restricted in other ways. You don't compete in tennis or even in tennis doubles (Q18123885) - you compete in Womens professional tennis doubles. What properties should we use to indicate these restrictions? Joe Filceolaire (talk) 14:51, 4 September 2015 (UTC)

@Filceolaire: It's a sports league (Q623109) (league (P118). I always wondered if a sport was actually assimilated with the rules you follow on a match for example, which would mean basketball = basketball match for example. author  TomT0m / talk page 14:58, 4 September 2015 (UTC)
Yes TomT0m, Labette Cardinals (Q17015538) is in the Kansas Jayhawk Community College Conference (Q16998117) league but they compete in 6 different sports, some men only, some women only, some both sexes. I think the 'Spirit squad' may even be mixed., I accept that we should create a separate item for each thing the Olympics give medals for but do we need to do the same for Kansas Jayhawk Community College Conference (Q16998117) or can we use qualifiers?
How do we say
  • if it is men/women/mixed;
  • if it is amateur or professional;
  • if it is local, regional, national or international;
  • if it is under 12s, under 18s, under 21s, veterans;
  • if it is one of those categories the paralympics have;
  • if it is 100m, 200m, 400m, 800m?
Is there an existing property/qualifier we could use? Do we need a new property? Do we need a new property for each of these? Joe Filceolaire (talk) 22:16, 4 September 2015 (UTC)
To me 800m is a sport by itself, who is a part of the more general athletics (Q542) one, for example: usually there is a 800m competition in a meeting. author  TomT0m / talk page 07:24, 5 September 2015 (UTC)

problems with Lee and LEE

please look at Lee (Q218125) and LEE (Q6457631). i noticed that at the moment it is (partly) impossible to add/change labels/descriptions of items which differ only in lower/upper case spelling. This was possible in the past.

for example please try out at Lee (Q218125) to change en-description "Wikipedia disambiguation page" to "Wikimedia disambiguation page". Then an error note appears telling that LEE (Q6457631) already has the same description (even if its label is spelled differently).

does someone know more about this, or can help? Holger1959 (talk) 11:30, 30 August 2015 (UTC)

It's a feature of the software that no two items can have the same Label and Description in one language. "Wikimedia disambiguation page (but see 'LEE' also)" and "Wikimedia disambiguation page (but see also 'Lee')" should work. Joe Filceolaire (talk) 16:17, 30 August 2015 (UTC)
Just curious if such "feature" could be disabled. It much simple to enter nothing into description at all than duplicating description. -- Vlsergey (talk) 16:38, 30 August 2015 (UTC)
@Filceolaire: thank you for reply! do you know who decided this and where the change was announced?
Problem: we have thousands (!) of different items like Abcd and ABCD, mainly because enwiki (but also some other Wikipedias) has different pages in many such cases (often disambiguations, but there are also "CamelCase" items vs. not "Camelcase" items, etc).
I mean, it would be easy for me to add "(but see also …)" to the English label, and "(aber siehe auch …)" to the German label, but what should i add for all the other possible languages if every label in every language must differ? this really looks like a bad change to me. Holger1959 (talk) 16:46, 30 August 2015 (UTC)
Holger: The developers put this feature into the code but you will have to ask Lydia Pintscher (WMDE) where (or if) that was announced (she leads the developers team). The problem mainly relates to Disambiguation pages - other types of page are unlikely to have the same description for both pages. If Lydia doesn't get back to you here you can ask on Wikidata:Contact_the_development_team or you can also register a bug/feature request on [1]. Hope that helps. Joe Filceolaire (talk) 17:50, 31 August 2015 (UTC)

@Lydia Pintscher (WMDE): can you say something about this issue? to see the "API Error" problem, please try eg. changing the Dutch/nl description of Lee (Q218125) to the standard string "Wikimedia-doorverwijspagina", or try changing the English/en description to "Wikimedia disambiguation page". Another example could be the 2 items for UOC (UOC (Q6478014)) und UoC (UoC (Q7897866)).

For editing such items after the software change, we have to either change their labels in all languages (eg. for English from "LEE" to "LEE (all capitals)", or for German to "LEE (in Großbuchstaben)", etc), or we have to change their descriptions in all languages (see Filceolaire's suggestion above). For both options i don't know what to add eg. in Spanish, or in many other languages i don't speak. For the "change labels option", this would also break the rule that labels generally should only contain the term itself (no additional strings). For the "change descriptions option", this would break the rule for disambiguations, that they should be uniform (all the same, not differing, at least in Latin Script languages), and this would also break the standardized set of disambiguation descriptions used by many bots, tools and users. Holger1959 (talk) 09:57, 4 September 2015 (UTC)

The label plus the description should be unique and identify an entity clearly. Difference only in case is not enough. We should give more information imho. --Lydia Pintscher (WMDE) (talk) 10:41, 4 September 2015 (UTC)
@Lydia Pintscher (WMDE): what do you mean with "We should give more information" – who is "we"? the developers (on how to handle such problems), or the community (on which information)?
i would be happy if you (or someone else) could tell or show me a working solution for the problem. i don't know the exact number, but i think enwiki alone has more than 1000 such "double" disambiguations where we at Wikidata have two items like "Lee/LEE" or "UoC/UOC".
of course one could also say "only one of them is allowed", but i don't know if we could really force enwiki (and other projects) to merge all these pages? Holger1959 (talk) 09:42, 5 September 2015 (UTC)

Find properties for infoboxes

How could we help users to find properties that are relevant to specific infoboxes? Ideally a project would list them, but I think we don't have that many projects for that to be possible. --- Jura 04:57, 5 September 2015 (UTC)

@Jura1: Assuming an infobox is mapped to a class, or some subclasses of a class, templates like {{PropertyForThisType}} and more importantly the corresponding property should help. author  TomT0m / talk page 07:21, 5 September 2015 (UTC)
Yes, properties for this type (P1963) on the item mentioned in template has topic (P1423) should do. Thanks. --- Jura 07:32, 5 September 2015 (UTC)
A dynamic approach might be to run related_properties.php on items with instance of (P31) = the item mentioned in template has topic (P1423). --- Jura 07:53, 5 September 2015 (UTC)

Open call for Individual Engagement Grants

Greetings! The Individual Engagement Grants program is accepting proposals from August 31st to September 29th to fund new tools, community-building processes, and other experimental ideas that enhance the work of Wikimedia volunteers. Whether you need a small or large amount of funds (up to $30,000 USD), Individual Engagement Grants can support you and your team’s project development time in addition to project expenses such as materials, travel, and rental space.

I JethroBT (WMF), 09:34, 5 September 2015 (UTC)

WQS optimisation for tree intersections ?

Hi. Does anyone know if there are optimisations possible or needed to enable tree query intersections on the Wikidata Query Service beta (ie the SPARQL beta) ?

The query I'm trying to run is to try to discover why The Guardian (Q11148) was being returned as a member of the class-tree of occurrence (Q1190554) in this Autolist query. So I'm trying to query to return all the items in that class-tree, between The Guardian (Q11148) and occurrence (Q1190554).

First I tried to use WDQ, with this query; but it only returns the single item The Guardian (Q11148) itself.

Having tried another similar WDQ query, this query to discover items in the corresponding class-tree for Hundred Years' War (Q12551), it seems WDQ may be returning only partial response-sets on queries like this, apparently when it reaches an item in the tree with more than one parent (details in this ping to Magnus.)

So I thought I'd try the SPARQL beta instead. After a couple of hours trying to get to grips with how to do SPARQL, I think this is a corresponding SPARQL query, at least in a basic form, [2] (seemingly having clicked the link you may then need to hit return by hand in the browser bar to load the query; then hit the 'execute' button below) -- but the query is timing out, presumably when it tries to create a full descending class-tree for occurrence (Q1190554).

Can anyone check whether my SPARQL is right, and/or suggest any optimisation that may help -- eg perhaps to hint to it to try to run the upward query from The Guardian (Q11148) first, and then an upward query from each of the items in that set, rather than a full descending query from occurrence (Q1190554) ? Alternatively, does anyone know what @Magnus Manske: did to get queries like this to work, and whether any similar heuristics could or are being planned for the SPARQL version ?

I appreciate that I'm a SPARQL newbie, so my query may be wrong. But this seem (at least at my newbie level of understanding) a reasonably straightforward example of the sort of query that we would seek to be able to run successfully and routinely.

Thanks in advance for any help anyone can offer. All best, Jheald (talk) 21:06, 5 September 2015 (UTC)

Maybe it's because The Guardian (Q11148) is instance of newspaper (Q11032) which is subclass of publication (Q732577) which is instance of process (Q10843872) which is subclass of process (Q3249551) which is subclass of occurrence (Q1190554)? Mushroom (talk) 21:27, 5 September 2015 (UTC)
@Mushroom: Thanks!! I'd missed the "instance of process (Q10843872)" when I was trying to follow the chain by hand. (Which may imply I'm going to need something more sophisticated than just TREE in the original WDQ query, eg perhaps this).
On the other hand, I still don't see why my subsequent WDQ and SPARQL queries didn't work to try to find the items in the chain -- looking at some of the Autolist results from the queries at Wikidata:WikiProject_UK_and_Ireland#Stats, there's a lot more than just The Guardian (Q11148) which seems anomalous, so anything to make the investigation process less manual is valuable. Jheald (talk) 21:44, 5 September 2015 (UTC)
@Mushroom, Jheald: This chain subclass/instance seems bogus. I usually insist that for something like this to work correctly, in the spirit of Help:Classification, we need to take care that a class as class as real world of object should always be a subclass of real world object classes, and that class of class of real world object should always be subclasses of class of class of real world object ... events or processes subclasses of events or process classes and so on. Can we stop a bit here and think about this in this particular case ? author  TomT0m / talk page 07:35, 6 September 2015 (UTC)
@TomT0m, Mushroom: I have had a bit more of a think here about the statement that seems to have caused this.
The problematic statement appears to be
publication (Q732577) ... is an instance of (P31) ... process (Q10843872)
whereas what I think may have been intended might have been something a bit more like
scientific style (Q7433830) ... is part of (P361) ... engineering (Q11023)
Making this change would leave publication (Q732577) entirely focussed on the notion of representing a publication, a physical object; rather than the process of publication, the (or an) act of publishing. This would solve the class-tree issue -- though I am a little worried that the classification might not be reflecting the actual full content and scope of the directly sitelinked articles.
I would suggest further discussion specifically on the classification of publication (Q732577) could be appropriately located at Wikidata:Classification_noticeboard#Publication, and leave this thread here for any discussion / hints / tips as to how to get best results for SPARQL queries. Best, Jheald (talk) 15:05, 6 September 2015 (UTC)

statistics

Hoi, Statistics on Wikidata seems to be in dire straits when I notice no updates for august.. What is going on, what can be done to improve things? Thanks, GerardM (talk) 06:36, 6 September 2015 (UTC)

Announcing the first StrepHit dataset for the primary sources tool

Hi everyone,


As Wikidatans, we all know how much data quality matters.
We all know what high quality stands for: statements need to be validated via references to external, non-wiki, trustworthy sources.

That's why the primary sources tool is being developed.
And that's why I am preparing the StrepHit IEG proposal.

StrepHit (pronounced "strep hit", means "Statement? repherence it!") is a Natural Language Processing pipeline that understands human language, extracts structured data from raw text and produces Wikidata statements with reference URLs.

As a demonstration to support the IEG proposal, you can find the FBK-strephit-soccer dataset uploaded to the primary sources tool backend. It's a small dataset serving the soccer domain use case.
Please follow the instructions on the project page to activate it and start playing with the data.

What is the biggest difference that sets StrepHit datasets apart from the currently uploaded ones?
At least one reference URL is always guaranteed for each statement.
This means that if StrepHit finds some new statement that was not there in Wikidata before, it will always propose its external references.
We do not want to manually reject all the new statements with no reference, right?

If you like the idea, please endorse the StrepHit IEG proposal!


Cheers,

--Hjfocs (talk) 14:47, 4 September 2015 (UTC)


The danger of blanket statements is that they are often easy to refute. No. Quality is not determined by sources. Sources do lie.
When you want quality, you seek sources where they matter most. It is not by going for "all" of them, it is where Wikidata differs from other sources.
Arguably and I do make that argument. Wikidata is so much underdeveloped in the statement department that having more data with a reasonable expectation of quality will trump quality for a much smaller dataset.
I do not understand at all why your tool must feed into the primary sources tool. Imho it is anathema to the instant gratification that is posting in Wikidata itself.
"Rejecting new statements with no reference"? That is plain insane. It is just another argument why I hate the presentation of your idea.
Thanks, GerardM (talk) 17:39, 4 September 2015 (UTC)


Thanks GerardM for your criticism, let me reply to your concerns.
  • "Quality is not determined by sources. Sources do lie. When you want quality, you seek sources where they matter most."
  • "Rejecting new statements with no reference"? That is plain insane."
The following references contrast the above points. I got inspired by them when developing the idea:
  • "It is not by going for "all" of them"
I completely agree with you that many sources can be flawed. I may have neglected the term "trustworthy" before "sources" and added it.
The project will also include an investigation phase to select a set of authoritative sources, see the first task in the proposal work package.
  • "I do not understand at all why your tool must feed into the primary sources tool."
As stated in the project page, the primary sources tool is made for data donations, and StrepHit would consist exactly of that.
Cheers,
--Hjfocs (talk) 12:13, 5 September 2015 (UTC)
Phabricator is very much a development environment. What goes on there is not familiar to many people like myself that are deeply involved in Wikidata. Certainly not when suggestions like compulsory references to statements are made. The sad thing is that the tool that you may be working on becomes in this way an aggressive statement, not into a welcome plan for a tool that may integrate with existing tools. Integration is what we need. Both quality and quantity is what we need badly.
In my blogpost I suggest that StrepHit will be awesome when it is more like Kian. It actively seeks integration both by playing nice with other tools and by seeking cooperation and not confrontation. There is plenty of room for StrepHit, it is a choice of its creators what it is they want it to be. Thanks, GerardM (talk) 06:14, 6 September 2015 (UTC)


  • "The sad thing is that the tool that you may be working on becomes in this way an aggressive statement"
The primary sources tool actually asks the user to approve or reject new references for an existing statement or new statements extracted from external sources (which always come with references in the StrepHit case).
Under this perspective, I think it is way less aggressive than directly adding statements into the knowledge base.
Cheers,
--Hjfocs (talk) 11:21, 7 September 2015 (UTC)

Wikidata weekly summary #174

 
Here's your quick overview of what has been happening around Wikidata over the last week.

NEW: Classification noticeboard

After the suggestion was made at the London science meet-up yesterday, I have been bold and created a new Wikidata:Classification noticeboard, in similar spirit to the existing Wikidata:Interwiki conflicts page for sitelink issues. On the new page I hope people will feel encouraged to bring apparent issues in the classification tree for discussion -- for example, instance of (P31) / subclass of (P279) chains that appear to lead to paradoxical results or nonsense search results; discussions about best choices of classification statements for particular items; and any other specific class-tree examples that just smell wrong.

It seemed to me that it would be useful to have a dedicated page that such issues could be brought to without swamping Project Chat. I know it's an area that myself I don't feel 100% confident in, so felt it would be good to have somewhere to bring things that don't smell quite right, for feedback and review if I think I can see how things might be fixed, or to be able to involve experts if I can't. There is certainly a lot that doesn't seem quite right at the moment -- the Autolist searches at Wikidata:WikiProject_UK_and_Ireland#Stats currently bring up a lot of anomalous results, which I hope to work through somewhat as a project over the next few months. But with luck many such of these anomalies will turn out to be related, so that starting to look at a few, and resolving issues in the tree of classes that cause those, may then make many of the other anomalies go away as results of the same cause. My hope is that with a dedicated page, it will act as a magnet for people to bring forward such anomalies, leading in turn to more of the most far-reaching underlying oddities being fixed and sooner.

Some initial threads I've opened there for discussion include:

... and I am sure over the next few weeks there will be many more.

One thing that could certainly be improved would be the machinery of the page -- eg an easy button to create a new thread, permalinks for threads, auto-archiving of threads recently marked resolved, etc. I don't know whether it would be straightforward to add such functionality, such as is typically found on noticeboard pages on larger wikis. One option might be to go to Flow, rather than to develop a specific dedicated infrastructure for the page; though I have to say the edit-box in Flow feels rather unstable compared to the trusty text-area input of an ordinary wiki page, plus of course the recent de-prioritisation of development work on Flow; so any move to go with Flow would seem to come with a fair number of pros and cons.

But anyway, the page is now up; so I hope people will approve of its creation, and bring issues to solve, and solutions to fix them.

What you want to do is to define an ontology. You have two ways to do it: the bottom-up or the top-down approach. Your page proposes a bottom-up approach starting by examples and classifying them by building a classification structure.
The best in my opinion is to start using a top-down approach with the two or three first levels at the top in order to have some objectives when starting to classify an element.
Have a look at en:Upper ontology before starting anything and before doing anything new perhaps can we find an existing ontology which can meet our requirements.
But we add to answer some questions before doing any choice:
  • what kind of granularity we want (what is the smallest concept we want to describe as instance)
  • the question of the inheritance
Instead of a Wikidata page, we need a project for that and as we are speaking about general knowledge we need a general project to organize the whole structure but specific projects have to develop domain classifications because nobody can have an overview about everything. Snipre (talk) 16:46, 6 September 2015 (UTC)
The situation is that we have a load of items with classifications that do not pass the smell test. That is a matter of quality of what we have, not of defining an ontology. When as a consequence of stuff that does not pass the smell tests changes are proposed and implemented it is NOT either bottom to top or top to bottom it is from somewhere within the mess outwards. If this leads to clarity, I am all for it.
You have been bold and it may be of a benefit. You mention a meeting in London but you do not report on it, all you do is present us with what was suggested.. I am interested in the lines of thought. Thanks, GerardM (talk) 17:39, 6 September 2015 (UTC)
Good ideal ! ({ {Ping|Snipre}} : before having something ideal in some sense, let's already try to have something that works ;) ) author  TomT0m / talk page 18:33, 6 September 2015 (UTC)
PS: related to this : Help:Modeling, I launched quite a while ago but did not really work. author  TomT0m / talk page 18:49, 6 September 2015 (UTC)
Hi @GerardM: not much to report, in truth. It was only a small hacking session on Saturday, following the earlier Wikipedia Science Conference on Wednesday and Thursday (which I did not get to), and hackathon near Cambridge on Friday (which I did not get to either).
There were only about half a dozen of us there on Saturday, but at one point, talking with User:Daniel Mietchen and User:NavinoEvans, the talk turned to Wikidata, and the importance of including the full subclass tree when running searches instance <instance of X>, but then somebody said how you had to be careful doing that because there were so many odd things in the class tree, eg the Guardian newspaper coming up as an instance of "event", and what could we do to try to clean it up; and I think that's when somebody suggested that creating a dedicated noticeboard might act as an organising focus and also encourage activity, just as we have Wikidata:Interwiki conflicts already, or all the various noticeboards on en-wiki. The conversation didn't really develop any more than that, just the suggestion being raised, but it seemed to me not a bad idea, so I thought why not give it a try. That's about it really. Jheald (talk) 22:23, 6 September 2015 (UTC)
Yeah, not much to report, except that people in the room had no difficulty coming up with examples where such a noticeboard might be useful. My notes from the meeting (on bibliographic metadata, not classification) are here. --Daniel Mietchen (talk) 23:08, 6 September 2015 (UTC)
@TomT0m: Nice draft page -- brings together a lot of useful stuff in one place. I've added it to the 'resources' at the top of the noticeboard page. Jheald (talk) 22:27, 6 September 2015 (UTC)
@Snipre: I take your point, to some extent, but we do now have 250,000 subclass of (P279) statements already in play, and very nearly 14 million instance of (P31) statements, so there's a lot we already have in place; but it would be nice to iron out at least some of the more noticeable oddities, to make the searches work better.
You're also right that it can sometimes make a lot of sense to clean up top-down -- for example, even at the first level, the class tree of occurrence (Q1190554) shows some pretty odd inclusions. [3]
You're also right that there is certainly scope for refinement, both at the top end, and as regards choices of granularity -- for example, for people we insist that every item for a human being be an instance of human (Q5), with all other defining information being carried by properties; whereas for example for films there is really quite an extensive sub-tree of classes that an instance might belong to. No doubt our ontology will continue to evolve, such decisions could certainly be revisited, and I am sure that as time goes on we shall compare and benchmark against as many other ontologies as we legally can. But for the moment, rather than throwing everything out and starting again from scratch, it does seem to me to make some sense to try to eliminate some of the odder anomalies from what we've already got, to at least make the search queries work a bit better. Jheald (talk) 22:45, 6 September 2015 (UTC)
@All, regarding Flow it seems I may have have stacked the picture a bit over-cautiously against it. Proclamations of its demise appear to be exaggerated. According to its original developer Jorm here, and this clarification message, it has not been abandoned, or even indefinitely put on ice -- dev suppport for it is not going to go away. But instead the focus from now is going to be how to better provide and support workflow features of typical project and noticeboard pages -- pages such as this very noticeboard -- rather than more general discussion-editing bells and whistles. So perhaps adopting Flow for the page should not be so quickly dismissed, but should still be seriously considered. Jheald (talk) 23:22, 6 September 2015 (UTC)
@Jheald I never said that we should start from the scratch. But I tink your noticeboard will treat individual problems and you risk to modify after some weeks a classification branch you already modified previously and by the way you will destroy what you tried to correct. You need at some point a overview of the structure. Snipre (talk) 19:57, 7 September 2015 (UTC)
Hi all, I've written up the notes here for the discussions at the Hackathon, but nothing extra on this topic that hasn't been brought up (note: Wikimedia UK has an issue with the site's security certificate so you will probably get a warning from your browser if you follow the link). Many thanks @Jheald: being bold and setting up the notice board! NavinoEvans (talk) 01:35, 7 September 2015 (UTC)

Entity, article, item

(I'm going to be using the WD:Glossary definitions of these terms here, even though they seem to have been a bit mixed up for some people (myself included on occasion) along the way. For clarification: Item means the real-world topic/subject, and entity is the thing on Wikidata. If I've gotten these mixed up right now, my apologies.)

The statements on a Wikidata entity represent claims regarding the item of that entity, not the entity itself. We don't put instance of (P31) Wikidata item (Q16222597) everywhere. Similarly, the statements are not regarding the Wikimedia articles linked to the entity. We don't (normally) put instance of (P31) Wikimedia article page (Q15138389) or its subclasses. There are exceptions to this, such as instances of Wikimedia list article (Q13406463). These pages tend to have unusually formed titles, that don't describe a topic, and instead describe the article page itself. For example, when one navigates to the page "List of sovereign states", the article doesn't go on to tell you all about an interesting list. The title describes the article page itself, not the topic, which is a decent way of telling that something's a bit different with this kind of article. Wikimedia article page (Q15138389), along with Wikimedia project page (Q14204246), Wikimedia main page (Q5296), and others, are subclasses of Wikimedia page outside the main knowledge tree (Q17379835). Wikidata item (Q16222597), strangely, is not, and is instead a subclass of Wikimedia internal item (Q17442446).

With that out of the way...

I think we really need to be clearer on the distinction between entities linked to articles about the items and entities linked to articles that are the items. Wikimedia page outside the main knowledge tree (Q17379835) and Wikimedia internal item (Q17442446) (which should be merged, imo) should not be subclasses of MediaWiki page (Q15474042) or its numerous superclasses. No subclasses of either of these should also be subclasses of items within the main knowledge tree. Entities that have actual topics should not have any sort of should not be subclasses of these. Some professor whose area of study was history of China (Q82972) was (hopefully) not studying the aspect of history (Q17524420). The notable event known as the death of Osama bin Laden (Q19085) that occurred on 2 May 2011 was not a certain aspects of a person's life (Q20127274). The Canadian detainees at Guantanamo Bay (Q5030746) are not a Wikipedia overview article (Q20136634). The city of Luanda (Q3897) is definitely not a Wikinews article (Q17633526).

I propose that all statements similar to these be removed, and the following pages be deleted:

--Yair rand (talk) 02:00, 25 August 2015 (UTC)

  Support this effort. author  TomT0m / talk page 08:15, 25 August 2015 (UTC)
Glossary
"Item means the real-world topic/subject, and entity is the thing on Wikidata" No, item is a special type of entities (properties are entities). An item is equally the real world concept and the stuff on Wikidata it maps to :) The glossary is not the easiest piece to write sometimes with successive rewrinting nonsenses emerges (next time I read it the (way secondary reification concept got totally out of control, don't know what happened.)author  TomT0m / talk page 08:11, 25 August 2015 (UTC)
Ah, got it. I don't suppose we actually have an unambiguous term specifically for the real-world topic/subject? --Yair rand (talk) 08:56, 25 August 2015 (UTC)
      • 'Wikidata Entity' = 'Wikidata item (with Q number)' OR 'wikidata Property (with P number)' OR future Wikidata query (with some other reference)'.
      • A Wikidata item is about a single topic and has statements about that topic. A ststatement has a property and a value. The value can be an item or a wikimedia Commons file or a web URL or a string or monolingual text (with specified language) or a date or a number or a number with specified unit. A statement may have qualifiers and references each with a property and a value.
      • A wikidata item can be an instance or a class. An instance is a single individual and has a statement using the instance of (P31) property to link it to the class it is a member of. A class is a group of instances and has a statement using the subclass of (P279) property (or the parent taxon (P171) property for a biological taxon) linking it to a larger class. Some items are both a class and an instance of a particular type (or class) of class.
      • For Encyclopedias and biographical dictionaries on wikisource the topic that the item is about (or is an instance of (P31)) will be an 'encyclopedic article' or a 'biography'.
      • A wikidata item has sitelinks to pages on wikimedia wikis which are about the same topic. The wikimedia wikis use these sitelinks to add interlanguage links between wikis in diffferent languages. Each wikidata item can have only one sitelink to a wiki. This is so that other wikis have only one language link to that wiki.
      • Because each wikidata item has only one sitelink to each wiki there can be a problem when a wiki has more than one page about a topic - separate wikidata items have to be created for each of these pages. Wiki Category pages, for instance, are on a different wikidata item from the item for the topic of that Category.
      • Some wikimedia wiki pages describe a topic which is internal to wikimedia and not related to a real world topic, such as Template pages, disambiguation pages, Help pages etc. These pages also need language links so they need wikidata items with sitelinks. These items have statements identifying them as "instance of (P31):MediaWiki page (Q15474042)" or various subclasses of MediaWiki page (Q15474042).
Hope that helps. That Glossary really needs a rewrite. Joe Filceolaire (talk) 11:43, 25 August 2015 (UTC)


"political career of Arnold Schwarzenegger"

  • Independently of the re-write of the glossary, what value for instance of (P31) do you suggest for "political career of Arnold Schwarzenegger" (Q3660742). Personally, I think "partial biography article" (Q20127274) is suitable. Obviously we can discuss if "article" should be in the item's label or not. @YMS:. --- Jura 08:47, 25 August 2015 (UTC)
I'm okay with anything that is a consistent system. Using items like Q20127274 was an easy solution for articles which are otherwise hard to classify. What would you do with first inauguration of Barack Obama (Q25094)? With religious views of Adolf Hitler (Q392636)? With Death and funeral of Pedro II of Brazil (Q5420282)? Or with the cited death of Osama bin Laden (Q19085), which actually is an interwiki conflict, with some articles describing a military operation, some the death of a person? What with Napoleon and the Jews (Q2916488), Abraham Lincoln in the Black Hawk War (Q4669055), later life of Isaac Newton (Q6495541)? All those are items for Wikipedia articles just describing some arbitrary aspects of some larger topic, and to find a suitable "is a" will be a guesswork every time. --YMS (talk) 14:24, 25 August 2015 (UTC)
They are all instances of events - or sequences or sets of events, aka. story ? , for sure. A death is an instance of death, an investiture is an instance of investiture ... Put the most precise you can think of. As the classes are hierarchically ordered (an investiture is a political event which is a US political event, ...), querying instances of a general one will return also the more specific ones. author  TomT0m / talk page 14:35, 25 August 2015 (UTC)
I have some difficulties seeing "Napoleon and the Jews" or "Abraham Lincoln in the Black Hawk War" as events. In fact, I have some difficulties seeing those as real-life entities at all, for me they are just Wikipedia articles. If that's my personal problem, I won't dare to block any progress here. But it seems like may others seem to have similar problems to find the right category to use, as most of the items of these kinds used to be classified as humans basically since the beginning of Wikidata. --YMS (talk) 16:05, 25 August 2015 (UTC)
It depicts and analyses a class of events in the case of Abraham Lincoln, all the events of the Black Hawk War involving Abraham Lincoln, or in which Abraham Lincoln influences. Otherwise it's an analysis of what he thought or how he acted in the Black Hawk War ... What does "Napoleon and the Jews" certainly describes a set of interaction (direct or indirect, through law voting) with the Jews ? Otherwise it describes a set of decisions Napoleon took on the Jews topic, I don't know. Interesting topic though :) Apart from that, these articles have definitely a defined topic. author  TomT0m / talk page 16:16, 25 August 2015 (UTC)
Hm, this is a rather complicated issue. Maybe "Napoleon and the Jews" is a relationship of some sort? --Yair rand (talk) 01:41, 26 August 2015 (UTC)
I would go with <instance of:political career> or maybe <instance of:career(applies to part:politics)> and <main subject (P921):Arnold Schwarzenegger (Q2685)>. I think <field of this profession:politics>is wrong because a "career" is not a profession. I think <main subject (P921):Arnold Schwarzenegger (Q2685)> is better than <of:Arnold Schwarzenegger (Q2685)> because it can be an independent statement and not just a qualifier. Joe Filceolaire (talk) 14:12, 25 August 2015 (UTC)
I'd go with "political career" over simple "career" with an extra qualifier, but that's not such an important point. It definitely should not use main subject (P921). The primary point I've been trying to make is that these items should not be treated as works/articles. They don't have languages, publication dates, subjects, authors, etc. The distinction between the topic and the article itself is important. Arnold Schwarzenegger's career in politics did not have a main subject. A biographical article may have a subject, but that is not what the item is about. --Yair rand (talk) 01:41, 26 August 2015 (UTC)

"History of"

There are some 2916 item listed as instance of (P31) aspect of history (Q17524420). These are definitely linked to articles about a topic, so they shouldn't be treated as Wikimedia articles themselves. I'm unsure about how they should be treated, though. Is "history of France" a part of "history of Europe"? Maybe a subclass of it? Neither? Is it an instance of "history"? "Time period"? 121 items are listed as subclass of (P279) history of a country or state (Q17544377), and using it would mean that subclass of (P279) couldn't be used for lower-level history articles. 34 items have instance of (P31) history (Q309), and 143 items have subclass of (P279) history (Q309) statements. (Many of the former to actually be museums, strangely.) Thoughts? --Yair rand (talk) 01:53, 1 September 2015 (UTC)

@Yair rand: History of france is a part of history of europe. History of france and history of europe are subclasses of history, in general. History of france is a subclass of "history of a country", and maybe "history of a European country" to refine further. author  TomT0m / talk page 08:00, 1 September 2015 (UTC)
Would history of the United States from 1980 to 1991 (Q2633338) be part of history of the United States (Q131110)? Would 1981 in the United States (Q2812799) be part of that? Or maybe both of those would use subclass? Would history of the Poles in the United States (Q16843880) be part of history of the United States (Q131110)? Also, would any of these have any instance of (P31) statements? --Yair rand (talk) 22:17, 1 September 2015 (UTC)
  • Q17524420 has just been relabeled from "Wikimedia history article" to "aspect of history" by User:Michiel1972, with the description changed to "article with an aspect of the history of something". If the item is still to be used, the new label is very much an improvement, but the description still has the problem of treating items as Wikipedia articles. I'm also unclear on where this item fits into the general workings of history items. @Michiel1972: Could you comment on this? --Yair rand (talk) 01:01, 4 September 2015 (UTC)
I do agree with the discussion above, these items are not included in wikidata because they are 'wikipedia articles' but because they are describing something (a detail, an aspect, a year, a part) of history of something. So I changed the label so it better fits the P31 usage, as I was mainly inspired by the change of label of certain aspects of a person's life (Q20127274) . Michiel1972 (talk) 11:50, 8 September 2015 (UTC)

edit summaries; and other language labels

Could somebody quickly remind me:

  • how to add an edit summary to an edit ? &
  • how to see (and edit) the labels and descriptions in all languages for a particular item, not just my selected favourite languages?

Thanks in advance, Jheald (talk) 10:32, 8 September 2015 (UTC)

So apparently I can only add an edit summary when undoing or reverting an edit (Help:FAQ#Editing). Which is unfortunate; but I suppose I can make a very minor edit, then revert it, if there's a block of edits I want to explain... Jheald (talk) 11:03, 8 September 2015 (UTC)
@Jheald:
T47224 (Story) Allow detailed summaries in GUI
T92759 allow editing of more languages in the in other languages box than the ones defined via babel boxes Open
T61905 It should be possible to add a label in any eligible language Closed, Resolved
I have no idea why manual edit summaries are not allowed at wikidata. It' s bad. --Atlasowa (talk) 11:29, 8 September 2015 (UTC)
Custom edit summaries are not possible yet (without hacks) because Wikidata is a multilingual project and we try to make as much information available to the user in their language as possible. This includes the edit summary. --Lydia Pintscher (WMDE) (talk) 16:15, 8 September 2015 (UTC)

Next office hour

Hey everyone :)

We'll be doing the next Wikidata office hour on September 23rd at 17:00 UTC. See here for your timezone. We'll be meeting in #wikimedia-office on Freenode IRC. As usual I'll start with an overview of what's been happening around Wikidata since the last office hour and then we'll have time for questions and discussions. If there is a particular topic you'd like to have on the agenda please let me know. Hope to see many of you there.

Cheers --Lydia Pintscher (WMDE) (talk) 10:34, 8 September 2015 (UTC)

be_x_oldwiki

Hi everybody. I tried to merge two items Q6511312 with Q4300924. However because of moving be-x-old.wikipedia.org domain to be-tarask.wikipedia.org something broke. It writes now The external client site 'be_x_oldwiki' did not provide page information for page 'Малоткавічы' Can anyone help me? --Jarash (talk) 17:31, 8 September 2015 (UTC)

They also created ticket in wiki bug tracking system. --Jarash (talk) 17:33, 8 September 2015 (UTC)

/doc in Arabic Wikipedia entry

Hey, some of the links to Arabic Wikipedia in Wikidata use the documentation template. For example Q6717769 has a link the documentation template in Arabic Wikipedia instead of the main template. Do you have a way to get a list or fix it by bot ? We need to delete all (/شرح which means /doc) in Arabic Wikipedia intery. The link قالب:صندوق جبن/شرح will be then قالب:صندوق جبن. --Helmoony (talk) 01:56, 7 September 2015 (UTC)

@Helmoony: This may be a useful list. --Daniel Mietchen (talk) 02:24, 7 September 2015 (UTC)
@Daniel Mietchen Thank you. I've made some changes. Just one question, how can you explain that the example I mentionned above Q6717769 doesn't appear on google results ? --Helmoony (talk) 03:21, 7 September 2015 (UTC)
@Helmoony: I have created a database query for you. There is also a regularly updated list which seems not to support Arabic suffixes. Matěj Suchánek (talk) 15:25, 7 September 2015 (UTC)
@Matěj Suchánek That's extraordinary usefull. Thank you a lot. --Helmoony (talk) 22:14, 8 September 2015 (UTC)

Searching for items with no reference on specific properties ?

Hello,

I'd like to search systematically items with unsourced statements on specific properties (date of birth (P569) and date of death (P570) for a beginning).

Is there a way to build a WDQ query checking if the statement is sourced or not ? and if not, is there another tool that could please help me ? --Hsarrazin (talk) 17:25, 7 September 2015 (UTC)

You should be able to do this with http://wdqs-beta.wmflabs.org/ . It uses SPARQL (Q54871) but not a lot of us are good with that. Multichill (talk) 19:47, 7 September 2015 (UTC)
ok, this is like Chinese to me — certainly very clear for a dev', but completely unreadable to ordinary contributors…  :D
is there someone who could help me build a query, please ?
the idea would be something like date of birth (P569)2004 (Q2014) and unsourced - ideal would be and linked to xxwiki. Thanks for help. --Hsarrazin (talk) 23:06, 7 September 2015 (UTC)
@Hsarrazin: you can try something like this: https://phabricator.wikimedia.org/P1993 (note the limit - there are a lot of unsourced statements so without limit it will probably time out). If you also want to check for sitelinks then it's like this: https://phabricator.wikimedia.org/P1994 . Note it checks for language, if you want specific wiki then so far the only way is to filter URLs with SPARQL string functions.  – The preceding unsigned comment was added by Smalyshev (WMF) (talk • contribs).
Combined with a date range filter, that would be this. --Yair rand (talk) 07:56, 8 September 2015 (UTC)
Thanks a lot, this works fine, and the limit number for answers is ok, since I intend to complete those :)
one small question though, I had results that already had imported from Wikimedia project (P143) or reference URL (P854) sourcing. How can I filter those with no reference at all, or those with existing imported from Wikimedia project (P143) or reference URL (P854), but no stated in (P248) ? --Hsarrazin (talk) 18:39, 9 September 2015 (UTC)

WQS: Searching for items with reference to "Le Figaro"

From wikidata mailinglist: Searching for "stated in" "Le Figaro" (P:P248, Q216047):

See also: mw:Wikibase/Indexing/SPARQL_Query_Examples. --Atlasowa (talk) 09:44, 8 September 2015 (UTC)

Can someone improve the query? better use count(?statement) rather than count(?ref), right? and post the WQS link? --Atlasowa (talk) 09:56, 8 September 2015 (UTC)
@Atlasowa: Interestingly, changing count(?statement) to count(?ref) makes no difference -- count() here is essentially counting the number of identifiable cases where the subsequent algebra is true (ie the number of lines you would get from a SELECT query), regardless of whether it is ?ref or ?statement in that case which is being selected and then counted. So, even though it may not be what one might image at first sight, both are in fact equivalent to asking for just (count(*) as ?count), like this, without specifying either ?ref or ?statement.
To count the number of individual statements or refs, use DISTINCT -- giving 337 distinct statements, to 267 distinct references. Jheald (talk) 10:19, 8 September 2015 (UTC)
Thank you, Jheald! Can i ask some more?
1) What is the difference between 337 distinct statements and 267 distinct references? Does it mean that wikidata has 267 items with LeFigaro references and 337 distinct statements with LeFigaro references (= some of the 267 items have several statements with LeFigaro references)? Is this interpretation correct?
2) Can i get a list of the statements with reference to Le Figaro (or a list of items or both)?
3) What about references to Le Figaro that do not use "stated in" (P:P248) but "reference URL" (P:P854, "should be used for internet URLs as references") like http://www.lefigaro.fr/flash-actu/2015/09/08/97001-20150908FILWWW00125-abe-reelu-a-la-tete-de-son-parti.php ? (See Help:Sources#Web_page) Is it possible to search for those references too?
Sorry for piling on more questions ;-) --Atlasowa (talk) 11:02, 8 September 2015 (UTC)
@Atlasowa: I am only a newbie at this, but I think one can discover the subject, the property, and the object of the statement with a query like this: http:// tinyurl.com/nmkee2z (Can't link directly to the URL generated by the 'short URL' button on the query page, because the URL-shortener is blocked by the wiki software). [direct link]
There may well be neater ways to do this; and I haven't checked to see if it copes with statements with multiple values; but it's at least a start. Jheald (talk) 12:56, 8 September 2015 (UTC)
An example of where the same reference is used to support more than one statement occurs in the place of burial (P119) of Honoré Daumier (Q187506), as can be seen if we lock the '?subject' variable to Q187506 as in this query: tinyurl.com/p87rsd5. Jheald (talk) 13:29, 8 September 2015 (UTC)
Here's the query tweaked to find statements that occur more than once: tinyurl.com/qcgwtty; investigating further, this occurs where the same statement is supported by more than one reference that fits the criteria, eg place of burial (P119) for Benoît Malon (Q2015151) as per this query, tinyurl.com/nmbblwp. Jheald (talk) 14:14, 8 September 2015 (UTC)
Awesome, Jheald! Here's the [direct link], many thanks! --Atlasowa (talk) 15:18, 8 September 2015 (UTC)
@Atlasowa: I think that this: tinyurl.com/pxlrkd7 is how one would do a search for a reference link including 'lefigaro.fr'. However, as you can see, the search is timing out. I think this is because the search service will be trying to build a list of all URLs used by reference URL (P854) before scanning it for lefigaro; but this first stage is taking too long.
Interestingly, the LIMIT 10 directive, telling it to stop as soon as it's found ten, doesn't seem to be helping -- with a conventional SQL search this can often at least let you have a peek (and confirm your query is correct). I also tried searching for '^http://www.lefigaro.fr' -- ie this exact phrase specifically at the start of the URL, which should be slightly quicker to filter by; but probably not nearly enough to make the query fast enough. Even just trying to count the first ten also fails. tinyurl.com/q7y77bm Jheald (talk) 15:41, 8 September 2015 (UTC)
Thank you very much for your investigation, Jheald! There is Special:LinkSearch, which suggests that wikidata should find some links: https://www.wikidata.org/w/index.php?title=Special:LinkSearch&limit=100&offset=0&target=http%3A%2F%2Fwww.lefigaro.fr For example date of death (P:P570) of Q3313825. --Atlasowa (talk) 08:49, 9 September 2015 (UTC)

What is the purpose of 2747 (Q19242825) and such?

I see the there is a lot of "number" entities in Wikidata like this: 2747 (Q19242825) - basically for every number. What is the purpose of those? Are we going to store all natural numbers in Wikidata one by one? That seems kind of pointless. They are created by User:PoulpyBot though the bot's description pages have no mention of why it is done. Anybody has explanation about this? --Laboramus (talk) 00:03, 9 September 2015 (UTC)

Quite a lot of numbers are interesting/notable, so it would seem to be convenient to have the complete set up to a certain limit and then additional ones as necessary. How far to go? Well 1000 might have been a good place to stop; it seems that 10000 was chosen by User:Poulpy. Not sure it matters a great deal. MSGJ (talk) 14:07, 9 September 2015 (UTC)

more data for more sister projects coming right up

Hey folks :)

It is time to give some more sister projects access to the sitelink or data part of Wikidata. The schedule is as follows:

  • September 22nd: Wikibooks will get access to statements (aka Phase 2)
  • October 20th: Meta, Mediawiki and Wikispecies will get access to the sitelinks (aka Phase 1)

Please give them a warm welcome and watch Wikidata:Wikibooks, Wikidata:Meta-Wiki and Wikidata:Wikispecies for questions.

Cheers --Lydia Pintscher (WMDE) (talk) 16:12, 9 September 2015 (UTC)

URL shortener?

Since tinyurl.com is banned here, is there any other solution for URL shortening that is OK here? I understand the potential problems with spam, etc. but this seems to be a convenient way to post URLs for e.g. long SPARQL queries without having to actually paste in the whole query (which may also conflict with wiki formatting and cause other inconveniences). Any suggestions? --Smalyshev (WMF) (talk) 20:25, 9 September 2015 (UTC)

Smalyshev (WMF) I think all URL shorteners are banned because of the cost to community time when abused. I see no way to URL shortening without someone making a very strong case for some exception. Blue Rasberry (talk) 20:33, 9 September 2015 (UTC)
@Smalyshev (WMF): Why not create our own URL shortener, limited to URLs on WMF domains? Jheald (talk) 20:41, 9 September 2015 (UTC)
Would love a wiki.data one. ;) Sjoerd de Bruin (talk) 20:43, 9 September 2015 (UTC)
In beta test: http://urlshortener.wmflabs.org/wiki/Main_Page i.e. us.wmflabs.org/u Raymond (talk) 21:42, 9 September 2015 (UTC)
Let's see: https://us.wmflabs.org/v --Smalyshev (WMF) (talk) 22:00, 9 September 2015 (UTC)

Fusion problems

Can someone fusion en:Template:IsraelForeignMin (Q10630122) with German de:Vorlage:Navigationsleiste Außenminister Israels ? 2567Mara3425 (talk) 20:28, 9 September 2015 (UTC)

  Merged. Jared Preston (talk) 20:36, 9 September 2015 (UTC)

Can someone also fusion en:Template:IsraelDefenseMin (Q10629703) with German de:Vorlage:Navigationsleiste Verteidigungsminister Israels  ? 2567Mara3425 (talk) 21:10, 9 September 2015 (UTC)

  Done. Jared Preston (talk) 05:04, 10 September 2015 (UTC)

Fetching an item alias

I am wondering how/if I can fetch an items aliases into a template. I am currently using |{{#invoke:Wikidata to pull items properties, but not sure of the syntax needed to get data out of the description such as aliases. Thanks in advance for any useful guidance. Julialturner (talk) 22:23, 9 September 2015 (UTC)

@Julialturner: Look at the module at cswiki (getAliases()). Matěj Suchánek (talk) 14:12, 10 September 2015 (UTC)

metre (Q13175867) is metre (Q11573)

Is there any reason why metre (Q13175867) exists as separate entity (with said to be the same as (P460)) and not just merged into metre (Q11573)? From the only site link it seems to be the same thing... --Laboramus (talk) 23:50, 9 September 2015 (UTC)

There is a sitelink to kuwiki on both of them. Popcorndude (talk) 00:29, 10 September 2015 (UTC)
@Laboramus:  Done Elements have been combined.Thnk you --ديفيد عادل وهبة خليل 2 (talk) 11:23, 10 September 2015 (UTC)

Vote for Wikidata at "Deutschland - Land der Ideen"

Wikidata ist in der Endrunde des Wettbewerbs "Deutschland - Land der Ideen". Hier kann man bis zum 15. September täglich abstimmen und den Gesamtsieger küren: http://hauptvoting.welt.de/mainvoting/list. --Stefan Weil (talk) 04:56, 10 September 2015 (UTC)

Wikidata is in the final round of competition "Deutschland - Land der Ideen". Here you can vote every day until 15 September, and nominate the overall winner: List to vote. Please support us --Crazy1880 (talk) 17:52, 10 September 2015 (UTC)

Big mistake?

OK, if we're not talking about the facts, that I haven't added birth dates and full death dates, did I done something else stupid/wrong by adding date of death (P570) in such format: +00000002008-00-00T00:00:00Z? Just asking, because when I preview Wikidata module in Wikipedia page (w:lv:Jānis Ivars Stradiņš) I get wrong death date: {{#invoke:Wikidata|getDateValue|P570|FETCH_WIKIDATA|y}}. Of course, if I did something stupid, I can revert my actions if there isn't easier way to do that. --Edgars2007 (talk) 19:43, 10 September 2015 (UTC)

I think there is something wrong with w:lv:Module:Wikidata, because {{#property:p570}} returns the correct value. MSGJ (talk) 20:36, 10 September 2015 (UTC)
This sounds related to phab:T103378. —Galaktos (talk) 22:18, 10 September 2015 (UTC)
This sounds the same, thanks Galaktos! Well, I'm glad, that I didn't screw up anything :) For my current needs, it will be OK to have incorrect year, because currently I just want to check, who is dead in Wikidata, but alive in Latvian Wikipedia. Thanks! Edgars2007 (talk) 07:31, 11 September 2015 (UTC)

Unmarried versus missing spouse data

Is there a way to distinguish between someone who is unmarried versus missing spouse data? --Richard Arthur Norton (1958- ) (talk) 04:52, 11 September 2015 (UTC)

@Richard Arthur Norton (1958- ): One can set spouse (P26) to the special value no value. Currently the query tinyurl.com/po9nxyv reveals 125 incidences where that has been done. Here is also the equivalent search on WDQ, Autolist, which uses a special fake item number to represent no value.
On the edit screen for an item, start to add the property, then when the input box comes up, hit the icon immediately to the left of the input box with three little horizontal bars. This will reveal the special option "no value", allowing you to make an edit like this: (diff). Jheald (talk) 06:29, 11 September 2015 (UTC)
How can I filter "no value" to just get the ones without any value? --- Jura 08:39, 11 September 2015 (UTC)

Internationalization for IPs

The Universal Languag Selector is not really suited to non-logged in users. In Commons, there is a script that appears to do the job all right, commons:MediaWiki:AnonymousI18N.js, can we import it to Wikidata ? (@HB: who raised the topic in frwiki). --Zolo (talk) 09:51, 11 September 2015 (UTC)

@Zolo: It used to be here... Matěj Suchánek (talk) 12:44, 11 September 2015 (UTC)

Q20901608

This item may be deleted as it is empty. --Hindustanilanguage (talk) 10:53, 11 September 2015 (UTC)

We have WD:RfD for those type of requests. Mbch331 (talk) 12:07, 11 September 2015 (UTC)
  Done for this time, but next time, please use the link in my previous comment. Mbch331 (talk) 12:08, 11 September 2015 (UTC)

Redundancy?

Is transmission (Q16259746) and line shaft (Q192696) the same? --2A02:810D:27C0:5CC:F1F0:D92B:87DF:76BD 17:11, 11 September 2015 (UTC)

It could be, but both items contain a link to kkwiki and srwiki. Sjoerd de Bruin (talk) 17:24, 11 September 2015 (UTC)
+ also Q6497460 (the same name on the article on enwiki as the simplewiki at Q16259746). --Stryn (talk) 17:29, 11 September 2015 (UTC)
Moved en:Transmission (mechanics) from Q6497460 to Q16259746. (Q6497460 should link to en:Gearbox but it's only a redirect) --2A02:810D:27C0:5CC:F1F0:D92B:87DF:76BD 17:53, 11 September 2015 (UTC)
Fixed Serbian as well (cf my contributions) --2A02:810D:27C0:5CC:F1F0:D92B:87DF:76BD 17:58, 11 September 2015 (UTC)
OK, it isn't redundant. Step by step I get it. Moved German article "Transmission" to "Line shaft". But there are more articles that are not sorted in correctly. --2A02:810D:27C0:5CC:F1F0:D92B:87DF:76BD 18:09, 11 September 2015 (UTC)
Moved around other languages. Looks better now (probably still not 100 % perfect) --2A02:810D:27C0:5CC:F1F0:D92B:87DF:76BD 18:21, 11 September 2015 (UTC)

School of education

I am very confused with what's going on with "school of education". We seem to have two entries with that English label intended to mean the generic term for a college that trains teachers - school of education and professor seminary. The first lists just 'de' and 'en' wikipedia links, both of which look like they describe what I would expect. The second also has a 'de' link to a different page - "Lehrerseminar" which seems to be a historical entity rather than the current term for a German teacher's college; and it has many other language links as well (but not 'en'). Meanwhile there seem to be some other inconsistent language links on the various language-specific pages, for example I would have expected the page 'École supérieure de pédagogie' in French wikipedia to be linked as it is the French label for school of education, but that is instead linked to Q17000323 which is indicated as a disambiguation page... Any suggestions how to fix this all up? I think school of education is right and should just be linked to more pages in different languages somehow? ArthurPSmith (talk) 18:17, 11 September 2015 (UTC)

Help:FAQ#Other Wikimedia sites

In Help:FAQ#Other Wikimedia sites, #1, there is actually no answer given. It would be really helpfull, if someone that knows the whole development of Wikidata, offer some date, when the process of integration is expected to end, for example midst of 2017 or in about 3 or 4 years... Many linking work would be spared in local wikis knowing that. --Janezdrilc (talk) 01:58, 8 September 2015 (UTC)

There is currently no progress towards integration with Wikispecies or most Wiktionaries, so no date can be offered. Wikispecies has no interest in integration, and the challenges of Wiktionary integration have hit major snags, in part because the data structure of the Wiktionaries does not match that used by Wikidata. --EncycloPetey (talk) 03:26, 8 September 2015 (UTC)

Thank you for this information. Just one more question: Is there any plan to introduce existing sister project links into all local wikis (like this way or maybe this way) and when about is that going to take place? --Janezdrilc (talk) 11:52, 8 September 2015 (UTC)

Yes that is planned. The ticket for it is phabricator:T103102. There is no timeline for it yet. --Lydia Pintscher (WMDE) (talk) 16:14, 8 September 2015 (UTC)
There are also plans for the wikidata software to be used on Commons to encode the metadata for each file and internationalise it. This is on hold for the moment but is (as I understand it) a fairly high priority. Joe Filceolaire (talk) 13:12, 12 September 2015 (UTC)

Errors VIAF/Wikidata

I tried to ask to the Italian Bar, but no answer. So the problem is the links to VIAF to some Italian buildings are totally wrong, and I don't know how to correct this, so I am leaving this note.

  • Palazzo Alberti Q16163465 on Wikidiata is a palace in Florence, on Viaf in Sansepolcro
  • Palazzo Bardi-Serzelli Q16585968, on Viaf is actually palazzo Busini Bardi Q3889596
  • Palazzo Bargagli Q16585972, on Wikidata is a palace in Florence, on Viaf in Siena (sarebbe Q3889601)
  • Palazzo Mancini Q16586227, on wikidata is a palace in Florence, on Viaf in Roma (sarebbe Q3360896)
  • Palazzo Strigelli Q16586343, on wikidata is a palace in Florence, on Viaf in Palazzuolo sul Senio
  • Palazzo Bonaparte Q19721646, on wikidata is a palace in Florence, on Viaf in Rome (sarebbe Q3889693)
  • Via Garibaldi Q16620946, on Wikidata is a street in Siena, on Viaf in Genoa (sarebbe Q2343106)
  • Collegio degli Scolopi Q19984666, on Wikidata is a palace in Florence, on Viaf in Pieve di Cento e Urbino
  • Museo dell'Opera Metropolitana del Duomo Q869130, on wikidata in a museum in Siena, on Viaf in Perugia (sarebbe Q3868002)

--Sailko (talk) 21:23, 14 August 2015 (UTC)

So they are not well identified. The obvious solution remove the VIAF identification for now. GerardM (talk) 06:42, 12 September 2015 (UTC)
There are a lot of wrong identifications at the side of VIAF. I still don't know why a bot imported them all without checking. Sjoerd de Bruin (talk) 12:11, 12 September 2015 (UTC)
I am glad they were imported. Yes, there may be errors, but now they are checked and otherwise nobody would bother. Thanks, GerardM (talk) 12:41, 12 September 2015 (UTC)

Contradiction

Hello.Is there a contradiction between this item and this Wikipedia template?This template prevents interwiki links from appearing.Because it determines the appearance of a particular copies --ديفيد عادل وهبة خليل 2 (talk) 10:54, 12 September 2015 (UTC)

In which sitelink(s)? --- Jura 12:03, 12 September 2015 (UTC)
I just note your addition to your comment. It's likely that this is something that needs to be fixed at Wikipedia, not Wikidata. --- Jura 16:06, 12 September 2015 (UTC)

locale for language versions

Originated from this set of fixes for the Russian version of the Main page. Because of locale set globally to en-US (I assume), native language specific formatting tools are useless. So for number formatting or plural forms I had to write independent modules. Say instead of just {{NUMBEROFARTICLES}} having {{#invoke:String|replace|{{NUMBEROFARTICLES}}|,| }}. It would be more convenient to set locale to ru-RU for the page, or to have an optional argument overriding default locale like {{NUMBEROFARTICLES|locale=ru-RU}} I am not sure if this is within the scope of this forum yet to mention the problem. --NeoLexx (talk) 13:47, 12 September 2015 (UTC)

How do I disable Flow?

It's so fucking slow. And why does it not allow me to just type {{Q}} or {{P}}?--Kopiersperre (talk) 17:10, 12 September 2015 (UTC)

There is a button to switch between wikitext and visual mode. It can't be disabled, because it's enabled per page. Sjoerd de Bruin (talk) 17:12, 12 September 2015 (UTC)

About stalagmite (Q181312) and stalactite (Q177197)

Distribution of items (2015-09-07)
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210 (QID)
  •   items without claims/statements
  •   items with claims/statements
  •   „missing” items (deleted, redirected or not created)
 

This is an updated and more realistic version of my earlier Claimless diagram. --Succu (talk) 21:45, 9 September 2015 (UTC)

Beautiful :-) but what does it mean? ;-) I understand that the blue "No claims" (= wikidata item without any statements) are the same as in the earlier diagram (see below), but can you explain „Missing” items for me, Succu? For example, „Missing” items in the interval [Q19200000, Q19299999] would be 5970 items deleted after creation? --Atlasowa (talk) 07:46, 10 September 2015 (UTC)
(note: per below: I inverted blue and red --- Jura 14:18, 10 September 2015 (UTC)
„Missing” refers to the json dump, where redirs and deleted items are not present. So the red bar is the sum of redirected and deleted items. An execption is the last bar, which contains not yet created items. --Succu (talk) 11:05, 10 September 2015 (UTC)
 
User:Succu/Statistics/NoClaims/20150330: Distribution of items without a claim (2015-03-30). This statistic was made from the last dump (2015-03-30). Item numbers are subdivided into intervals of 100,000. So for instance (X,Y) = (192,73720) represents the interval [Q19200000, Q19299999] having 73,720 out of 100,000 items without a single claim. --Succu (talk) 21:27, 2 April 2015 (UTC)
I like it. It gives a chronological perspective on the no-statements problem. Personally, I might have used the colors differently. I think the variation in missing items is odd, but less important than missing statements. --- Jura 11:43, 10 September 2015 (UTC)
Jura, feel free to change the colors in that line: | colors = #0000FF : #FFFFFF : #FF0000. --Succu (talk) 12:54, 10 September 2015 (UTC)
Ok, I inverted blue and red, so we still have stalagmites and stalactites. --- Jura 14:18, 10 September 2015 (UTC)
Very nice user:Succu! Make sure to put this somewhere more permanent too and findable from Wikidata:Statistics.
How hard is it to produce these kind of nice diagrams for a specific subset of items? Say for example I would like to have it for instance of (P31) -> painting (Q3305213)? Multichill (talk) 16:26, 10 September 2015 (UTC)
If the items all have instance of (P31): painting (Q3305213) then they all exist (not blue) and they would all have at least 1 statement (not red). So then all the items would be white, or are you asking about slightly different? Popcorndude (talk) 16:47, 10 September 2015 (UTC)
I guess he wouldn't get any stalactites, but one could still try to do batches of equal size by QID to visualize the number of statements per item, a bit like this (which isn't by QID). --- Jura 17:08, 10 September 2015 (UTC)
This should not be to hard to implement, but not out of the box. A solution could be to use a fourth bar (one reducing the size of the "white" bar). --Succu (talk) 17:31, 10 September 2015 (UTC)
But, Multichill, showing around 106,000 items this way would not be very impressive. ---Succu (talk) 17:45, 10 September 2015 (UTC)
To impress is not the goal Succu :-). I would love to have it to see the progress of Wikidata:WikiProject sum of all paintings. Multichill (talk) 14:15, 13 September 2015 (UTC)
A fourth bar would look like in Distribution of P225. --Succu (talk) 13:44, 11 September 2015 (UTC)
Jura, I created a color variant on my user page. Meaning green = good, yellow = attention, red = bad ;) --Succu (talk) 18:04, 10 September 2015 (UTC)
I guess we don't agree what is bad  ;) --- Jura 18:07, 10 September 2015 (UTC)
>10% of redirects/deletions (2.4 million items): it does seem like a lot. --- Jura 18:36, 10 September 2015 (UTC)
I would call this often unnecessary part as „bad”. Items generated by automated processes that say only „We have your sitelink” are of limted help. I think the old intewiki bots did a better job. --Succu (talk) 20:46, 10 September 2015 (UTC)

Sorting sequence for entities in the SPARQL service ?

Moved to the Wikidata:SPARQL query service/suggestions page

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows. @Smalyshev (WMF): Is there a sorting order for entities in the SPARQL service ? Could you add one ?

For exampe, consider the following query:

tinyurl.com/nwc47gj

In the query ?fname is a firstname, like Dave (Q16275316), and ?altname is an alternate name said to be the same as (P460) as the firstname, eg David (Q18057751).

I would like to be able to canonicalise firstnames to find the firstname with the lowest Q-number, (to be called ?name) -- eg to count the number of 'families' of firstnames, or count each family's total incidence.

However, it seems although I can find a sample alternative name (using SAMPLE), I can't reliably create a canonical one (using MIN) -- the query engine throws a Vocab(2) exception. I presume this means that there's no sequence order been declared for wikibase:Item

Is this right? and/or is there a good way round this? Jheald (talk) 14:39, 11 September 2015 (UTC)

To sort by numeric value in the QID, try
    SELECT (xsd:integer(STRAFTER(str(?fname), "Q")) As ?numid) ?fname (SAMPLE(?altname) AS ?name) WHERE {
       ?altname wdt:P460 ?fname . 
       ?fname wdt:P31 wd:Q12308941
    } GROUP BY ?fname
      ORDER BY ?numid
.--- Jura 19:10, 11 September 2015 (UTC)
@Jura1: Thanks, the xsd:integer(STRAFTER(str(?altname), "Q") is really useful.
So I can now count the number of distinct first-name families tinyurl.com/ppqekmd. Next I just need to see whether I can turn that ?numid back into an item reference, to see whether I can count the number of incidences for each family, without all the duplicate entries produced by this code tinyurl.com/p6xc6yv (which did just complete, to produce this set of counts, but at the expense of counting and reporting each family all 'n' times over -- both messy and expensive!) Cheers, Jheald (talk) 23:25, 11 September 2015 (UTC) (Of course it would be nice just to be able to use MIN() to produce an item reference directly; but this round-the-houses route may get there in the end).
First-name families with the largest number of name variants: tinyurl.com/ok5ke4q. Now I still just need to work out how to turn the numeric IDs back into items. Jheald (talk) 16:15, 12 September 2015 (UTC)
Here's a version that does now have name labels tinyurl.com/n9pfqno But this workaround because MIN doesn't work directly on items still seems messier than it ought to need to be. Jheald (talk) 09:58, 13 September 2015 (UTC)
The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.

Do we have a process for deleting items that become orphaned?

I am curious if we have a process for deleting items that become orphaned? For example, an item that has only one sitelink then that page is deleted on Wikipedia (or sister project) for reasons such as lack of notability / spam. Now the item is orphaned (e.g. unused on Wikidata) and also probably not notable here.

I know that the site link automatically gets deleted, but then what? Aude (talk) 13:52, 12 September 2015 (UTC)

There a few queries trying to track these (focusing on nearly empty items) and eventually they end up on Wikidata:Requests for deletions where they get nuked. --- Jura 14:12, 12 September 2015 (UTC)
the fact that an item does not exist on wikipedia doesn't mean it has to be deleted. Such items can be used to store info (I think of people, especially), in order to be able to use them for "lists of", even if there is not enough info to build a real sourced wp-article ;) --Hsarrazin (talk) 15:51, 12 September 2015 (UTC)
Well, sometimes we get several items as they keep deleting and re-creating articles at WP. ;) --- Jura 16:07, 12 September 2015 (UTC)
Good point about items suitable for a list, though I see a lot that probably don't fit that (e.g. stuff deleted for promotional / spam reasons).
In going through deletion requests on enwiki, I realized this might be somewhat an issue here and thought maybe there could be a tool or something to help, though sounds like we already have something. (where?) I don't think we would want to automatically delete items, but report them so humans can review them. Aude (talk) 17:05, 12 September 2015 (UTC)
A good style is that if you close an AfD request on enwiki as delete, and there are no interwiki links, to delete the Wikidata item as a part of the deletion process.--Ymblanter (talk) 17:42, 12 September 2015 (UTC)
That's what I do and agree would be nice, but in practice I think most admins don't take care of this. (they probably don't realize or think of it) Aude (talk) 08:27, 13 September 2015 (UTC)
No, most of them do not. This is why we have the instruments listed below.--Ymblanter (talk) 13:04, 13 September 2015 (UTC)
See User:Pasleim/Items for deletion/Page deleted‎], there are bot runs on the easy ones. Sjoerd de Bruin (talk) 17:08, 12 September 2015 (UTC)
There is also User:Pasleim/Items for deletion/Category for categories, which requires some extra care because categories which have been speedy moved also end up there, but they should not be deleted.--Ymblanter (talk) 17:42, 12 September 2015 (UTC)
Thanks for the links :) Aude (talk) 08:27, 13 September 2015 (UTC)

Changing interwiki - risk of erroneous deletions?

I just changed the interwiki link for one of our lemmas on sw-wiki. I noticed that I was close to making a mistake - as I luckily did not proceed to the end I am not sure if there is a kind of brake on that risk.

I went on the "edit links" button on my swwiki-page and get on the "wikipedia"-box at the "ecology (Q7150)" entry, hit edit and see 139 entries. So i start removing languages, because I do not want to have my sw-lemma linked to these. After hitting delete 20,30 times I grow tired and think "There must be another way". So I start looking for faqs, find nothing and go back to my edit window and ONLY NOW remember that I just have to remove my sw-lemma from this screen, so that I can later link it to another item on wikidata.

I feel relieved that I did not hit the "save" button yet. Would I have erased the interwiki link to (Q7150) for these 20,30 languages ? Did you notice that this happens sometimes or am I the only idiot?

Well I did change interwiki links before, but not often and this morning I was wired the other (wrong) way round. And for my morning mindset I could not find any hint how to change an interwiki link for a lemma that belongs somewhere else. Presently I feel like putting a short warning / hint just on top of the list you see after you hit "edit links" on your wikipedia page. "For a change of interwiki links remove your language from this list, then go back to your article and link anew to the desired item". Kipala (talk) 07:06, 13 September 2015 (UTC)

I don't see what's the problem here? Sjoerd de Bruin (talk) 08:16, 13 September 2015 (UTC)
If it happened only to me - no problem. If it happens more often - put in some brake. Kipala (talk) 08:23, 13 September 2015 (UTC)
Don't worry. It's a wiki, so you can edit it, but you shouldn't be able to break permanently ;)
BTW, I think there is an edit filter running on that. --- Jura 08:26, 13 September 2015 (UTC)
Maybe was a problem on label. Now I changed it for sw lang. --ValterVB (talk) 08:35, 13 September 2015 (UTC)
But is this the reason why I see so much users removing all sitelinks? Sjoerd de Bruin (talk) 08:41, 13 September 2015 (UTC)
Well, I guess if your interface language is set to a language that doesn't have a label for the item and you come from an article convinced that you are on the item for that article, the logical thing to do could be to remove all conflicting sitelinks? --- Jura 08:54, 13 September 2015 (UTC)

Yesterday, while cleaning data-twins (see above), I noticed another problem of the same kind. When, on a wp, the need for for disambiguation appears, the original name generally becomes the disambiguation page.

In this case, I found 3 cases, yesterday, where the original human (Q5) items (with iw-links) held the disambiguation. The new page for the person was created with a new name, and another wd item created with data added, hence creating a duplicated item. Those are rather difficult to trace if the new item is not completed with claims. There may be a lot of them, unnoticed. They require quite a lot of work to clean up, and could lead to accidental wipeouts. :/ --Hsarrazin (talk) 17:55, 13 September 2015 (UTC)

It's getting harder and harder to propose properties

This is the second time I am experiencing difficulty 1) finding where to propose a property (searching the discussion page of similar properties does not lead me where I need to go) and 2) proposing the property with the example template (syntax errors). Can somebody please fix this for me: https://www.wikidata.org/wiki/Wikidata:Property_proposal/Creative_work#Bildindex thx, --Jane023 (talk) 10:28, 13 September 2015 (UTC)

Fixed by Jura1 here. --Stryn (talk) 11:11, 13 September 2015 (UTC)

Mass / Weight property input

Please head over to Wikidata:Property_proposal/Generic#Generic_or_Specific:_Vote_here and give your opinion on how many mass properties we need. --Tobias1984 (talk) 10:43, 13 September 2015 (UTC)

Filter for Commons categories?

Is technically possible to create a filter to detect addition of sitelinks to Commons categories in entries compused solely of entries on NS:0 from Wikipedia (and/or NS:AUTHOR from Wikisource; that have an id number inconsistency across Wikisources)? This will be helpfull to detect changes as [4]+[5]. Lugusto (talk) 17:01, 12 September 2015 (UTC)

There are a lot of people on Wikidata, who are focussing on Commons sitelinks. But the most don't know what they are doing, removing links and merging categories with articles... Sjoerd de Bruin (talk) 17:05, 12 September 2015 (UTC)
ok, I thought Commons categories were to be added only in Commons category (P373) on items which are not Wikimedia category (Q4167836) (i.e. subjects of the Commoncat), and to Commons wikilinks on Wikimedia category (Q4167836). Am I right ? is there a page with recommandations about that, or is there no consensus yet ? --Hsarrazin (talk) 02:04, 13 September 2015 (UTC)
@555: The really important thing is to add a Commons category (P373) statement in such cases, as that is the canonical way to show an item has a related Commons category.
If we do want to stop people making sitelinks from article-items to Commons categories (the conclusion that was sort-of adopted from this discussion), we need a bot to enforce it. As the statistics I posted a couple of weeks ago show, over 100,000 such 'cross-namespace' sitelinks were created in the last 12 months -- a more than doubling in number.
We need to take a decision, either to enforce the no cross-namespace links rule, or to abandon it. Doing nothing is effectively the latter.
But the one thing that is important is to add the Commons category (P373)s, as the one guaranteed way to be able to show such a linkage. Jheald (talk) 09:13, 13 September 2015 (UTC)
@Hsarrazin: Note that it is also worth adding a Commons category (P373) on category-like items here that have a corresponding category on Commonscat, as with the present sitelinking muddle, it cannot be guaranteed that such category <-> category relationships with be discoverable by a sitelink (the sitelink may have been taken by an article <-> commonscat sitelink). Currently, per the statistics linked just above, about 80% of category-items with known related commonscats have Commons category (P373)s. It would be good if this were nearer 100%. Jheald (talk) 09:25, 13 September 2015 (UTC)

"If we do want to stop people making sitelinks from article-items to Commons categories (the conclusion that was sort-of adopted from this discussion)" <- This is a false representation of the situation. This is what Jheald wants, but it is NOT the conclusion. See for instance:

--Atlasowa (talk) 20:14, 13 September 2015 (UTC)

Polders

In the Netherlands polders often had dikes that broke at some stage. How do I register such events? Thanks, GerardM (talk)

significant event (P793)? Multichill (talk) 14:16, 13 September 2015 (UTC)
Thanks, like this ?
err... do you think "dijkdoorbraak" is "en" ? shouldn't it be "nl" rather ? if you got "en" too, would be nicer to translate in other languages ;) --Hsarrazin (talk) 21:04, 13 September 2015 (UTC)

Use of Wikidata in frwiki

We do not communicate the much on how Wikidata is used in various Wikis, so here is a quick summary about French Wikipedia:

  • Wikidata has trigged length, loopy debates. The "biographie2" template in particular has been criticized by some, and widely used by others. The bot adding it has been stopped. Criticisms of Wikidata usage are very diverse, with some main clusters:
  • we are frwiki, we don't want to depend on another site, and certainly not a geeky database (that seems to be a minority)
  • data are low quality and badly sourced
  • Wikidata is too hard to understand, I don't want to have to spend my time there
  • edits are hard to track in watchlist, and not visible in page history (the most cogent arguments imo)

Heated debates there but it seems that when Wikidata has been implemented in a template for a while, things settle pretty well. --Zolo (talk) 09:51, 11 September 2015 (UTC)

LUA-templates can avoid use of data that has Wikipedia as source or no source at all, right? That seems like a good step first. Sjoerd de Bruin (talk) 10:02, 11 September 2015 (UTC)
It can, but that would exclude just about everything. Even though there are complaints about data quality, unsourced data is usually better than no data. --Zolo (talk) 10:11, 11 September 2015 (UTC)
I guess that depends on the what statement you use. -- Innocent bystander (talk) 13:22, 11 September 2015 (UTC)
In many cases, automatically uploaded data, often from infoboxes, has a reference to xxwiki. That in my opinion is a poor (secondary) source and should not qualify as such. At least the rank of such a reference should be "history". Poul G (talk) 09:37, 12 September 2015 (UTC)
In an effort to update da:Module:Wikidata I did a review of some of the other implementations. By design they are so different, that it's hardly fair to claim, that they are really the same (same Q). I found the french to be the best design. I would support an effort to converge, ultimately with the same Lua-code in the core modules. That would require the module-names, including sub-modules, published functions and documentation to be in the same language (english). Texts shown to our readers should come from language-configurations, eg. i18n = mw.loadData("Wikidata/i18n/"..mw.language.getContentLanguage():code), with fallbacks from mw.language:getFallbackLanguages. Default formatting should rely on mw.language with posible escapes parameterized like value-module in the spanish version (inspiration for our danish, but now only readable by spanish-speaking people). Poul G (talk) 09:37, 12 September 2015 (UTC)
Somewhat funny that WMFDE develops it and it gets used first on frwiki, but not on dewiki --- Jura 06:54, 13 September 2015 (UTC)
I have tried to write the frwiki module in such a way that it can be exported, but as it came to rely on existing local modules, it did not work 100%. Surely we can draw inspiration from each other, but as long as we can't store modules in a central place, keeping them really in synch sounds a bit difficult. Another complicating factor is the particular habits of each Wiki (in the present case, the way frwiki likes to pack three internal links into every single date). --Zolo (talk) 15:12, 13 September 2015 (UTC)
The value-module/-function can handle dates with multiple links; maybe it could be reduced to one field, like formatter=Module.function. Poul G (talk) 08:07, 14 September 2015 (UTC)

UCI cyclists, cycling teams and event results

WP's like wp:de, wp:es, wp:en, wp:nl, wp:it etc. individually generate almost identical tables of the yearly changing team members array and the event (race) results from one source: Union Cycliste International. Isn't this a field of data processing that can be restructured and simplified here? --MaxxL (talk) 09:39, 14 September 2015 (UTC)

@MaxxL: Please contact user:Jérémy-Günther-Heinz Jähnick: he is working a lot to develop a data structure and infobox in lua in the field of cyclism.
@Jérémy-Günther-Heinz Jähnick: You can provide some info I think. Snipre (talk) 09:58, 14 September 2015 (UTC)
@MaxxL: Hi. The project Wikidata:Cycling is larger that you can guess. I have write a small presentation in English here. To respond at your question, we list cyclists thanks to has part(s) (P527) in the seasons item, like CCT-Champion System 2015 (Q19336144). We can add precisions thanks to qualifiers. Infos like the country or the previous team will be contain in the item of the cyclists. These informations were also used when we will write classifications. In the articles on Wikipedia, the entries of the table will be ordered thanks to P1964 (P1964), contained in cyclist'item. It lacks me the solution (a template) to establish this table on an article in the French Wikipedia. After tests, it will be translate in around 20 languages. For example, today, I translate this template in norge. If you have other questions, I am here. Jérémy-Günther-Heinz Jähnick (talk) 10:27, 14 September 2015 (UTC)
And for cycling races, two modules already permit to display datas in infoboxes, as you can see for 19e étape du Tour d'Espagne 2015 and Grand Prix de Fourmies 2015. I can also list teams thanks to participating team (P1923), I can count them in the infobox, but I don't yet have the solution to display the table in the article. I can realise a classification, like general classification of the 2015 Coupe Sels (Q20895314), but it lacks me the time (3 h 56 min 30 s for example, absolutely in one line, I launch a proposal), and, again, the solution to display it on the article. It is not a problem, it permits me to think and to discuss, and to illustrate races (the weather seems bad for wednesday, I surely rest at home). Jérémy-Günther-Heinz Jähnick (talk) 11:17, 14 September 2015 (UTC)

Fusion problems

Can someone fusion en:Ghattas (Q16783843) with German de:Ghattas ? FelicitasFlorenzeLupo (talk) 13:34, 14 September 2015 (UTC)

AFAICT they’re already merged… did someone already do it? Or what’s the problem? —Galaktos (talk) 13:41, 14 September 2015 (UTC)
It seems that FelicitasFlorenzeLupo found out himself how to do it. Mbch331 (talk) 18:33, 14 September 2015 (UTC)
Actually, I checked again, and FelicitasFlorenzeLupo only moved the page link. I completed the merge – Ghattas (Q16783843) now redirects to Ghattas (Q1521469). —Galaktos (talk) 18:36, 14 September 2015 (UTC)

Wikidata weekly summary #175

 
Here's your quick overview of what has been happening around Wikidata over the last week.

Data-twins

Do we have any "instances of human" with identical data sets? How would I search for them? I imagine that people with minimal information might have all the fields filled with the same data. They may have been born and died in the same years and and in the same cities. I imagine the fewer fields filled-in the higher the probability. I would like to work to fill in more information so they are better disambiguated. --Richard Arthur Norton (1958- ) (talk) 01:20, 9 September 2015 (UTC)

What would you consider to be part of the data set? Same labels, descriptions, aliases, all languages, sitelinks (or lack of them, since the only way for two items to have identical sitelinks is for both to have none), everything? Or just identical statements? --Yair rand (talk) 02:48, 9 September 2015 (UTC)
  • Sounds like a use case for "Wikidata Query Service" (sample: How about different items with claim[31:5] and A:P569=B:P569 and A:P570=B:P570 ). --- Jura 07:18, 9 September 2015 (UTC)
  • Here's a start towards writing such a query: tinyurl.com/nmz3mj6, which gives a count of numbers of identical birth and death dates. Next step is to refine it though, to return only dates that are known exactly, rather than 1900-2000 (ie born 19th century / died 20th century), which in the naive query comes out top of the list. Jheald (talk) 10:04, 9 September 2015 (UTC)
Here's a query counting humans with nationality:UK with birth-dates and death-dates that match to the day: tinyurl.com/pzddgsw It finds 32 hits. (Query times out if the restriction to the UK is dropped). Jheald (talk) 11:07, 9 September 2015 (UTC)
I think that this is how to add item numbers and labels to the previous query, tinyurl.com/onbsh37, but it is timing out. I don't see why it should really take much longer than the previous query. One attempt (without the ordering, or the label lookup, and limited to six returns) did complete, but took just over 60 seconds -- ie double the advertised time-out threshold. Jheald (talk) 12:07, 9 September 2015 (UTC)
Some other results: France (Q142) 71 multiples; Italy (Q38) 14 multiples; Spain (Q29) 1 multiple - with a lookup which did complete (in 60 seconds!) tinyurl.com/qaqgv4x - Jacobo Fitz-James Stuart, 17th Duke of Alba (Q1677724) and Jacobo Fitz-James Stuart, 17th Duke of Alba (Q11977861) -- and is a genuine duplicate (at it is now, until somebody merges it!) Jheald (talk) 12:26, 9 September 2015 (UTC)
  Merged Popcorndude (talk) 12:36, 9 September 2015 (UTC)
Some more counts: United States of America (Q30) timed out; Germany (Q183) 24 (but took 101 seconds); Sweden (Q34) 28; Netherlands (Q55) 18; Kingdom of the Netherlands (Q29999) 0; Belgium (Q31) 17; Canada (Q16) 12; Russia (Q159) 7; New Zealand (Q664) 7; Poland (Q36) 5; Austria (Q40) 4; Hungary (Q28) 2 (now merged); Finland (Q33) 2, checked, see below; India (Q668) 2 (1 merge, 1 coincidence 1 incorrect birth/death dates); Portugal (Q45) 1 (incorrect birth+death dates on one item); Republic of Ireland (Q27) 0; Australia (Q408) 0. Jheald (talk) 15:07, 9 September 2015 (UTC)
Note these may sometimes be trickier than they first appear.
For example, for Finland, Olavi Merinen (Q11885292) and Olavi Merinen (Q17382548) appear to be two articles for the same person on Finnish wikipedia.
In contrast, Yrjö Saarinen (Q11903052) and Yrjö R. Saarinen (Q11903045) were in fact different people (with different articles on fi.wiki); however the English language article en:Yrjö Saarinen was actually sitelinked to the wrong Finnish article, so the dates for the wrong Yrjö had been scraped from en-wiki and put onto Yrjö Saarinen (Q11903052). The lesson, I suppose, is to be careful: correspondences shouldn't always be merged blindly, even when all details match, but sitelinks should be checked. Jheald (talk) 15:53, 9 September 2015 (UTC)
Revised version of my counting query for a particular country: tinyurl.com/qzh2den
There was a subtlety I had missed: two different statements on the same item could sometimes resolve to the same date, if one was in the Julian calendar and the other was in the Gregorian calendar. Th issue this raised is now resolved by being more careful to count separate items with duplicated dates rather than statement-combinations.
Reasonator has quite a good date search, eg https://tools.wmflabs.org/reasonator/?date=1737-05-08 , which can be used to turn the duplicated dates into duplicated items. I'm now working through the UK ones. Jheald (talk) 17:35, 9 September 2015 (UTC) UK list now done, apart from the match below. Jheald (talk) 20:00, 9 September 2015 (UTC)
This one's going to need more research: same man? Or misidentified dates of birth/death? en:Freddie Osborn / en:Frederick_Osborn_(cricketer) ? Jheald (talk) 18:35, 9 September 2015 (UTC)
That's a splendid tool, thanks Jheald. I've added it to my favorites ;)
and I'm working on the French dupes - already 1 problematic dupe/fusion of clean up found — I reported it on frwiki to solve :)
there may also be dupes with different nationalities, as peoples are sometimes wrongly tagged from one country, or they have moved in their life, and are claimed by 2 different countries. Those will probably be harder to find… --Hsarrazin (talk) 23:40, 9 September 2015 (UTC)
and twins [6] :D --Hsarrazin (talk) 00:01, 10 September 2015 (UTC)
Incidentally, here's a query for the most common countries of nationality tinyurl.com/n97fsvw. The main point of adding the nationality criterion was pragmatic, to try to limit the number of items being date-searched against other items to something that the query service could cope with. 193,188 (nationality:Germany) is apparently just possible; 290,061 (nationality:USA) apparently too many.
An alternative filter could be occupation tinyurl.com/qyz4y2t. Too many with occupation:politician (Q82955) (256,901), but association football player (Q937857) (185,400), actor (Q33999) (133,197), writer (Q36180) (68,815), journalist (Q1930187) (57,940), singer (Q177220) (42,824), composer (Q36834) (40,787), author (Q482980) (34,146), athlete (Q2066131) (33,991), priest (Q42603) (27,227), baseball player (Q10871364) (27,188), and lawyer (Q40348) (26,174) should all run. I found that there were 45 twins for painters, and have offered them to WikiProject sum of all paintings.
Of course, what was originally suggested was trying to find items to merge in people with minimal information. Instead, if we're insisting on an shared accurate day of birth, day of death, and nationality or occupation, that's requiring both items to already have quite detailed information. So I'm not sure whether the number of returns really was quite so "amazingly few" as you wrote. But it's a start; and the accurary rate as a detector for issues to sort out does seem very high. User:Hsarrazin's pair of twins seems to be the only true false positive so far. Jheald (talk) 11:09, 10 September 2015 (UTC)
indeed... it's very rare that true twins die the same day ;D - those 2 were executed... --Hsarrazin (talk) 11:54, 10 September 2015 (UTC)
therefore, we have a second pair of twins, who died the same day, André Bernand (Q19953021) and André Bernand (Q19953021). :D --Hsarrazin (talk) 21:10, 12 September 2015 (UTC)
I tried to run a query to list the properties most commonly applied to humans, tinyurl.com/pfuyoc6 but it timed out. Here's the equivalent query run for dogs (number of instances = 106) tinyurl.com/pafabwa, but it times out when I try to apply it to art historians (no of instances = 5013) tinyurl.com/nndtn2x Jheald (talk) 11:49, 10 September 2015 (UTC)
related_properties can give you that. For larger sets, it takes the first 50k. --- Jura 12:20, 10 September 2015 (UTC)
Very nice!! It fell over when I tried to apply it to all items with CLAIM[31:5] (exceeded its memory allowance); but from 50,000 French people, [7], given name (P735) or place of birth (P19) might also be worth trying. Jheald (talk) 12:41, 10 September 2015 (UTC)
Here's the top 500 given names, tinyurl.com/qcj75ln. There are 7 with over 15,000 incidences that might be worth doing. Jheald (talk) 14:24, 10 September 2015 (UTC)
It should match the list at WikiProject Names. --- Jura 14:28, 10 September 2015 (UTC)
@Jura1: Yes, same top 10 in the same order with very similar counts. That's reassuring! Is there a general tool, like the 'related properties' you linked to earlier, that lists the most common values for a particular property for a WDQ solution set? It would be a useful thing to be able to link to. (And a useful addition to the existing 'related properties' tool, eg alongside the Autolist link). Jheald (talk) 14:38, 10 September 2015 (UTC)
I don't think there is. Some constraint reports list most common values. We could make a general link from one of your queries and add that to property documentation template. --- Jura 14:45, 10 September 2015 (UTC)
@Jura1: It would be nicer to have a standalone service, like the 'related properties' tool, and link to that -- which shouldn't be too hard to create (I would think), if the code for 'related properties' is available to adapt. Jheald (talk) 21:44, 10 September 2015 (UTC)
And here are the top places of birth: tinyurl.com/p94gzho. Only four with over 10,000 incidences. Jheald (talk) 14:29, 10 September 2015 (UTC)
-- page now all formatted up and operational. Jheald (talk) 21:33, 10 September 2015 (UTC)
I thought I got one Gilbert Bécaud (Q160433) and Gabriel Béchir (Q5515555) but Béchir's dates are not properly sourced (Imdb does not give dates contrary to what is stated on enwiki). Probably a copy/paste contamination somewhere... Bécaud and Béchir being very near alphabetically :/
don't know how to warn enwiki, though :( --Hsarrazin (talk) 20:12, 14 September 2015 (UTC)
I found three other ones and added them to Wikidata:Database reports/identical birth and death dates/False positives. --- Jura 08:40, 15 September 2015 (UTC)

Use heritage designation (P1435) for UNESCO World Heritage Sites instead of instance of (P31) ?

Currently, all World Heritage Sites in Wikidata are indicated using instance of (P31) World Heritage Site (Q9259). I'm sure we should use heritage designation (P1435) for this instead. Any objections to me changing all of them? e.g. Beni Hammad Fort (Q500367)heritage designation (P1435)World Heritage Site (Q9259) instead of Beni Hammad Fort (Q500367)instance of (P31)World Heritage Site (Q9259) NavinoEvans (talk) 22:10, 10 September 2015 (UTC)

Why do not we use both? I thought it is best practice to have P31 in every item.--Ymblanter (talk) 06:49, 11 September 2015 (UTC)
Indeed we should have P31 for every item, but perhaps something more precise than World Heritage Site (Q9259). /ℇsquilo 09:46, 11 September 2015 (UTC)
Yes this is bad use of the P31 property. Some World Heritage Sites are buildings and others are parks. Heritage status should be used instead of P31. --Jane023 (talk) 16:56, 11 September 2015 (UTC)
Many thanks for the feedback. I should clarify that all examples I've seen so far already have another instance of (P31) statement to indicate what the item is (as it should be), we'd simply be removing the extra instance of (P31) statement and using heritage designation (P1435) instead. NavinoEvans (talk) 19:46, 11 September 2015 (UTC)
@Jane023, NavinoEvans: I'll repeat my usual bla bla but we can totally conceptually have classes for say World heritage parks, for example. author  TomT0m / talk page 08:27, 12 September 2015 (UTC)
Yes that is an even better idea. Much cleaner. --Jane023 (talk) 17:13, 12 September 2015 (UTC)

@Jane023, TomT0m: I can see that it works conceptually, but it seems like a very bad idea in this case. For a start there are lots of different cultural heritage lists (e.g. monument historique inscrit (Q10387575), class A Swiss cultural property of national significance (Q8274529)), national historic site of Canada (Q1568567)), so to be consistent, wouldn't we need to create an item for every combination of type of site with notable heritage list! ? For example, "national historic site of Canada park", "UNESCO World Heritage park etc etc... It's so much simpler to just use heritage designation (P1435) to show it's a World Heritage Site and let P31 describe what type of thing it is as it already does (park, castle, lake etc.) NavinoEvans (talk) 19:42, 12 September 2015 (UTC)

@NavinoEvans: No we would not. That's actually one of the main advantage of classes : we can define them the way we want, which gives a lot of freedom compared to the strict heritage list the monument is listed in. This could allow, for example, to regroup buildings of listed on analog lists of several country (just saying) in the same classified building class. Also note that the two are not mutually exclusive, of course. author  TomT0m / talk page 20:32, 12 September 2015 (UTC)
@TomT0m: I fully understand the point about not being mutually exclusive, absolutely true :) But I still don't get why it's necessary make new classes. Surely we can run any query we like by using whatever combination of P31 and heritage designation (P1435) that is needed? I also don't understand why your suggestion would not involve creating lots of new items so they could be used in classification statements (like the "World Heritage parks" example that you've given). Can you please explain further? NavinoEvans (talk) 08:56, 13 September 2015 (UTC)
@NavinoEvans: The creation (or not) of new items is not really a point here. There is two ways to create a query : Using the live query point, but this needs to be written in SPARQL, so it's not doable for anyone, or follow the initial Wikidata plan, that will probably be fulfilled in the future : creating "query entities", like ther is property entites and items. Then the question of which query to create is the same as the question of whether or not create class items. (Also note that if we can use query entities in statements we might also be able to use them in instance of (P31) one, which would open some interisting workflow doors). So I don't think that, even if they are great, queries will solve magically this kind of problems :) The underlying question of whether or not create equivalents of the old so called "intersected categories" will not totally vanish. But what I fight against, in this context, is that we will have to create every possible combinations ... all the interesting part, and the exercise is called "classification", is to choose the one to create. author  TomT0m / talk page 13:26, 13 September 2015 (UTC)
@TomT0m: Apologies for the delayed response. Thanks for the extra information, some very interesting points about the future query namespace :) I can see the merit of using classes here, but are you saying that we should only use classes to describe this? or use classes in addition to heritage designation (P1435)? If there's no objections to using heritage designation (P1435) then I can add all of those statement but leave the P31 statements as they are for now - If there's any further discussion needed about the correct use of P31 in this case then I think it would be a good idea to move it over to the classification noticeboard? NavinoEvans (talk) 15:41, 15 September 2015 (UTC)

One advantage of not using instance of (P31) World Heritage Site (Q9259) is that then items will get a "missing P31" constraint violation, if they do not say what sort of site they are -- a useful check. Jheald (talk) 09:31, 13 September 2015 (UTC)

Country an administrative unit

Hi I learned from Property talk:P131#Country an administrative unit? not to use country as administrative unit, therefore I reverted this edit. I do not see why this should be needed by constraint on P150, as Vigneron told. As I learned all other French regions have located in the administrative territorial entity (P131) France, which IMHO should be changed as well. I do not see why located in the administrative territorial entity (P131) = Metropolitan France (Q212429) should be removed and do not understand what adding exceptions on contains the administrative territorial entity (P150) should mean.--Oursana (talk) 19:54, 14 September 2015 (UTC)

I don't know about P150 constraints but I can tell you Metropolitan France (Q212429) isn't an administrative unit. The upper level for a French region would be France (Q142).  – The preceding unsigned comment was added by Casper Tinan (talk • contribs).
I suppose it depends how you define "administrative unit". --- Jura 09:16, 15 September 2015 (UTC)
@Oursana: thank you for raising this question. I can wait to have a clear answer.
@Casper Tinan: true, I've planned to remove these located in the administrative territorial entity (P131) = Metropolitan France (Q212429) claims (of there is not opposition of course).
@Jura1: apparently, in most definition, country is not an administrative unit (see the link discussion on the P131 property talk page). Fine, It's maybe a good call but since there is the contains the administrative territorial entity (P150) constraints, it's quite a mess…
Cdlt, VIGNERON (talk) 10:11, 15 September 2015 (UTC)
I meant to refer to Metropolitan France (Q212429). --- Jura 10:13, 15 September 2015 (UTC)
Metropolitan France (Q212429) isn't a country but a geographic entity. In France, there is no administrative body that covers Metropolitan France (Q212429).--Casper Tinan (talk) 12:22, 15 September 2015 (UTC)
As there is billions of administrative units in the french administration, I'm pretty sure we can find at least one matching the Metropolitan France (Q212429) ; but the item Metropolitan France (Q212429) is not about a admistrative unit. The only claim right now is , wich is probably improvable (at least it should be consistent with and other claims), so for the time being it's illogical to use Metropolitan France (Q212429) in located in the administrative territorial entity (P131). Same question for France (Q142), is it an administrative unit? and if so we should 1. re-start this discussion Property talk:P131#Country an administrative unit? and 2. remove or more probably add exception on the inverse constraint of Property talk:P150. Cdlt, VIGNERON (talk) 16:58, 15 September 2015 (UTC)
A country is not an administrative division. This is why located in the administrative territorial entity (P131) refers to an entity instead of a division - so that P131 can be used to to track the hierarchy all the way up to the country (and higher in some cases). For me a country is definitely an administrative territorial entity. Joe Filceolaire (talk) 18:23, 15 September 2015 (UTC)

Interwiki conflict

I am trying to interwiki link nlwiki:Hazenpootje (paddenstoel) with enwiki:Coprinopsis lagopus, but this is not allowed because it says it is already used by item (Q13703047). I fail to understand what I should do. Can someone please help? Dwergenpaartje (talk) 14:15, 15 September 2015 (UTC)

I   Merged Coprinopsis lagopus (Q13703047) with Coprinopsis lagopus (Q162599) --Pasleim (talk) 15:45, 15 September 2015 (UTC)

death location of Carlos Herrera

Carlos Herrera y Luna (Q436815) died in Guatemala according to two es and de, in Paris according to two en and fr, so in case anyone has good sources at hand to fix this. More generally, we would need to think about some better workflow for this kind of situation (dunno, like an "ask for another look" button that would add the page to an equivalent of WD:Interwiki conflicts). --Zolo (talk) 06:38, 15 September 2015 (UTC)

Such pages don't work, as we don't have enough users that focus on such things. Sjoerd de Bruin (talk) 06:59, 15 September 2015 (UTC)
Adding Template:Citation needed (Q5312535) to all involved articles seems a lot of work. --- Jura 08:58, 15 September 2015 (UTC)
Property-based constraint violation reports are widely used, that could be a guidance about what to do. I mean, we could have "report an issue" basic form, and it would be automatically transcluded to the property talk page so that people would know they should tackle the case in priority. Don't know how well that would work, but if anyone is willing to write a tool, I think it would be worth a try. --Zolo (talk) 21:01, 15 September 2015 (UTC)

How do you call a room whose main purpose is to contain a stair ?

I created stairwell (Q20972918) in order to document that the legal protection of a building only applied to it, but I can't find any English word. Apparently, "staircase" only means that flights of stairs themselves, not the whole room. --Zolo (talk) 18:03, 15 September 2015 (UTC)

I believe in English a staircase in contained in a stairwell. Popcorndude (talk) 19:04, 15 September 2015 (UTC)
stairwell (Q2451751) --Diwas (talk) 19:22, 15 September 2015 (UTC)
Ah that seems to be it, thanks everyone ! --Zolo (talk) 20:31, 15 September 2015 (UTC)

Wikidata Query Service

We've just announced the beta release of Wikidata Query Service. Please see the full announcement: https://lists.wikimedia.org/pipermail/wikidata/2015-September/007042.html

TLDR version:

Enjoy!  – The preceding unsigned comment was added by Smalyshev (WMF) (talk • contribs) at 06:50, 8 September 2015‎ (UTC).

This is really exciting news. I stumbled on the beta a couple of days ago (see thread above), and the documentation and examples and the translator from WDQ made it a lot easier to pick up SPARQL for the first time.
One thing for people to perhaps be aware of, though, is that this new beta service may sometimes be a lot slower than the current version of WDQ. For example, here's a query to count items in the sub-class tree of occurrence (Q1190554), using WDQ and using the new WQS beta. The WQS is over 30 times slower, taking 2164 ms, compared to 63ms for WDQ.
As a result, it's rather easy for queries to hit to 30-second compute time limit on WQS, even if it seems WDQ can manage the same query with no problem. For example, extending the query above to count the instances as well as the subclasses currently times out on WQS, whereas with WDQ although it takes a hundred times longer than the prvious query, WDQ does manage to complete it in about 6 seconds.
So this explains why the more complex query I was trying to run earlier was failing to complete on the SPARQL service.
But I have to say, I am already very impressed by the beta WQS service, and presumably from now on it will only get faster and faster, as it gets more and more scaled up and more and more optimised. So big kudos for this to all the team. Jheald (talk) 09:25, 8 September 2015 (UTC)
Looks complicated to me (compared to WDQ) but after some studying I will probably be able to make queries. But will there also be a functionality as in Autolist to make edits (new claims) on the result set? Michiel1972 (talk) 13:26, 8 September 2015 (UTC)
Yes, it is slower than WDQ for some queries and SPARQL has steeper learning curve (though see the links on the home page of the service for WDQ to SPARQL sytax translator). The advantage is you can run more advanced queries. This is a usual trade-off between narrow specialization and greater flexibility, so for some queries WDQ is better now, some it can't do and SPARQL can. Of course, we're just starting with the SPARQL engine so it will be getting better with time. --Smalyshev (WMF) (talk) 19:22, 8 September 2015 (UTC)
See also the thread above, Wikidata:Project_chat#WQS:_Searching_for_items_with_reference_to_.22Le_Figaro.22, for some more query examples. Jheald (talk) 19:41, 8 September 2015 (UTC)
Nice! Very very minor nitpick though: From an aesthetical standpoint, it annoys me that the URIs for Wikidata entities, properties etc. must be given with http:// and don’t work with https://. Could the interface be changed to support HTTPS URIs too?
(Yes, I’m aware it makes no technical difference, and my browser isn’t actually fetching anything over HTTP. I just have a universal aversion against the text “http://” in any context.) —Galaktos (talk) 21:53, 8 September 2015 (UTC)
@Galaktos: This is a bit deep topic for which there is no well established rules in the data world AFAIK, but basically there is two things here in play, which look the same but are different things. The first thing is object identifiers, or IRIs. They are unique - i.e., different IRI means different object. So, https://www.wikidata.org/entity/Q1 and http://www.wikidata.org/entity/Q1 are different IRIs. Making them work the same I guess could be possible, but it would deviate from RDF/SPARQL standards. The second thing is URIs under which these objects can be found, e.g. on wikidata.org. Those can be https and http, and point to the same resource, but it's not the same as resource identifier. So for identifier we have to choose one. We chose http as most schemes out there use http. That does not prevent one from transmitting data about the entity over HTTPS - indeed, if you use http://www.wikidata.org/entity/Q1 as URI, the data will still be transmitted over HTTPS as wikidata.org produces necessary redirects. --Smalyshev (WMF) (talk) 23:27, 8 September 2015 (UTC)

Suggestions for Wikidata Query Service (WQS)

Suggestions moved to Wikidata:SPARQL query service/suggestions -- feel free to re-factor them and add more thoughts there. Jheald (talk) 09:01, 16 September 2015 (UTC)

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.

https://tools.wmflabs.org/pagepile/ Good one, thanks. --Atlasowa (talk) 08:39, 9 September 2015 (UTC)
 

On the wikidata mailinglist was the suggestion that we need to catalogue the queries that are most important for our use cases and that do not do well on the SPARQL endpoint. We also need "a page on-wiki to collect these" (Wikidata:Wikidata Query Service? Why is the manual only on mediawiki?) Here are some general suggestions:

  • Don't provide URL-shorteners like this: http:// tinyurl.com/nmkee2z (Can't link directly to the URL generated by the 'short URL' button on the query page, because the URL-shortener is blocked by the wiki software). But providing perma-links to queries is a must-have.
  • Make it easy to publish successful queries that others can reuse/modify, with a short summary/description of what it does. See for example https://quarry.wmflabs.org/query/runs/all where you can also fave/"star" queries (but don't publish all failed queries too, too much noise).
  • Querying sources/references is the biggest benefit of WQS, imho. Including strings/reference URL (P:P854) (Wikidata does have Special:LinkSearch, but...) If i want to consider using wikidata property values in Wikipedia infoboxes or lists, i want to know: Does the data have references/sources? Are 99% of those references the dreaded "imported from wikipedia" type? How many have "real" references and what are they?

--Atlasowa (talk) 08:32, 9 September 2015 (UTC)

I agree with all of these :) - One more suggestion, though, very visual :
  • change the font and size of the result list : on my screen it is very tiring and difficult to click on very small letters. - a sans-serif would be better.
  • develop a more user-friendly interface for making query. I am now able to modify the queries that were given to me, to change year or language link, but I don't understand how it queries reference, or how to modify the result display. :/
thanks a lot --Hsarrazin (talk) 19:25, 9 September 2015 (UTC)
  • It would probably be good to set up a permanent suggestions page on this wiki -- much better than a Google form, because then we can read and think about what everybody else is saying.
One change I would like to see is in the output formatting of the results. It would be nice if I could simply copy-and-paste results into a text editor, and have them in a nice tab-separated columns. Instead there's an odd mixture with carriage returns and additional stars that then needs clean-up with a text editor.
Another nice addition to the output would be Reasonator links, for items and dates. Jheald (talk) 10:09, 13 September 2015 (UTC)
+ links opening by default in a new tab. --- Jura 10:23, 13 September 2015 (UTC)
  • It would be nice if the labels service could also find labels for properties in the wdt: namespace.
For example, this query tinyurl.com/o6jky56 to find the most common properties that point to Wikipedia disambiguation pages works, but times out when I try to look up the corresponding property-entity in the wd: namespace to add a label. Jheald (talk) 08:11, 16 September 2015 (UTC)
The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.

Wikimedia administration category page

What's the inclusion criteria for Q15647814? Why , for example, Q3740, Q6741743, Q9195632, Q6378333 or Q20010800 are not in there? Paucabot (talk) 06:32, 15 September 2015 (UTC)

Wikimedia user language category (Q20010800) is a subclass of Wikimedia user category (Q20769287) which is a subclass of Wikimedia administration category (Q15647814). The other items are all older than Wikimedia administration category (Q15647814) and presumably just haven't been changed yet. - Nikki (talk) 11:11, 15 September 2015 (UTC)
Thanks, Nikki. Paucabot (talk) 11:24, 16 September 2015 (UTC)

Item descriptions in all languages?

Hi All

I'm currently working as Wikimedian in Residence at UNESCO and would like to encourage translations of item descriptions within UNESCO programmes (e.g all the World Heritage sites), what would be the best way for potential contributors to see which item descriptions were in each language and what was missing?

Thanks

Mrjohncummings (talk) 09:44, 16 September 2015 (UTC)

It might just be my preference, but, I'd prioritize contributions to Wikidata in this order:
(1) create items, (2) add labels in one language, (3) add statements, (4) add labels in English if not done, (4) add statements, (5) add labels in ar/ja/ru/zh, (6) add statements, (7) add labels in any other language, (8) add descriptions.
If an item (A) has statements (B) and these statements have labels, it's possible to auto-generate descriptions for items (A) merely based on the labels of the statements (B).
If you are ok to share content of http://whc.unesco.org/en/list/ it would be great to add the elements about their various parts into Wikidata. It might be possible to find a way to include the WHS short_descriptions into Wikidata as well (preferably as statements). --- Jura 12:49, 16 September 2015 (UTC)
I agree. Adding statements is much more important than adding descriptions because statements are queryable, can be used to identify the subject and are structured data. Descriptions however are ambigous, not structured and only valuable for the humans that read them. In lots of cases, one could create a description automatically if there is at least the instance of property set. Other properties can enhance that further. -- Bene* talk 13:34, 16 September 2015 (UTC)
Is there a central place where people organise translations on Wikidata? The translators noticeboard? It would be great if I can point people towards instructions for doing the translations. I'll answer the other part next week hopefully.
Cheers
Mrjohncummings (talk) 13:37, 16 September 2015 (UTC)

Arbitrary access done on enwp and more

Hey folks :)

We just enabled arbitrary access on enwp and more \o/ Wikidata:Arbitrary access has the full list. --Lydia Pintscher (WMDE) (talk) 10:49, 16 September 2015 (UTC)

Great, I love it! :) --Stryn (talk) 11:01, 16 September 2015 (UTC)
Great! Are there any wikis left that don’t have arbitrary access yet? —Galaktos (talk) 11:25, 16 September 2015 (UTC)
I _think_ we have all Wikipedias by now. Commons is definitely still missing. Not sure if any other project is still waiting. Will need to check with Katie. --Lydia Pintscher (WMDE) (talk) 11:51, 16 September 2015 (UTC)
@Lydia Pintscher (WMDE): What's the timescale foreseen for Commons, that we could be thinking about trying to get Commons:Template:Creator, Commons:Template:Institution and Commons:Template:Artwork ready for? Jheald (talk) 16:47, 16 September 2015 (UTC)
I am trying to figure that out right now. I just today talked with my team about this and there are some more blockers we need to tackle. phabricator:T98307 is tracking that. I hope we can get to working on it within the next month. --Lydia Pintscher (WMDE) (talk) 16:54, 16 September 2015 (UTC)

Which classes of items do properties most often connect together?

What classes of items are most often connected by a property prop ?

Courtesy of the new SPARQL query service, here are some queries to find out for some of the properties that are most frequently used.

The results can be quite interesting! Mostly they make very good sense, but some can be a little ... odd. In each case, follow the link and then hit the 'Execute' button to run the query.

To look up your own favourite property, click on one of the links above, then scroll the query window down and change the line BIND (wdt:Pxxxx AS ?p) to the property you desire. Choose any property with type 'wikibase-item' from the list at Wikidata:Database_reports/Constraint_violations/All_properties.

Properties linking to disambigation pages, lists, categories etc

Quite a few properties in the reports above seem to link rather often to items in class Wikimedia disambiguation page (Q4167410), or Wikimedia list article (Q13406463), or Wikimedia category (Q4167836), when perhaps they shouldn't. Here are some queries to find out which properties link to them the most:

I wasn't yet unfortunately able to come up with a way to add the class labels to the queries above without them going over their time limit, so here are some results from this morning for the most common inbound properties to

Wikimedia disambiguation page (Q4167410):

Here are queries for the disambiguation items that are the most common targets of those properties:
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property family name (P734)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property given name (P735)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property said to be the same as (P460)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property cast member (P161)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property instance of (P31)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property category's main topic (P301)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property place of birth (P19)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property located in the administrative territorial entity (P131)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property has part(s) (P527)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property place of death (P20)
most connected-to items in class Wikimedia disambiguation page (Q4167410) that are the target of property subclass of (P279)

Wikimedia list article (Q13406463):

Again, queries for the items that are the most common list items that are the target of these properties:
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property instance of (P31)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property category's main topic (P301)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property transport network (P16)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property list of monuments (P1456)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property minor planet group (P196)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property position held (P39)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property award received (P166)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property part of (P361)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property follows (P155)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property followed by (P156)
most connected-to items in class Wikimedia list article (Q13406463) that are the target of property subclass of (P279)
In most cases the items on this list should be turned into an item for the thing itself, with a new item created for the Wikimedia list article -- which is actually true for most of these properties: very few of them (with the obvious exception of list of monuments (P1456)) should actually be pointing to list articles.

Wikimedia category (Q4167836):

Queries to investigate the items that are the targets of properties that, on the face of it, might not be expected to be pointing to categories:
most connected-to items in class Wikimedia category (Q4167836) that are the target of property subclass of (P279) -- mostly sane: identifies sub-categories
most connected-to items in class Wikimedia category (Q4167836) that are the target of property follows (P155) -- sane: mostly sequential categories by date
most connected-to items in class Wikimedia category (Q4167836) that are the target of property followed by (P156) -- sane: mostly sequential categories by date
most connected-to items in class Wikimedia category (Q4167836) that are the target of property part of (P361) -- sane(?): more subcategories
most connected-to items in class Wikimedia category (Q4167836) that are the target of property instance of (P31)
-- Being used for things like days of the month (see eg the 'instance of' in-links for Category:August 2005 (Reasonator  ) Is this right?

The templates {{MostConnectedClasses}}, {{InboundPropertiesToClass}} and {{MostConnectedToInstances}} used above are still rather rough-and-ready, so if anyone wants to improve them, and bring them up to a more appropriately shiny standard, feel free!

There's quite a lot there in the reports above; but there are some interesting things (and some revealing anomalies) when one starts to drill into them. Jheald (talk) 14:36, 16 September 2015 (UTC)

Entries that have weird names and return on queries that shouldn't

The three following entries keep returning on sparql queries using regex, queries that they shouldn't return to. https://www.wikidata.org/wiki/Q2608398 https://www.wikidata.org/wiki/Q689438 https://www.wikidata.org/wiki/Q844059

For example, in http://lod.openlinksw.com/sparql try the following query 

  PREFIX wd: <http://www.wikidata.org/entity/>
  SELECT DISTINCT ?p, ?title WHERE {
   ?p a ?c.
   ?c rdfs:subClassOf* wd:Q2431196.
   ?p rdfs:label ?title .
   FILTER REGEX(?title,"^(Vamp)(( [(].*[)])?)$")
   ?title <bif:regexp_match> "^(Vamp)(( [(].*[)])?)$"
  }

I think it has to do with the characters in the GOT language. How should I proceed? remove the unidentified characters? --Bernardofbbraga (talk) 15:14, 16 September 2015 (UTC)

have you tried filtering on language in the SPARQL query? Do you really want the title query to match in any language? Otherwise use something like FILTER(lang(?title) = 'en'). ArthurPSmith (talk) 15:49, 16 September 2015 (UTC)
well, I suppose I could filter, but the question is: aren't those entries ill-formed? They return on many different queries and seem counter intuitive to me.--Bernardofbbraga (talk) 18:06, 16 September 2015 (UTC)
I think that this is the equivalent query on the WMF service, and for me the regular expression is returning no hits. (But that might be because I have the language switched to English).
Can you replicate the hits you're getting using this query?
Then we would be able to try to work out whether what is strange is only occurring on the lod.openlinksw.com, or whether it is more systematic.
Regular expressions are a notoriously easy place to make subtle bugs, that can take forever to definitively track down. If you can replicate the strangeness you're seeing on the service here, then that would be a significant step forward to isolating where the unexpected may be getting in. Jheald (talk) 18:40, 16 September 2015 (UTC)

Autodesc on search results...

Hello,

I thought that Autodesc was activated by default for all users, on search page results[8], and indeed, it's activated for me, but Thibaut120094 tells me he doesn't see descriptions when there is no label. I've been used to it for so long that I can't remember when and how it was activated…

So, is Autodesc default-activated ? and if not, how do you do to activate it ? Thanks a lot. --Hsarrazin (talk) 13:25, 12 September 2015 (UTC)

Hoi, it is not. Imho that is an embarrassment. Thanks, GerardM (talk) 13:40, 12 September 2015 (UTC)
Ok, I found the source in my common.js
 mw.loader.load('//tools.wmflabs.org/wikidata-todo/autodesc.js');
GerardM, the fact that you don't like it does not mean it is not very useful for a lot of people, especially to easily see which items have some claims or not, on the fly, not searching for it.
I am surprised it is not even a Gadget (thus easy to activate or de-activate at will). I remember seeing the question somewhere, months ago.
so, please, why is it not a Gadget, yet ? --Hsarrazin (talk) 13:46, 12 September 2015 (UTC)
Actually the disgrace is that autodescriptions is not what we use. Thanks, GerardM (talk) 15:50, 12 September 2015 (UTC)
GerardM, what do you mean what we use... please don't think that everybody works like you :) most items have no description at all in many languages… - I use it everyday to target items that should be completed a minima — most items have no description at all in many languages… :) --Hsarrazin (talk) 15:54, 12 September 2015 (UTC)
@Hsarrazin: we should default include code hosted on toollabs. It should be move to one of the production sites first. Multichill (talk) 19:40, 16 September 2015 (UTC)

Labels for nameless persons

Is there some usual way to handle situations when the name is currently unknown to the science? Namely stumbled upon the mother of en:Treniota, Q20740118. Giving a description as a label like "Treniota's mother" sounds weird to me. Plus when plunged into Wikipedia infocards (say here) it makes it look like a schoolbook for very special children, "Treniota's mother is mother of Treniota". --NeoLexx (talk) 19:10, 13 September 2015 (UTC)

I think I’ve seen some “mother of X” items, for fictional characters IIRC. Can’t find them now though. —Galaktos (talk) 19:41, 13 September 2015 (UTC)
@Neolexx: For artists we use notname (Q1747829). Multichill (talk) 19:35, 16 September 2015 (UTC)

Decision

OK, nobody else, so I go for it. At the Russian forum also reminded about nameless daughters and sons of Chinese emperors ("16th son of..."). So "mother of Treniota is mother of Treniota", "5th daughter of Chinese emperor X is 5th daughter of Chinese emperor X" and so on. Captain Obvious is applauding respectfully... But no big harm in a little fun for readers after all. --NeoLexx (talk) 22:27, 14 September 2015 (UTC)

Property:P571

Is it really wanted, that Property:P571 is inserted in every single property item with its creation date? Cf [9], [10], [11], [12], [13] and so on. This is contradictory to what I thought Wikidata was about: describing real things, not describing itself. How does the creation date of a Wikidata property item characterize the property's content/subject? P.S. See also User talk:Thierry Caro#Property:P571 --тнояsтеn ⇔ 06:24, 15 September 2015 (UTC)

+1. The Date of page creation is given in the Page information (example). This is sufficient not misleading to readers as opposed to the use of inception (P571). --Leyo 08:39, 15 September 2015 (UTC)
Others think so, too -> [14] --тнояsтеn ⇔ 18:29, 15 September 2015 (UTC)
+2, removed it. Multichill (talk) 18:59, 15 September 2015 (UTC)
I think it was an interesting idea to store in properties' metadata the date of creation. I wonder whether there is another way of retrieving it different from accessing it from revision table. Matěj Suchánek (talk) 19:10, 15 September 2015 (UTC)
Using the api? This request does it: [15]. -- Bene* talk 13:30, 16 September 2015 (UTC)
Aren't statements on properties about the property? Besides, I don't think pageinfo is part of Wikibase. --- Jura 19:38, 15 September 2015 (UTC)
Yes, but those are specific "property" properties which should only be used on properties. On items, P571 also is about the creation date of the subject, not of the item itself. -- Bene* talk 13:30, 16 September 2015 (UTC)
P31 is widely used on properties. --- Jura 14:42, 17 September 2015 (UTC)
тнояsтеn: Wikidata is about representing "real" things represented by items with Qnumbers. Many properties have equivalent items linked by Wikidata item of this property (P1629). These items represent the real things but the Properties do not represent "real things". Properties are a tool we use to create statements to describe "real things". This tool can also be used to create statements to describe properties. Hope that helps. Joe Filceolaire (talk) 17:12, 17 September 2015 (UTC)

pb with VIAF links

Since yesterday, there is a pb with VIAF AC-links.

Clicking on a VIAF link of a person, leads to a page "Échec de la connexion sécurisée" (failure of secured connection).

Don't know the reason of that, but it's very problematic to check or import other AC. --Hsarrazin (talk) 20:21, 15 September 2015 (UTC)

What is a pb ? GerardM (talk) 15:08, 17 September 2015 (UTC)
Problem, I think it has something to do with the expired certificate of wmf labs that day. Sjoerd de Bruin (talk) 15:13, 17 September 2015 (UTC)

Unit support on Wikipedias

Apparently the enwiki template Infobox telescope now takes the altitude from Wikidata as well: see for example en:South Pole Telescope. Unfortunately, the unit support doesn’t seem to work all that well… it just displays the URL (IRI?) of the unit :D
Is this a known issue, tracked somewhere? —Galaktos (talk) 22:05, 15 September 2015 (UTC)

Oh shoot. I thought we tested that and it worked -.- I opened phabricator:T112737. --Lydia Pintscher (WMDE) (talk) 07:59, 16 September 2015 (UTC)
Not sur how that is supposed to work in the core: just retrieving the label would miss the "s" for the plural form. 2.8±0.1 sounds rather weird anyhow. I think en:Module:Wikidata would need to do more work by itself. In cas it can help, the French version simply sends the amount and unit to Module:Convert (Q14033926), and let it handle things. --Zolo (talk) 08:20, 16 September 2015 (UTC)
This is fixed now: The label is being shown instead of the item's URL for quantities with units. - Hoo man (talk) 17:05, 17 September 2015 (UTC)
@Galaktos, Lydia Pintscher (WMDE), Zolo: fr:WP partially solves the problem by creating a conversion table which translate the Q number of the units from the json code into string. See the module doing this conversion. Snipre (talk) 12:45, 16 September 2015 (UTC)

Number of firms per sector

On French cheeses items, I would like to add the number of milk producers, cheeses producers... I didn't find any property that could be used for that. And of course, I will not ask for a specific property about the number of milk producers ;) What should be created ? Pyb (talk) 19:42, 16 September 2015 (UTC)

Perhaps enlarge the use of field of work (P101). Snipre (talk) 11:18, 17 September 2015 (UTC)
@Pyb: My option would be to create a class item cheese producer, subclass of firm, with a statement like produce <cheese>, and another number of instances : n. author  TomT0m / talk page 11:48, 17 September 2015 (UTC)
Blessed are the cheesemakers. Joe Filceolaire (talk) 16:55, 17 September 2015 (UTC)

Wikidata:Database reports/Recent deaths

When I try to manually update Wikidata:Database reports/Recent deaths I get the error No template match No template. Can anyone fix this ? --Racklever (talk) 07:38, 17 September 2015 (UTC)

The page is too large. Sjoerd de Bruin (talk) 08:40, 17 September 2015 (UTC)
@Magnus Manske: It is the listeria site that gives the error (when you follow the link "manually update" at the top of the table). Can you check what's going on?
@Racklever: The link goes to an external site of a tool designed by Magnus Manske. I've pinged him so he can check what's going on. Mbch331 (talk) 08:43, 17 September 2015 (UTC)
It turns out that the page was too large. It is now fixed.--Racklever (talk) 10:41, 17 September 2015 (UTC)
Magnus changed the output format recently. It now gets too large much quicker. --- Jura 10:59, 17 September 2015 (UTC)
Working on solutions ;-) --Magnus Manske (talk) 14:41, 17 September 2015 (UTC)

Import IW-links to a WP article

Hi!

Is there a way to import a list of interwiki-links to a Wikipedia article or discussion?

I've only done this manually, and this is what it looks like: 

The article about Balder (roller coaster) is also avaliable in de, fr, sv.

Code: 

The article about [[Balder (roller coaster)]] is also avaliable in [[:de:Balder (Liseberg)|de]], [[:fr:Balder (Liseberg)|fr]], [[:sv:Balder (berg- och dalbana)|sv]].

I've used this in different tables for projects and such (like here on svwp), but when there are many links (like Jóhanna Sigurðardóttir, Q57772, who has an article in 60 languages) it takes too much time. I want to create a template (on SVWP) that makes this automatically. -abbedabbtalk 12:41, 17 September 2015 (UTC)

If the articles are linked to the same Wikidata-item, the links appear in the left column. If you want them in the article (which as far as I know most Wiki's don't want), I think it has to be done through lua. Mbch331 (talk) 13:12, 17 September 2015 (UTC)

familyname versus surname

Should the description of "instance of family name" be "family name" or "surname". It seems like no label is standard and it would be nice if it was. "Surname" is ok, but the plain English "family name" conveys the information more clearly, especially where not all readers will have English as first language. The other property is "given name", so they should match. --Richard Arthur Norton (1958- ) (talk) 22:34, 10 September 2015 (UTC)

In some countries the family name is first. Using surname can cause confusion in those cases so I prefer family name. Joe Filceolaire (talk) 22:25, 12 September 2015 (UTC)
Should we change all of them from "surname to "family name" in the label field? --Richard Arthur Norton (1958- ) (talk) 21:22, 17 September 2015 (UTC)

Requesting deletion of a non-item page ?

How do I request deletion of a non-item page?

Template:QySPARQL and Template:QySPARQL2 were a couple of experimental templates that I created yesterday, that didn't work and I have now found a better approach for.

To request their removal, is there a particular tag I should put on them? (And also in future, eg for user sub-pages when these are no longer needed?)

Thanks, Jheald (talk) 17:17, 18 September 2015 (UTC)

{{Delete}} is usually enough. Matěj Suchánek (talk) 18:02, 18 September 2015 (UTC)

Units are live! \o/

Hey everyone :)

As promised we just enabled support for quantities with units on Wikidata. So from now on you'll be able to store fancy things like the height of a mountain or the boiling point of an element. Quite a few properties have been waiting on unit support before they are created. I assume they will be created in the next hours and then you can go ahead and add all of the measurements.

Cheers --Lydia Pintscher (WMDE) (talk) 19:49, 9 September 2015 (UTC)

I think the new datatype still needs to be added to Module:I18n/datatype. Mbch331 (talk) 20:09, 9 September 2015 (UTC)
Is there documentation? I would like to expand Wikidata:Units. Blue Rasberry (talk) 20:25, 9 September 2015 (UTC)
It is not obvious what property one would add to a mountain to give the elevation of the summit. Jc3s5h (talk) 20:31, 9 September 2015 (UTC)
I used elevation above sea level (P2044) and proposed a property to record what the height is relative to (here: some sea level). —Galaktos (talk) 21:01, 9 September 2015 (UTC)
Thanks Lydia! I see length (P2043) and elevation above sea level (P2044) have been created. Can somebody with appropriate permissions look through the list here: Wikidata:Property_proposal/Pending/2 and get more of these in? ArthurPSmith (talk) 20:53, 9 September 2015 (UTC)
This link shows all newly created Quantity properties since the Units rollout. Explore! —Galaktos (talk) 20:59, 9 September 2015 (UTC)
I have changed the name of height to elevation above sea level (P2044), so that this property matches a specific height proposal. Note that I have had to change labels in some languages I don't speak, so please check and change these as needed. If you are creating properties, please be sure to follow exactly what was proposed and accepted. Danrok (talk) 22:13, 9 September 2015 (UTC)
There’s no single sea level. This dewiki page lists 28 different levels, with a variance up to 230 cm. I think it’s better to specify the exact level the value is relative to with a qualifier. —Galaktos (talk) 22:55, 9 September 2015 (UTC)
I agree, measurements are not so simplistic, and we need to be following accepted standards. Bear in mind that I simply put this property in line with an approved proposal. Danrok (talk) 23:54, 18 September 2015 (UTC)

Great news. I did not follow precisely the tests about quantities with units. So from what I understand, we can now add a number associated to an unit (that is an item itself).In the future, is it planned to enable the possibility to translate unit to another one? For example, if someone adds an unit in meter, will it be possible, at some time, to display automatically the value in feet for example. Pamputt (talk) 05:08, 10 September 2015 (UTC)

Yes :) --Lydia Pintscher (WMDE) (talk) 06:16, 10 September 2015 (UTC)

Great, but since the unit is an item, the unit maybe should be a link, to easy verify which definition of meter is used (see below)? -- Innocent bystander (talk) 06:46, 10 September 2015 (UTC)

Excellent. I think it's good that this has gone live. It might be easier to sort out some of the remaining issues.
BTW it seems that number properties have become properties with units as well. That's odd. Especially after all that talk about not using number for quantity ;) --- Jura 12:58, 10 September 2015 (UTC)

Sorry, I'm late. What a wonderful day for Wikidata! \o/ :D \o/ :D \o/ :D \o/ :D \o/ :D \o/ :D --AmaryllisGardener talk 16:24, 14 September 2015 (UTC)

@Lydia Pintscher (WMDE): @AmaryllisGardener: @Tobias1984: : I give you just two examples of the use of datas with units from Wikidata in the French Wikipedia, it is about cycling : Grand Prix de Fourmies 2015 (I also take photos) and 19e étape du Tour d'Espagne 2015. Modules are here and here. Three other modules are used in cycling (here, here and here). Jérémy-Günther-Heinz Jähnick (talk) 17:32, 14 September 2015 (UTC)
@Jérémy-Günther-Heinz Jähnick: That is really great! The pages and pictures look amazing and hopefully the infobox-usage will catch on. --Tobias1984 (talk) 17:41, 14 September 2015 (UTC)
Thank you. As you can see, infoboxes-of-one-line are now a reality even for complex infoboxes. Jérémy-Günther-Heinz Jähnick (talk) 17:56, 14 September 2015 (UTC)

Property creators

Dear property creators: Please only create the properties from the page Wikidata:Property_proposal/Pending/2 that meet the current standards of Wikidata. There is some stuff on that page which is highly out of date and was discussed prior to a good concept of the datatype. Make sure you transfer the information to the new Template:Property documentation. Also make sure you move the properties that you create or dismiss to the archive marking them as either   Done or   Not done. --Tobias1984 (talk) 17:39, 11 September 2015 (UTC)

Also: Try to archive duplicate proposals next to each other (under the same headline). And it is the job of the property creator to inform the relevant WikiProject for very specific properties. --Tobias1984 (talk) 17:43, 11 September 2015 (UTC)

Wu Zhen : People's Republic of China painter (1280–1354)  ?????

Hello,

Is there an item for "Imperial China", that lasted for 2 milleniums... as it is really a problem to consider long dead people as citizens of People's Republic of China (Q148)...

--Hsarrazin (talk) 01:50, 13 September 2015 (UTC)

and if you follow Imperial China (Q9008156) back to one of those disambiguation pages you will reach a list of lots of different items for different chinese empires any one of which can be the value for a country of citizenship claim such as <country of citizenship (P27):Yuan dynasty (Q7313)> Joe Filceolaire (talk) 03:13, 14 September 2015 (UTC)
(there is a related discussion on WD:Bistro)
The meaning of Yuan dynasty (Q7313) should be clarified, it can't be an instance of state and an instance of dynasty at the same time. But "state" may not be the right word anyway. It would rather be something like "political regime", and thus would not really apply to P27. When the dynasty changes, people do not change their nationality. But the case is complicated, because "citizenship" does not seem like the right concept for that place and time. For more modern usage, I find the wide use of Weimar Republic (Q41304) in P27 a bit puzzling. --Zolo (talk) 04:38, 14 September 2015 (UTC)
Zolo: Why can't it be a state and a dynasty at the same time? Didn't they treat the state as more less the personal property of the emperor back then? Who says that when the dynasty changes the nationality doesn't change? Remember those people were not citizens of a state they were subjects of an emperor. I agree that there are subtle differences that may need to be modelled here but there are also broad brush statements and classifications that need to be made while we are waiting for the subtle modelling to be done and Q7313 is definitely the item for the state that Wu Zhen lived in even if that item has an English label of "Yuan dynasty". Don't confuse the label with the underlying concept. Joe Filceolaire (talk) 19:25, 15 September 2015 (UTC)
@Filceolaire: do you mean that every Chinese person should be linked to the dynasty item of it's time of living ? this is mad.
if this must be done for China, it should also be done for every other kingdom or empire in the world, which mean that French people would not be French people but citizen of King X, and so on for every country ?
country of citizenship (P27) is for the Country or Empire, or Kingdom that the King or Dynasty ruled, not for the name of the Dynasty or King !! --Hsarrazin (talk) 20:26, 15 September 2015 (UTC)
@Filceolaire: (edit conflict) That the state is the property of the Emperor does not mean we can mix the item about the owner and the item about the property. We could probably go on about the particulars of China. Supporting the 'dynasty equivalent to state' view, we can notice that in times of political fragmentation, kingdoms are named afer their ruling families. On the other hand there is a sense that China has a continuing existence, even when the Mandate of Heaven is handed over to a different family.
I think we need to have more general approach here however. In ancient times, there is often the idea that things revolve around personal bounds rather than abstract states (X is a "subject of sovereign Y" rather than a "citizen of state Z"). Still, for Wikidata purpose, I think we have to differentiate dynasties -sequences of people- from spatial entities. Those are disjoint classes calling for different properties. Louis VI of France (Q165883) member of (P463) Capetian dynasty (Q179544) sounds ok, as does Louis VI of France (Q165883) country of citizenship (P27) France, but you can't swap them around.
Now, a specific issue about China is that historiography is typically divided according to dynasties, and, as a result, Wikipedia article en:Yuan dynasty is a much about China under the Yuan Dynasty as about the dynasty itself. Still, in Wikidata, I think we mustn't equate the two things. "citizenship: China under the Yuan Dynasty" might be ok, but "citizenship: Yuan Dynasty" is not. --Zolo (talk) 07:09, 16 September 2015 (UTC)
Hsarrazin, Zolo: There has been a lot of talk about Queen Elizabeth II recently and I learned that she doesn't have a passport. That is because she isn't a citizen of the UK - no one is they are all subjects of the queen except for QEII herself! The whole idea of citizenship is not really applicable more than a couple of hundred years back. I know that the idea of china is very clear today but I wonder how far back that idea goes and how much of modern china it extends to? The English and Irish crowns were united in 1800. The English and Scottish crowns in 1707. In 1921 part of Ireland got independence. There is a case for saying these countries changed on all those dates (the official name certainly did) and that is just one small country over three hundred years. China is much bigger and has a much longer history.
Louis VI of France (Q165883) country of citizenship (P27) France is just wrong. He was not a citizen of France. No one was till the revolution. Calling yourself a citizen was a revolutionary act back then. There is a case for saying that we need separate items for France under the revolution, under Napoleon, Under Louis Napoleon, Under the republic. Before the revolution there is reasonable case for a continuity under the King of France back to when Joan of Arc kicked out the English but before that there is a case for separate citizenship of Burgundy, English France etc. If you lived in Calais before 1558 you were English.
I know that tying people to countries is a useful thing to do and that P27 seems like a good property to use but citizenship is not as simple as the immigration authorities would have you believe and you should probably consider using 'country (P17)' or 'residence (P551)' instead in some cases. Joe Filceolaire (talk) 15:24, 18 September 2015 (UTC)
country (P17) is defined as to be used only for places, not for people :/, incompatible with human (Q5)
country of origin (P495) is too...
residence (P551) tells nothing about where the person came from… only where s/he lived ; for people who travelled a lot, or lived in numerous places this is not ok at all… there are MILLIONS of people who live in a country but have another nationality (not even speaking of DP people crisis)…
If country of citizenship (P27) is not to be used with human (Q5) older than 19th century, it should be defined as so, and explicitly explained, which is not the case on the property's page, neither in it's discussion initial frame where I read (country to which he/she belongs).
I agree that citizenship is a very historically connoted term. Maybe it should not have been chosen in the first place, as "homeland/patrie/heimat" is a very important property to define any human (Q5), and this is really becoming a mess for a lot of people… 
why not expand the definition of country of citizenship (P27) form "country of citizenship" (which is very 19-21 century term) to "country where the person is/was from" or "country to which he/she belongs", without linking it to citizenship ?, --Hsarrazin (talk) 08:44, 19 September 2015 (UTC)
the number of person who change from their "homeland" country to another citizenship is relatively rare (compared to the number of person who don't), and if there was a change of "citizenship", then a specific property could be used, with dates ?
--Hsarrazin (talk) 08:44, 19 September 2015 (UTC)
or maybe country of origin (P495) could be adapted to human (Q5), for patrie/homeland, thus avoiding the "citizenship" problem ? --Hsarrazin (talk) 10:42, 19 September 2015 (UTC)

editing interwiki - directed to "Set a sitelink"

I try to change the interwiki linking for swwiki at Q327954. I keep being redirected to "Set a sitelink". I had it before, but not recurring, Now I cannot continue, Anybody knows the problem? Kipala (talk) 08:06, 19 September 2015 (UTC)

I tried now 2 hours later - I could change, the sitelink page did not pop up. What is the problem? Server overload? script? Kipala (talk) 10:32, 19 September 2015 (UTC)
@Kipala: : please beware not to remove other people's edits when conflict-editing. I re-added your contribution, without erasing mine earlier. See history, and please be careful :) --Hsarrazin (talk) 10:44, 19 September 2015 (UTC)
I noticed the same problem with link adding and label editing for the past 2-3 weeks. It is not permanent, but very annoying :( --Hsarrazin (talk) 10:45, 19 September 2015 (UTC)
When the page looks like it's completely loaded, there could still be some loading going on in the background. If you try to edit a sitelink, the interface isn't ready yet and you are redirected to the backup interface (which is "Set a sitelink"). Only advice: be more patient. Mbch331 (talk) 16:12, 19 September 2015 (UTC)

Errors VIAF/Wikidata (again)

I tried to ask to the Italian Bar, but no answer. So the problem is the links to VIAF to some Italian buildings are totally wrong, and I don't know how to correct this, so I am leaving this note.

  • Palazzo Alberti Q16163465 on Wikidiata is a palace in Florence, on Viaf in Sansepolcro
  • Palazzo Bardi-Serzelli Q16585968, on Viaf is actually palazzo Busini Bardi Q3889596
  • Palazzo Bargagli Q16585972, on Wikidata is a palace in Florence, on Viaf in Siena (would be Q3889601)
  • Palazzo Mancini Q16586227, on wikidata is a palace in Florence, on Viaf in Roma (would be Q3360896)
  • Palazzo Strigelli Q16586343, on wikidata is a palace in Florence, on Viaf in Palazzuolo sul Senio
  • Palazzo Bonaparte Q19721646, on wikidata is a palace in Florence, on Viaf in Rome (would be Q3889693)
  • Via Garibaldi Q16620946, on Wikidata is a street in Siena, on Viaf in Genoa (would be Q2343106)
  • Collegio degli Scolopi Q19984666, on Wikidata is a palace in Florence, on Viaf in Pieve di Cento e Urbino
  • Museo dell'Opera Metropolitana del Duomo Q869130, on wikidata in a museum in Siena, on Viaf in Perugia (would be Q3868002)

--Sailko (talk) 21:23, 14 August 2015 (UTC)

So they are not well identified. The obvious solution remove the VIAF identification for now. GerardM (talk) 06:42, 12 September 2015 (UTC)
There are a lot of wrong identifications at the side of VIAF. I still don't know why a bot imported them all without checking. Sjoerd de Bruin (talk) 12:11, 12 September 2015 (UTC)
I am glad they were imported. Yes, there may be errors, but now they are checked and otherwise nobody would bother. Thanks, GerardM (talk) 12:41, 12 September 2015 (UTC)
Hi, your messages are so cool, but I asked if anybody could fix these errors as I don't know how... they seem that VIAF has imported the errors as well and I don't want to fix them and have a bot revert all of them, then. I provided this list of error twice, and they are all still there. After a week the discussion is trashed, so I am asking, can anybody help? Should I move this discussion somewhere else? Thank you. --Sailko (talk) 21:09, 19 September 2015 (UTC)
@Sailko, does this help: Wikidata:Tools/User_scripts#authority_control.js? (User_talk:Magnus_Manske/authority_control.js) It doesn't really work for me. --Atlasowa (talk) 21:41, 20 September 2015 (UTC)

Template or module to say whether an item is a redirect ?

Does anyone know if there exists a template or a module, eg {{IsRedirect}}, to say whether a particular Q-number points to a redirect ?

eg for Aldephonse Alexandre Félix du Jardin (Q2832278), template {{IsRedirect|Q2832278}} would return a "1" (or "Q2640437"), because that item is a redirect to Aldephonse Alexandre Félix du Jardin (Q2640437); but for Q2640437 it would just return a blank.

I wasn't sure how or where to look to see whether such a thing already existed. Jheald (talk) 09:12, 20 September 2015 (UTC)

Getting an entity via redirect, the target is loaded including its id. Comparing this id with the provided one...
local entity = mw.wikibase.getEntityObject(id)
if entity.id ~= id then isRedirect = true end
can tell us the result. Matěj Suchánek (talk) 09:27, 20 September 2015 (UTC)
Nice! Thank you, Jheald (talk) 10:05, 20 September 2015 (UTC)
Actually, that code does not work. I tested it in the debug console and it did confirm my suspicion. Use this code instead:
return mw.title.new(id).isRedirect

--Snaevar (talk) 22:00, 20 September 2015 (UTC)

It works if you expand my example a bit (as I have just tested), but it doesn't mind. Your example is better, thanks. Matěj Suchánek (talk) 05:35, 21 September 2015 (UTC)

Is wikidata malfunctioning on EN WP?

Please see discussion at wikipedia:Talk:2015_Thalys_train_attack#Wikidata_language_links for nonfunctioning links in language sidebar. Mathglot (talk) 22:50, 20 September 2015 (UTC)

Seems to be working now, but I’ve had the problem before on other pages. In my experience it can usually be fixed by purging the page (add ?action=purge to the URL). Hopefully someone else has more info… —Galaktos (talk) 23:55, 20 September 2015 (UTC)
Yes, just came back to say it resolved spontaneously, unless someone else saw this and fixed it? Thanks for the ?action=purge tip! Mathglot (talk) 00:32, 21 September 2015 (UTC)

Person x Human

Hello,

I've seen some outdated discussions on the two classes: "Person" (https://www.wikidata.org/wiki/Q215627) and "Human" (https://www.wikidata.org/wiki/Q5) regarding the need for two different classes.

The discussion is very confusing. It is still not clear to me the difference or when should I use one or the other.

I suggest adding this to the F.A.Q. or as commentary to the classes

Sorry if this is a duplicate question, if so, please direct me to the correct source as I was unable to find it.  – The preceding unsigned comment was added by Bernardofbbraga (talk • contribs).

You should generally use human if you are talking about individuals. person is a more abstract concept, and human is a subclass of person. Note for example corporation is in the subclass tree under person but generally not considered human. ArthurPSmith (talk) 19:01, 14 September 2015 (UTC)
Wikidata uses <instance of:human> for living and dead humans. Person is a wider class that also includes fictional humans and non-human fictional characters. The distinction is very useful in classification on Wikidata and we are not likely to abandon it. In practice you should almost never use person (Q215627); instead you should use one of it's many subclasses. Joe Filceolaire (talk) 18:32, 15 September 2015 (UTC)
Once "person" was used also for personages (fictional humans), yes, and it was quite convenient. Another possibility is to use person (Q215627) for some animals which have personality (e.g. Koko (Q1348219)). But now both links are cut, and distinction between the two classes are not clear. --Infovarius (talk) 10:00, 21 September 2015 (UTC)
What's sure about it is that this is a Chimp. It's likely that any chimp and a lot of other kind of animals have personality as well, and the line depends a lot of what we mean by personality, so the situation as it is pretty good. We rely on objective hence managable stuffs. author  TomT0m / talk page 10:54, 21 September 2015 (UTC)

Ranges of estimate and others

A few questions from a newcomer, about how certain kinds of data should be entered:

  1. A property of an item is estimated as "XXX to YYY", or "lower estimate XXX, upper estimate YYY", or has nuanced or split estimates. Examples:
    Tsunami -> Deaths -> estimated at 2000-2600, but most reports assess it as closer to 2500
    Tsunami -> Deaths -> estimated at about 30, comprising 10 from town A and about 20 from town B
    Tsunami -> Deaths -> estimated at about 30, including 17 children under 5
  2. "Commonly used estimates of a value". Examples:
    Pi -> commonly used estimate -> 3.14
    Pi -> commonly used estimate -> 22/7
    Speed of light -> commonly used estimate -> 3 x 108 m/sec)
  3. An event occurred which was related to a specific object, structure, person, discovery, etc, not a physical location. Examples:
    Discovery of Higgs Boson -> discovered at -> Large Hadron Collider
    9/11 attacks -> Happened at -> Twin towers
    Stux worm failure of centrifuges -> took place at facility -> nuclear facilities of Iran
    Person -> infection -> where transmitted/caught -> their dentist)
  4. Is there a way to enter free-text qualifiers, such as "reason-for-value" or "explanation-for-value", or isn't this done, if not, what is preferred? Example:
    Speed of light -> value ->
    (A) {2.2 x 108 m/sec; date -> 1675; Person-by -> Huygens; basis-of-value -> Measurement of moons of Jupiter}
    (B) {2.98 x 108 m/sec; date -> 1862; Person-by -> Foucault; basis-of-value -> Rotating mirror}
    (C) {2.998 x 108 m/sec; date -> 1926; Person-by -> Albert Michaelson; basis-of-value -> Michelson–Morley experiment}

Thanks! FT2 (talk) 11:20, 21 September 2015 (UTC)

My best guesses:
  1. multiple statements with different references
  2. no idea, not sure if this needs to be recorded
  3. location (P276) is a very general property and would IMO appply to the first three cases
  4. free-text qualifiers would be useless since you couldn’t do anything with them (an automated system, or someone who doesn’t speak whatever language was entered, would have no idea what the value means); for your examples, determination method (P459) —Galaktos (talk) 11:47, 21 September 2015 (UTC)
Tsunami deaths.
If there are multiple estimates then list each of them, with the uncertainty and reference that applies. Mark the best estimate as "preferred".
<number of deaths (P1120):30+-5>(preferred); <number of deaths (P1120):10>(applies to part (P518):'town A'); <{P|1120}}:20>(applies to part (P518):'town B').
<number of deaths (P1120):30+-5>(preferred); <number of deaths (P1120):17+-1>(applies to part (P518):'children under 5').
Not sure we even want to record "commonly used estimates'. Once we make the data machine readable no one need ever use these estimates ever again.
The Large Hadron Collider and the Twin Towers are locations with known coordinates. Iran nuclear facilities have a location (even if it is unknown) - use 'location (P276)'.
The original proposal for Wikidata did actually talk about having a 'comment' property but the editors here have resisted creating it because it will break the 'internationalisation' of wikidata. So far we have always found a way to do the job with properties - like 'determination method (P459)' - and by creating some carefully selected new items. Joe Filceolaire (talk) 12:59, 21 September 2015 (UTC)

Counties in administrative entities

It seems to me excessive to duplicate country in each administrative entity. Country could be deduced from higher level entities following Property:P131. Situation is very similar to biological classification. --EugeneZelenko (talk) 19:42, 19 September 2015 (UTC)

country (P17) is one of those funny properties that seem completely obvious to beginners but the deeper you go into the structured data the more dubious it seems. On the one hand it isn't really clear what it means in every case - what is the 'country' of a film? - but on the other hand it is information that gets put in an awful lot of descriptions and disambiguation notes so having a country (P17) claim will be very useful in automatically creating descriptions, when we start doing that. I say keep it. It does no harm and will get used for cool little toys. More serious users can always ignore it. Joe Filceolaire (talk) 13:13, 21 September 2015 (UTC)
What about historical changes when one AE can be in different countries in different periods? E.g. "district A" was in "region B" of "country C", but then B was divided and part of it went to "country D" and part stayed at C. P17 with time qualifiers helps us to know, in which periods A was in which country. --Infovarius (talk) 13:55, 21 September 2015 (UTC)
Yes indeed. country (P17) should have start date/end date qualifiers in those cases but the same qualifiers should also be included in the located in the administrative territorial entity (P131) statements, along with replaced/replaced by qualifiers so that the located in the administrative territorial entity (P131) statements give a complete picture of what happened. country (P17) is just for fun, for newbies to play with. Joe Filceolaire (talk) 00:21, 22 September 2015 (UTC)

guidebook for a fictional place

clicking on "random item", as I do, I came accross House of Steel: The Honorverse Companion (Q16386120). This is a guidebook to a fictional universe. It's not a novel - as it's not a story. I marked it as <instance of:guide book> but it probably needs something that says "fiction" too. Is there an existing item for this? What new item class should we create for works like this? Joe Filceolaire (talk) 12:30, 21 September 2015 (UTC)

This sounds like a good item to have a takes place in fictional universe (P1434) statement. —Galaktos (talk) 14:16, 21 September 2015 (UTC)
Looking at the results of the query most connected classes for takes place in fictional universe (P1434) (follow link, then hit execute), it would appear that we don't currently have a class to put items like this is, at least not one that is currently being used by items that have a takes place in fictional universe (P1434) statement.
It might be nice to have classes to distinguish fictional guidebooks about fictional places -- ie ones such as Fantastic Beasts and Where to Find Them (Q194424) or the Guardian's famous guide to San Serriffe (Q7415312) or the History of Castrovalva, that either add new elements to the fictional universes, or are part of the fictional narrative in themselves -- from primarily non-fiction works, which seek to document aspects of an existing fiction (and/or creative process), such as the work you are talking about? Jheald (talk) 16:03, 21 September 2015 (UTC)
I agree we need separate items for books written in-universe and those written as non-fiction works. Maybe fictional guidebooks and guidebooks? There are actually a lot of fictional guidebooks out there. It's actually, now I think about it, one of my favourite genres (Guy who like to edit wikidata likes to read guidebooks. What a surprise.) Joe Filceolaire (talk) 23:59, 21 September 2015 (UTC)

StrepHit IEG proposal: call for support

Dear all,

The StrepHit IEG proposal is now pretty much complete:
m:Grants:IEG/StrepHit:_Wikidata_Statements_Validation_via_References

We have already received support and feedback, but you are the most relevant community and the project needs your specific help.

Your voice is vital and it can be heard on the project page in multiple ways. If you:

  1. like the idea, please click on the "endorse" blue button;
  2. want to get involved, please click on the "join" blue button;
  3. share your thoughts, please click on the "give feedback" link.

Looking forward to your updates. Cheers! --Hjfocs (talk) 15:57, 21 September 2015 (UTC)

@Hjfocs: The name of the project does rather unfortunately sound as if it is allied to the rather unpleasant bug Streptococcus (Q190161), known as "Strep" for short. A rename might be in order? Jheald (talk) 16:26, 21 September 2015 (UTC)
You got the pun, @Jheald! The project solution is the serum (cf. its image in the proposal summary box, rather tiny by the way). :-) --Hjfocs (talk) 16:47, 21 September 2015 (UTC)
Well I'm not sure the pun is a good idea. (Just as I'm not sure that "Listeria" was necessarily the best name for Magnus's outstanding new tool). Association with something unpleasant can be somewhat dissuasive of people wanting to get involved or find out about it. IMO. Jheald (talk) 17:01, 21 September 2015 (UTC)
If you say something to this kind of guys, they will reply by doing it more ... /o\ author  TomT0m / talk page 17:06, 21 September 2015 (UTC)
@Jheald: I understand your point. If the project name is disturbing for you, I'm open to hear more pleasant alternatives! Of course, the pun was not intended to offend or discriminate anyone. --Hjfocs (talk) 19:00, 21 September 2015 (UTC)
P.S.: to be honest, some other bacteria may feel disregarded, but none of them seem to have complained so far.

Removing sitelinks on article deletion

I was using Duplicity to clean up Wikipedia articles without sitelinks, and I noticed with Q394867, that the sitelink was removed when the article was deleted. However, it was only deleted for a few minutes in order to impose semi-protection. I've seen a couple of other examples. I was wondering if there was a way of tracking this, and preventing it, or at least correcting it. It seems if sitelinks are removed when the article they point to is deleted, that should be reversed if the article is recreated shortly after (within 15 minutes). Has anyone looked into this before? Silverfish (talk) 00:38, 20 September 2015 (UTC)

It is a known problem that people delete articles and restore them again shortly after for various reasons. Semantically it's correct to unlink articles in these cases, yet I see that it would come in handy to have an easy way to restore these links. Relevant tickets for this topic: T75908, T56162. Cheers, Hoo man (talk) 09:31, 20 September 2015 (UTC)
I did an analysis of page deletions during the last 30 days. In total, 12,874 pages in wikipedia or in a sister project were deleted. Out of these, 511 pages were restored up to now but 310 pages are not yet again connected to Wikidata. 51 pages were restored within 15 minutes, 107 pages within a day. I think we should make users aware to readd the sitelink when restoring pages. -Pasleim (talk) 16:20, 21 September 2015 (UTC)
What is the impact of history merges at Wikipedia? Sample: es:Wikipedia:Tablón de anuncios de los bibliotecarios/Portal/Archivo/Fusión de historiales/Actual
Do we loose the link? --- Jura 16:09, 22 September 2015 (UTC)

How to get rid of redirects (in sitelinks)

Wikidata:Database reports/identical birth and death dates includes pairs of items where an article at a Wikipedia plus a redirect at the same Wikipedia have items, both with dates of birth and dates of deaths. As we don't need two items about the same person, the redirect can be deleted and the items merged.

This is likely only the tip of an iceberg. While redirects can't be added, pages that became redirects currently stay.

I think we need to find a solution to remove redirects when they lead to two separate items about the same. --- Jura 01:24, 20 September 2015 (UTC)

However, when you try to merge two items provided that one of them holds sitelink redirected to the one in the second item, merging them using Special:MergeItems is successful (not yet with the gadget). Matěj Suchánek (talk) 08:53, 20 September 2015 (UTC)
That's helpful! As Special:MergeItems fails on many other things, I assumed that didn't work either. --- Jura 09:30, 20 September 2015 (UTC)
I worked on this before (Wikidata:Requests for permissions/Bot/Hoo Bot 2) but someone complained about my approach, so I stopped resolving redirects as mentioned there. I wonder how to best (semi) automatize that. Cheers, Hoo man (talk) 08:55, 20 September 2015 (UTC)
@Jura1: It does seem, more generally, that having done a big import of items from a particular language wikipedia, we're not really keeping an eye on what happens on that wikipedia subsequently -- eg what new articles are created, what old articles are turned into redirects, etc etc. This is something it would be good to think about more systematically.
One complication is that sitelinks are sometimes deliberately made to redirects (even deliberately circumventing the sitelink-to-redirect restrictions) as a way to get around the Bonnie-and-Clyde problem. Not officially encouraged, but it does work. So there are some sitelinks-to-redirect which it might be worth keeping (though some other Wikidata people don't like the idea, and would prefer imperfectly-matched searches on wiki X to be sent eg to Reasonator, rather than to a local redirect). Jheald (talk) 09:05, 20 September 2015 (UTC)

Maybe there is something malfunctioning (or incorrect user input): at sv:Erik Söderberg the page was moved, but Q16650450 doesn't show the page move, resulting in a new item being created for the article at Q19976106. I came across several similar cases for svwiki. --- Jura 03:37, 21 September 2015 (UTC)

It seems to be a bug that was fixed in the meantime: https://phabricator.wikimedia.org/T92789 --- Jura 07:24, 21 September 2015 (UTC)
Supposedly all pagemoves in at least a month from the wikis mentioned in the bug didn't get processed. As I found one from cswiki, I suppose other wikis may be concerned as well. --- Jura 11:07, 21 September 2015 (UTC)

Here is a current count of redirects in sitelinks:

  • cs 613
  • ja 2053
  • sv 2332
  • pl 2923
  • es 3478
  • nl 3637
  • ru 5357
  • fr 7126
  • en 45575

(selected wikis only) --- Jura 12:01, 21 September 2015 (UTC)

There was a css code that allowed to directly see redirects as such. 

.mw-redirect { font-style:italic;color: #006633;}.mw-redirect:visited { color: #009900;}.mw-redirect:hover { color: #990000;}.mw-redirect:active { color: #990000;}

It still works for redirected items (merged), but not anymore for wikilinks. If it could be fixed, that would be a great help. --Hsarrazin (talk) 21:18, 21 September 2015 (UTC)

It could make it easier to visualize them.
The solution mentioned by Matěj Suchánek helped clean-up some of the entries on (the limited scope) Database reports/identical birth and death dates.
I think it would be worth doing the bot job mentioned by Hoo man for redirects on items affected by the bug T92789.
It wouldn't solve the bug for sv:Erik Söderberg-like pagemoves, meaning we have many sitelinks pointing to the incorrect items. I wonder if there are more items affected by this than the duplication through redirects. If yes, we need to address this at some point. --- Jura 11:21, 22 September 2015 (UTC)

Understanding SPARQL

I have just finished the first draft of an "Understanding SPARQL" section at

Wikidata:SPARQL_query_service/queries#Understanding_SPARQL

to try to introduce SPARQL quickly for people who already know Wikidata.

My hope is that after these introductory paragraphs, experienced Wikidata editors should find it no problem to then easily follow the further examples at mw:Wikibase/Indexing/SPARQL_Query_Examples.

I would be very grateful if people would consider looking over it and giving your thoughts -- is it helpful? Is it easy enough to follow? What is not good, and needs to be written again? -- etc. User:GerardM has already given some very useful feedback, encouraging me to try to make it start a bit more concretely, emphasise the query editor more and sooner, and think if there are any ways I can come up with to make it more visual. But he also said he loved it for being a great start :-)

I would very much welcome all the thoughts (and improvements, and rewrites) people can give, because I do think this is an incredibly powerful new resource, and we need to find ways to make it as easy as possible to introduce editors into using it. The page also includes some thoughts on how query-text for SPARQL queries can currently best be saved on wiki pages; and an introduction to some of the templated queries in Category:Template WDQS.

Please everyone do get involved and own the page, and add anything to it that would be informative or instructive. Jheald (talk) 10:55, 22 September 2015 (UTC)

Updated. I've now added a new, more concrete section at the start to try to address Gerard's spot-on comments. Jheald (talk) 17:16, 22 September 2015 (UTC)
vs.
You win! :-) --Atlasowa (talk) 11:16, 22 September 2015 (UTC)
@Atlasowa: Wow! That's really cool! Who made that? Jheald (talk) 11:43, 22 September 2015 (UTC)
It IS cool! :-) It's based on http://askplatyp.us (This query engine is the result of the Master's degree project of seven students), see also [16]
--Atlasowa (talk) 12:39, 22 September 2015 (UTC)

Wikibooks has data access

Hey everyone :)

We just enabled data access for Wikibooks as previously announced. Please welcome our newest sister and keep an eye on Wikidata:Wikibooks in case there are questions.

Cheers --Lydia Pintscher (WMDE) (talk) 15:52, 22 September 2015 (UTC)

Wikidata weekly summary #176

 
Here's your quick overview of what has been happening around Wikidata over the last week.
At first I thought "StrepHit" had something to do with streptococcal pharyngitis (Q840143). lol --AmaryllisGardener talk 16:08, 22 September 2015 (UTC)

P1705 and unknown language

Hello, I wanted to add the property native label (P1705) to Sirionó (Q3027953). Problem is "sirionó" is not available in the languages proposed. Are there only the languages used in any project of the Wikimedia foundation available here? How can we add other languages such as Sirionó? Pamputt (talk) 15:57, 20 September 2015 (UTC)

There was a plan to implement support for "unsupported" languages via some special language code and qualifier language of work or name (P407). This is probable still unresolved. Paweł Ziemian (talk) 18:50, 20 September 2015 (UTC)
Thank you for the information. Is there already a bug report opened somewhere for this? Pamputt (talk) 20:50, 20 September 2015 (UTC)
The language committee has agreed that Wikidata should be open for any and all recognised (ISO-639-3) language. So when a language exists, it should be a matter of asking for the addition of that language. Thanks, GerardM (talk) 10:59, 21 September 2015 (UTC)
Take a look at [17] how much difficult it seems to add a new language - 6 months and counting... Ahoerstemeier (talk) 15:53, 22 September 2015 (UTC)
Thank you for the link. I created a new task specifically for the Siriono language. Pamputt (talk) 20:40, 22 September 2015 (UTC)

Q21000000

Is created today. Congrats. By the way, long time no see:) --DangSunM (talk) 23:00, 22 September 2015 (UTC)

See also Special:Statistics. Joe Filceolaire (talk) 03:55, 23 September 2015 (UTC)

Using female (Q6581072) as a unit

See Wikidata:Property_proposal/Unsorted#proportion_of_population where I have proposed that we make a "proportion of population" property with datatype "number with units" so we can say a population is 0.512 female; 0.481 male; 0.23 Catholic; 0.01 asian etc. without needing a qualifier. This seems like a good idea to me.

@Lydia Pintscher (WMDE) : does this raise any technical issues?

Maybe we should change population (P1082) to this datatype too so we can say Population:1056 human; 503 female; 480 male; 806 white; 109 black; 305 catholic; 3 jedi etc. Joe Filceolaire (talk) 03:49, 23 September 2015 (UTC)

How do you select all jedi? --- Jura 08:20, 23 September 2015 (UTC)

There's no need to change the datatype of population (P1082), as for all quantity properties the unit is optional, and can be any item. So it's no technical issue, it's just how we want to use the property. It would however be nice if there would be a way to limit the unit for a quantity property, e.g. with a whitelist of possible units on the property - that'd prevent things like a area (P2046) without unit or with unit metre (Q11573). Ahoerstemeier (talk) 08:31, 23 September 2015 (UTC)

If there is no technical issue, can you explain how the selection is to be done? --- Jura 08:41, 23 September 2015 (UTC)
Same as when adding an area (P2046) - when adding/changing a quantity value a small edit box shows underneath to select the unit. Ahoerstemeier (talk) 09:59, 23 September 2015 (UTC)
I prefer to use applies to part (P518) as a qualifier.
population 1018 
population 503 applies to part: Women
population 2 applies to part: Jedi
population 1 applies to part: Jedi, applies to part: Women (for female jedis)
-- Innocent bystander (talk) 08:55, 23 September 2015 (UTC)
I really don't like this model. We might want to express the number of cats in the area for some reason, (for example I heard recently that the number of pet cats in an area is responsible for a drop in the number of birds). One smell test that it does not pass is that the main snak does not mean anything by itself for example. I would prefer some patterns like an item "living organisms in <area>", linked to the item of the area itself, which could have {{living organisms in <area>|has part|woman|number|10000}} ; {{living organisms in <area>|has part|cats|number|20000}} ; ... author  TomT0m / talk page 09:03, 23 September 2015 (UTC)
Then you maybe need units like "female black pet indoor cats with age 5-6 years"? A combination of qualifiers looks easier to me. -- Innocent bystander (talk) 09:08, 23 September 2015 (UTC)
@Innocent bystander: I think you did not understand my proposal. The pattern <has part : ?something> qualified with number implies that the number is a raw number, and that the number is the number of ?something, where something can be anything : men, woman, young people, wheels, ... So this does not require any unit as the number is just a count of ?something. author  TomT0m / talk page 09:26, 23 September 2015 (UTC)
TomT0m: What is the subject of these claims - the page they are on? Isn't it the area? So that area can have the claim <population:23, unit=cats>. Joe Filceolaire (talk) 10:53, 23 September 2015 (UTC)
I have conceptual troubles considering "cat" as a unit of anything, even if technically this seems elegant. As I said, this claim would apply to an item whod coulf be called "life in the area", for example "living organisms in Paris" or "Paris ecosystem" for example, would be made made of cats, birds, and that would be written
⟨ Paris ecosystem ⟩ has parts Search ⟨ cats ⟩
quantity (P1114)   ⟨  100 000 ⟩
for example.
To explain my conceptual problem with this, I think we must think of what is a measure, for example read the articles unit of measurement (Q47574). A unit is a conventional way of measuring something. But to measure something, we must first explain what we want to measure, for example to measure a length it's natural to measure in meters... if we want to measure the cat population, it's a natural unit to measure it in number of cats, of course. But ... here we it's implicit that by giving the unit we give the nature of what we want to measure. Which can legitimately make unconfortable. author  TomT0m / talk page 11:21, 23 September 2015 (UTC)

volumes with different publication dates

For Q19096589, I want to add information to clarify that 1910 was the publication date for Volume I, and that 1913 was the publication date for Volume II, but I cannot add that information when I try to do so. --EncycloPetey (talk) 04:17, 23 September 2015 (UTC)

I think you will have to create separate items for "Volume I" and for "Volume II" plus others for the original Greek works. use edition or translation of (P629) and has edition or translation (P747) to link between the items for the works and the items for the translations. Joe Filceolaire (talk) 10:45, 23 September 2015 (UTC)
Jup, separate items is mostly the way to go. Sjoerd de Bruin (talk) 11:41, 23 September 2015 (UTC)
Separate items for each volume? That's a hassle, since I'll need a means of connecting the main work's item, the volumes items, and the section items all connected, while also allowing for interwiki linking and such. I knew I would need to create separate items (eventually) for each translated play included in the volume, and I suppose that means that Wikidata will also have separate items for every individual article in the 1911 Encyclopaedia Brittanica, which will not be fun to coordinate since it too has the problem of multiple volumes and sections, just on a much grander scale. --EncycloPetey (talk) 13:27, 23 September 2015 (UTC)

WIkidata's Pillars

I think we have at least one pillar :

Let's look at the whole quote, TomT0m:
  • "... This makes it even more important to remember that Wikidata is not about truth, but about collecting referenced statements in a secondary database. The criterion for inclusion should not be veracity, but verifiability – a policy that has served Wikipedia very well."
That was in June 2013, in July there was Wikidata:Requests for comment/Sourcing requirements for bots and the result in August 2013 was:
  • "As a whole, the idea of requiring sources is rejected. Sources are still recommended though, and bots should try to use better/reliable sources when possible. Legoktm (talk) 06:04, 23 August 2013 (UTC)"
Looks like wikidata is neither about truth, nor about collecting referenced statements but all about hoarding data. --Atlasowa (talk) 11:31, 23 September 2015 (UTC)
@Atlasowa: Just as Verifiability is a pillar of Wikipédia but we don't delete immediately every unsourced or not sourced enough article. Nothing new or contradictory here. author  TomT0m / talk page 11:36, 23 September 2015 (UTC)

Anyone wanting to learn Pywikibot?

I am looking for people that would like to run a Pywikibot on Wikidata. The less you know about programming the better. - My goal is to write a super-simple to follow guide that teaches the various skills. But I need people to ask the important beginner-questions, check my English, (etc..). I started separate tutorial, so I can shave away all the old stuff (e.g. python2.7) and focus on one clear to follow learning path: Wikidata:Pywikibot - Python 3 Tutorial. --Tobias1984 (talk) 10:31, 18 September 2015 (UTC)

Well, I have just written my first (ever) Python script and am waiting for approval. Your tutorial helps me now, although it's drafted. Matěj Suchánek (talk) 11:44, 18 September 2015 (UTC)
@Matěj Suchánek: Perfect. Just post any question or requests you have and we can work on them together. I looked at your script and we could turn that into a chapter of the tutorial. --Tobias1984 (talk) 12:00, 18 September 2015 (UTC)
I am now stuck on adding quantities. I haven't found it documented anywhere and am not in the mood for looking into the resources and trying to understand what I do wrong / what is expected. This includes some other features. Matěj Suchánek (talk) 12:27, 18 September 2015 (UTC)
@Matěj Suchánek: I hope this can help you a little: Wikidata:Pywikibot_-_Python_3_Tutorial/Data_Harvest#Getting_claims. --Tobias1984 (talk) 14:16, 18 September 2015 (UTC)
Python and Pywikibot are definitely on my "to learn" list; but there are a few current jobs I want to get out of the way first, so I may need to continue soldiering on with Perl as my go-to language first for a little while longer. Jheald (talk) 14:26, 18 September 2015 (UTC)
 
BotAcademy 2015, swedish

Hi Tobias1984, Swedish Wikipedia has a Botacademy this year (and a cool logo!), maybe they can give some feedback, share their write-ups etc? (2014 by Jan Ainali?) --Atlasowa (talk) 21:55, 20 September 2015 (UTC)

@Atlasowa: That would be great! Will the botacademy use Python3 and Pywikibot-Core too? I would like to avoid the added complexity of having to explain legacy syntax and libraries. --Checkallthestrings bot (talk) 22:15, 20 September 2015 (UTC) --Tobias1984 (talk) 22:16, 20 September 2015 (UTC)
@Tobias1984: I know very little about the BotAcademy (but i like their logo :-) and the goal of educating better bots :-) ), maybe ask Jan Ainali or User:Anders Wennersten? --Atlasowa (talk) 12:20, 21 September 2015 (UTC)
Awesome wih new tutorials! We haven't decided yet the exact details of what we will go through, but we will for sure try to use as updated tools and libraries as possible. It is André Costa (WMSE) who is project manager and I am pretty sure he will happily receive input. By the way, there will be some (small) scholarships for travel to attend and the dates for it is 28-29 November. Jan Ainali (WMSE) (talk) 12:56, 21 September 2015 (UTC)
@Atlasowa, Jan Ainali (WMSE): Thanks for the information. I hope we can collaborate on the tutorials, or even merge them to reduce the workload. --Tobias1984 (talk) 14:11, 23 September 2015 (UTC)
@Atlasowa: Do they harvest wikipedias or „real sources”? --Succu (talk) 22:23, 20 September 2015 (UTC)
 
Logic Model Bot Academy
@Succu: It's a training for writing bots, not sure what they say about data sources to be used... (Wikimedia-l) Lsjbot+Bot Academy, anderswennersten Sep 16, 2014: "... we will now start what we call Bot Academy. A dozen of our experienced editors will, with the support from WMSE, learn more of running bots. First by sessions on basics, common knowledge stuff, in order for us to be able to use bot as a complement in our editing efforts. And after that we will have sessions for advanced use, taking in the learning from Lsjbot and Naskobot, in order to see if also we can find areas where we from excellent sources can generate articles." The current big bot-project on swedish WP is "all places" (sv:Användardiskussion:Lsjbot/Projekt_alla_platser, gtranslate) by LSJbot based on Geonames, which i wouldn't call a good source (user generated quality mix...), plus data copied from other Wikipedia-versions. Currently sv:Kategori:Robotskapade_geografiartiklar shows boticles for Macedonia: 10842; Nicaragua: 2706; South Sudan: 5514. See also (Wikimedia-l) LsJbot and geonames, anderswennersten, Sep 5, 2015. --Atlasowa (talk) 12:43, 21 September 2015 (UTC)
Arabic Wikipedia Bot Intro
@Tobias1984, André Costa (WMSE), Jan Ainali (WMSE): There is a Pywikibot video tutorial in arabic, ha! Awesome. --Atlasowa (talk) 12:56, 25 September 2015 (UTC)
@Atlasowa: Looks like the video explains all of the Windows and P2.7 caveats (encoding, env-variables). But I think the community should really make a push towards P3 now. The unicode strings make the handling of multi-linugual information so much easier. --Tobias1984 (talk) 14:25, 25 September 2015 (UTC)

Living vs. Date_of_death not filled in

Is there a way of distinguishing living people vs. date_of_death not filled in? Should date_of_death be filled in as "null" for living people so that we can find those people that need their death dates researched and filled in? --Richard Arthur Norton (1958- ) (talk) 00:18, 22 September 2015 (UTC)

I recall using "unknown" once where the date/year wasn't known, but I haven't done this every time because I'm also not sure. It should be blank (the property absent) if there is no death, obviously, no? "unknown" implies that there has been a death but its date isn't known, in my opinion. --BurritoBazooka (talk) 03:40, 22 September 2015 (UTC)
Date of death "unknown" (="somevalue") means we know the person died but we don't know when it happend. Date of death "novalue" means we know the person hasn't died. (more precise: we knew the person hadn't died by the time of the statement). Property absent means we don't know if the person has died or not, or we do know but nobody has added the statement yet.--Shlomo (talk) 08:07, 22 September 2015 (UTC)
I don't think there is an agreement to use "novalue" for living people. In any case, it would lead for most checks we currently do on death dates to fail. --- Jura 08:52, 22 September 2015 (UTC)
I do. You probably should addapt your checks so that they cope with it. Or do you have any other idea how to distinguish positively living people from positively death people with unknown date of death? --Shlomo (talk) 14:13, 24 September 2015 (UTC)
Any idea who wrote that and what website it comes from? --- Jura 14:17, 24 September 2015 (UTC)
Found it: Special:Diff/126530398. Not sure why it's there or what it's based on. For more recent discussion, see Property talk:P570.
BTW, Property:P1317 works for that. --- Jura 14:41, 24 September 2015 (UTC)

Is there a page that documents where Wikidata is reused outside Wikimedia?

Hi All

I can't seem to find a page on where Wikidata is reused outside Wikimedia projects, does one exist? I know Google reuse content in their knowledge graph and there are cool visualisation tools like Histropedia. Would it possible to create one in a non manual way?

Mrjohncummings (talk) 09:05, 22 September 2015 (UTC)

Mrjohncummings, start with Wikidata:Tools/External tools? --Atlasowa (talk) 14:54, 22 September 2015 (UTC)
Thanks very much John Cummings (talk) 07:55, 24 September 2015 (UTC)

How to enter a language family?

How do I say that Nyingwom (Q36753) is in the language family Savannas (Q4403672) and where do I discuss this? Subclass of, instance of, and part of all feel a bit off. Is there an existing discussion on a property for "language family" and whether it was rejected? I was trying to find it, but failed. --Denny (talk) 18:02, 23 September 2015 (UTC)

We settled on with I think. There used to be a property but it was deleted in favor to generic classification properties. author  TomT0m / talk page 18:09, 23 September 2015 (UTC)
Ah well, not sure I like the solution, but will do so. Is there a link to the decision? Thanks! --Denny (talk) 18:12, 23 September 2015 (UTC)
We had a property super language family. The deletion discussion you find here --Pasleim (talk) 18:14, 23 September 2015 (UTC)
@TomT0m: what it currently done is actually often the opposite:
with
Though it might logically equivalent, I think it sounds more sensible. Actually part of (P361) is used as well in English (Q1860). --Zolo (talk) 05:32, 24 September 2015 (UTC)
@Zolo: I don't understand, there might be mistakes in your comment, you used twice the same item in a statement. (It seems I was reading one of my comments :)) author  TomT0m / talk page 06:38, 24 September 2015 (UTC)
Uh actually, I had misread you comment, and hence miscopypasted :O. I guess I mostly agree this you that this is the best solution we have --Zolo (talk) 07:19, 24 September 2015 (UTC)
Apart from that, yes,
Visite fortuitement prolongée used to use both instance of : language and subclass of : <the language family> on the same item. I did not really agree with this, we had a chat with Pamputt. Unfortunately I just discover that (s)he was blocked indefinitely on frwiki :/ fr:Discussion_utilisatrice:Visite_fortuitement_prolongée. author  TomT0m / talk page 06:43, 24 September 2015 (UTC)

I'd just say here that I think a more specific property like "language family" would probably make more sense here, and lead to less confusion. Semantics like the one discussed here can still be layered on top of that, but the ground data would always be clear. But I am getting very tired of having this fight repeatedly. --Denny (talk) 20:48, 24 September 2015 (UTC)

@Denny: Sorry to say the thing like but we are also tired of people coming with THE good idea about properties. If you are fighting again and again, perhaps it is time to change your method: instead of proposing unrelated properties at periodic intervals, start to provide one big overview of how you want to describe a complete field of items. What WD is missing are people working in team inside Wikiprojects and trying to develop a model with a structured list of properties. If you already spoke with 2-3 contributors about your proposal before starting the proposal, you will have more chance to get their support that if you just put a proposal expecting that people will do the job of understanding how your proposal will match the needs of WD. Snipre (talk) 01:02, 26 September 2015 (UTC)
@Denny: And then we would return to the old "family/superfamily" pair ... It is very much parcimonious to have one consistent way to class things. It's always the same pattern that recurs in a lot of fields, and in the end everyone can understand Help:Classification. Plus in Wikidata it's pretty hard to make a new property accepted, way more than to get an infobox parameter accepted (for example, a property to link a science to the objects it studies can be in the pipe for acceptance for years without beeing accepted or rejected, see study of in Wikidata:Property_proposal/Natural_science which had its first birthday a few deays ago!) . So in the end, having powerful and generic solutions is giving the project oxygen. With templates like {{PropertyForThisType}} we can associate a list of property to a class which can help document the whole stuffs, and we can build things around classification generically that can benefit as is to other fields. By contrast having specific properties to do always the same thing, which mean an inflation in the classification property number with their specific guidelines, for ... what ? author  TomT0m / talk page 16:16, 26 September 2015 (UTC)

Item for unit of time "year"

Which items should be used as unit "year(s)"? year (Q577)?

From the description, it seems to be Q1092296, but its current label "Annum" doesn't really make it suitable.

Shall we create a new item? --- Jura 18:24, 23 September 2015 (UTC)

In some cases, typically those related to astronomy, the unit should be Julian year (Q217208). – The preceding unsigned comment was added by [[User:|?]] ([[User talk:|talk]] • contribs).
First we should make clear which "year" (or "annum" or whatever; the name in some particular language shouldn't be decisive) property describes which unity of time, and then we should apply the appropriate one to every particular statement. At this moment, I can see at least 3 approaches (and I assume there will be more of them...); I just can't identify them with particular items :(
  1. Year as a unit of time with a specified length (365.25 days, or 31557600 seconds?), so that 2 "years" from September 28th, 2015, 10 pm means September 28th, 2017, 10 am. Useful in physics, astronomy etc.
  2. Year as a period of time ending by the end of the day with the same description as the day of beginning of the period, so that 2 "years" from September 28th, 2015, 10 pm means September 28th, 2017, 24 pm. Useful for election periods.
  3. Year as a period of time with a specific description (e.g. 2015). Not very useful, but still possible; in some countries the duration of a copyright protection is 70 "years" (as described here) after the death of the author; it means the work of a person who died on September 28th, 2015, 10 pm will be copyrighted until December 31st, 2085, 12 pm. Alternatively, this concept may apply on periods starting and ending on other specified date than January 1st, like fiscal year starting in March or academic year starting in September.
Unless we make this clear, the discussion about which item to use for a "year" doesn't make much sense. At least for me.--Shlomo (talk) 19:30, 28 September 2015 (UTC)

Less than one week left for Individual Engagement Grant proposals!

There is less than one week left to submit Individual Engagement Grant (IEG) proposals before the September 29th deadline. If you have ideas for new tools, community-building processes, and other experimental projects that enhance the work of Wikimedia volunteers, start your proposal today! Please encourage others who have great ideas to apply as well. Support is available if you want help turning your idea into a grant request.

I JethroBT (WMF) (talk) 15:34, 24 September 2015 (UTC)

Dating relative to Big Bang

For dating "significant events" of the universe (e.g., the "start time" and "end time" of, say, the inflationary period, or the lepton epoch), using the Julian or Gregorian calendar is clearly ridiculous. We need a third option for dating events relative to the Big Bang. Or am I just not seeing how to do this with our current infrastructure? - dcljr (talk) 18:55, 24 September 2015 (UTC)

I think the closest way of doing it might be to do followed by (P156) and follows (P155). That's how items like Cretaceous (Q44626) are currently treated. --BurritoBazooka (talk) 19:56, 24 September 2015 (UTC)
We do not need a new datatype, though - it would be sufficient to have a property that means "after beginning of universe" with the quantity datatype, and the values could be "5 microseconds" or "100 Million years" etc. --Denny (talk) 20:25, 24 September 2015 (UTC)
Actually, I think at least two new properties are necessary: "time after beginning of universe" (for discrete events) and "start time after beginning of universe" (for periods of time) — "end time after beginning of universe" is not strictly necessary if we can co-opt duration (P2047) [was called "duration"] for that purpose. BTW, a similar issue arises if you want to date events in, say, human fetal development (in which case we would need to use times relative to conception — although I'm not sure how necessary that would be). I also considered a new property "start age", which would work (more or less) when dating the "Lepton epoch" in the item Universe (Q1), but wouldn't work when dating the same thing in the item lepton epoch (Q1079826) itself (because: age of what?). - dcljr (talk) 21:44, 24 September 2015 (UTC) [Nevermind: there is no such "co-opting". We would need a whole 'nother property for the length of time. - dcljr (talk) 23:01, 24 September 2015 (UTC)] [Also, I've renamed "duration" to the more appropriate "running time". - dcljr (talk) 18:15, 25 September 2015 (UTC)]

Kristiina Mäki (Q17275699)

I would like if someone could resolve this one. I don't want to edit war... Please see the item history and also User talk:Enzino. Thanks for your time, whatever you then decide to do or not. --Stryn (talk) 18:56, 24 September 2015 (UTC)

I removed all claims without references. BTW, please don't add qualifiers to human (Q5) --- Jura 19:26, 24 September 2015 (UTC)
I think it is correct now, we have country for sport (P1532) for situations like this. Sjoerd de Bruin (talk) 19:39, 24 September 2015 (UTC)

Q19085204

The DNB biographies like Q19085207, is the plan to convert these into instance of human when a Wikipedia page is made for them, or will they stay in instance of an edition and be used as a reference for the data? --Richard Arthur Norton (1958- ) (talk) 17:29, 25 September 2015 (UTC)

I would have thought they should normally be instance of (P31) biographical article (Q19389637), and maybe part of (P361) Dictionary of National Biography (Q1210343) (or some more specific part of it). They should *not* be converted into instance of (P31) human (Q5) -- start a separate item for the individual as an individual. Jheald (talk) 17:49, 25 September 2015 (UTC)
Compare for example Hobart, James (DNB00) (Q15985561) Jheald (talk) 19:08, 25 September 2015 (UTC)
When you read a Wikipedia article about them, they are exactly like this. Calling them a "biography" is weird. GerardM (talk) 07:26, 26 September 2015 (UTC)
@GerardM: The point of doing it this way is that these items were created to be sitelinked and in 1:1 correspondence with texts at Wikisource. So this is intended to be the item about the particular text. There may be many texts at Wikisource, all relating to the same person, but taken from different works. This approach allows them all to be sitelinked. It also means that a template at Wikisource can systematically use the main subject (P921) to store who the text is about. And having an item for the text is exactly what we need, if the text is to serve as a reference. So for all these reasons it's quiteuseful to have an item for the text, separate from the item for the individual. Jheald (talk) 08:19, 26 September 2015 (UTC)

Thanks for the clarification. I suppose it is the same as for "portrait of xxx" items. Many museums think their portrait painting should be the article about the person and often give the painting the name of the person. However a portrait is not the same as a person. We obviously have lots of backlog in this area, because I believe there are more portrait paintings in use to illustrate articles about people, than there are wikidata items for the portraits of those people. --Jane023 (talk) 08:31, 26 September 2015 (UTC)

I see your point. Thank you. GerardM (talk) 10:06, 26 September 2015 (UTC)
What Jheald said. When we first integrated wikisource into wikidata we looked at just linking the biographical articles into the same wikidata items as the subject of those articles. It was pointed out that one person can have multiple biographies on wikisource (a DNB entry, a Britannica entry, a Catholic Encyclopedia article etc.) so it was decided to link these wikisource articles to wikidata items about the articles rather than to items about the persons. This has the advantage that the wikidata item can have statements describing the wikisource article - what it is an edition of, author name, what it is a biography of etc. 23:45, 26 September 2015 (UTC)

Unit window

Is there a way to turn off the Unit window? It makes my life difficult... Xaris333 (talk) 15:31, 26 September 2015 (UTC)

Propably using CSS... why would you turn it off? How would you then add kilometers to length? Matěj Suchánek (talk) 16:26, 26 September 2015 (UTC)
I add only numbers, with out units. For example, [18]. The unit window shows everytime I put a number and that cost me time... (That ±1 also cost me time. Who thinks all these things; Puting number, without units, it should be the easiest, not so complicate.). Xaris333 (talk) 17:25, 26 September 2015 (UTC)
Regarding the ±1, see phabricator:T105623 where several people are trying to convince others that assumed margins of error are incorrect far more than they are correct. Your input there as an end user inconvenienced would be beneficial. Thryduulf (talk: local | en.wp | en.wikt) 17:42, 26 September 2015 (UTC)
For human users it has at least one advantage - it makes the user think about if there is a margin or he means an exact value - quite often source stating exact values omit the margins. However oddly lately a bot starts to add margins of ±1 [19] for data which is listed as exact value in the source (infobox in English WP). Besides, I doubt that the whole city is at the same elevation, making the whole datum a bit bogus. Ahoerstemeier (talk) 20:56, 27 September 2015 (UTC)
For human users it has at least one advantage - it makes the user think about if there is a margin or he means an exact value - quite often source stating exact values omit the margins My suggested separate field for uncertainty would prompt the user in the same manner without assuming the value they enter is incorrect, and without introducing errors into the data (e.g. at British Airways Flight 2276 (Q20962419) number of survivors, number of participants and number of injured are all exact figures and had to be manually correct from ±1. Where a source does not specify margins of uncertainty we need to record that as "unknown" as we cannot guess. For a city it might be an average height above sea level (±0.5 meters/feet, ±500 metres/feet or anywhere in between) or it might be the exact height above sea level of a standard reference point. Thryduulf (talk: local | en.wp | en.wikt) 21:17, 27 September 2015 (UTC)

Statistics show primary sources tool is a failure

I am happy to have learned that there are statistics for the primary sources tool after all. It shows that after all this time slightly more that one in a thousandth of its content found its way in Wikidata. Given that the quality of the content hidden in there is as good if not better than what we have in Wikidata, it is in my mind an extremely deplorable state of affairs. Particularly because it would have been much more productive to concentrate on workflows that would concentrate on the quality of Wikidata itself.

One tool that is proposed would be extremely worthwhile checking content in content in Wikidata but goes for inclusion of its data in the primary tool. Such an approach hides the technology away from us and consequently it will not realise the potential that it has.

Another fact that condemns the implementation of the primary sources tool is that it not even acknowledges the source the data came from...

In my opinion we should import everything into Wikidata that fits our criteria. The primary sources tool was a nice experiment and it deserves to be abandoned. When there is some merit after all, I love to hear it. Thanks, GerardM (talk) 16:20, 26 September 2015 (UTC)

19 165 new statements is definitely not a failure. --Jklamo (talk) 21:23, 26 September 2015 (UTC)
See Measures of Success. --Succu (talk) 21:45, 26 September 2015 (UTC)
When not even one percent, it is actually slightly over one in a thousand of all that data has been touched. How can you possibly say that it is a success? The Freebase data is as good as what we have. It is sheer discrimination that its data is largely ignored. Thanks, GerardM (talk) 07:37, 27 September 2015 (UTC)
did you GerardM involve in its success ?
I personally feel it could be better used and approved with a lighter interface. The complete refreshment of items after each approval or reject is very long and unfriendly. Does not help to work with items that have many primary source tools claims to add/remove :/ --Hsarrazin (talk) 09:20, 27 September 2015 (UTC)
I have often argued that the approach of the "primary sources tool" was flaky. I have been reassured time and again that everything will be well. Its UI is what is part of my Wikidata experience and it sucks in many ways. I have been told that this subject may be discussed and so I do. There are many arguments why this approach has failed and seriously, on a good day I can easily add as many statements as have been added from Freebase in all this time. In my honest opinion, it demonstrates clearly what failure it is. It would be polite to put it to pasture. Thanks, GerardM (talk) 11:00, 27 September 2015 (UTC)
Wow, that's less than 5 seconds per statement without a bot if you edit for all 24 hours of a day! Could you elaborate where I can get approval for performing edits this fast, which seems way past the usual rate limit and, I'd say, way past the limit at which a human can check his own edits? --Mineo (talk) 09:19, 30 September 2015 (UTC)
Yes, I do add from Freebase but I refuse to spend time on any particular statement ie I will not spend more time as for any other statement.

users copying bad examples from bots copying "imported from: English Wikipedia"

 
Wikidata References (Jan-June 2015)

Here. Looks like this is the wikidata standard source now? How often does this happen, that non-bots add "imported from: English Wikipedia"? --Atlasowa (talk) 04:26, 27 September 2015 (UTC)

Well, if the data is from a unreliable source like Wikipedia, I prefer that they use imported from Wikimedia project (P143), even if the it's done by a non-bot. I do that myself in rare occations. -- Innocent bystander (talk) 08:16, 27 September 2015 (UTC)
yes, I do that too, whenever I couldn't find on the wp page the source of the info. At least, this way, it is possible to find from which wp the info comes from (on some items wp disagree). :) --Hsarrazin (talk) 08:26, 27 September 2015 (UTC)
I've also been doing it, I assumed that it is to be done for anything where the source is the Wikimedia page(s) linked to the item. There was no documentation or warnings to imply that I should do otherwise or that it was only for bots. I suggest if this is not the purpose of this property, then documentation should be changed. Every now and then, I would go through Wikipedia pages to find more reliable sources, and add them to the item instead. I also sometimes would use "imported from: imdb" if there was already an IMDB ID linked to the item (similar with Musicbrainz). Again, the purpose of that property is not stated clearly in the first place. --BurritoBazooka (talk) 08:56, 27 September 2015 (UTC)
This is a crutch. You might use it, as long as there is nothing better, no better source. Bots might import contradicting statements from different Wkipedias, so there is a chance to see the contradiction, to get it right and to improve the wiki(s) with the wrong information. --Giftzwerg 88 (talk) 13:04, 27 September 2015 (UTC)
it would be better to be able to use stated in (P248) wp, when adding a human contributors adds it as wp says it, but I couldn't find wp's source, but there is a constraint that prevents to use stated in (P248) and forces imported from Wikimedia project (P143) :) --Hsarrazin (talk) 14:46, 27 September 2015 (UTC)

See also Property talk:P248#Receiving Warning on P248 (1 February 2015): Today, I have started getting this error message when setting a reference 'stated in': "An error occurred while saving. Your changes could not be completed. Details Warning: Use imported from (P143) instead." Sounds like a Special:AbuseFilter (but which one?) or something else? The warning shouldn't direct users to use imported from (P143) Wikipedia, but rather to use stated in with "real" external references (or, possibly reference URL (P854)). --Atlasowa (talk) 08:31, 28 September 2015 (UTC)

@Atlasowa: Special:AbuseFilter/54, see Wikidata:Project chat/Archive/2014/11#Abuse filter or Revert bot. Matěj Suchánek (talk) 09:52, 28 September 2015 (UTC)

Q18878536

@Magnus Manske: Something went wrong with this item. It has an identifier which is duplicate of another. The second one is however held by another item which seems to be the correct one. Thus the next step should be deleting the identifier in Q18878536 (everything already resolved in Mix'n'Match) but that would be problem because Charles Matthews changed the data in the item a bit. So I am not sure know how to fix this problem. Matěj Suchánek (talk) 09:23, 27 September 2015 (UTC)

Disambiguation pages and Commons category (P373)

Is it correct to add Commons category (P373) to Disambiguation items? like these cases?Yamaha5 (talk) 12:49, 27 September 2015 (UTC)

No. How can a DP have a commonscat? A DP doesn't refer to 1 subject. Mbch331 (talk) 12:57, 27 September 2015 (UTC)
So we should clean this query also many of Disambiguation items have incorrect p31. many items have wiki links combination of Disambiguation and normal pagesYamaha5 (talk) 13:05, 27 September 2015 (UTC)
Some Disambiguation pages are pure collections of things that just happen to be spelled similarly but there are also some Disambiguation pages which are a list (or class) of things that are thematically linked and should really have a 'subclass of' claim rather than 'instance of:disambiguation page' claim. As such they may well have an associated Commons Category. This applies to some of the items that Yamaha5 links to above but not to others. Joe Filceolaire (talk) 13:55, 27 September 2015 (UTC)
and there are other I found that aren't disambiguation pages at all. Joe Filceolaire (talk) 14:53, 27 September 2015 (UTC)
I said also many of Disambiguation items have incorrect p31 Yamaha5 (talk) 15:25, 27 September 2015 (UTC)

Sometimes disambiguation pages link to a commons category disambiguation page on the same word. In such cases, Commons category (P373) should be OK, isn't it?--Pere prlpz (talk) 12:01, 28 September 2015 (UTC)

How to access a single value as integer

Hi, I want to access the latest population of Germany and use it to do a calculation with it but {{#property:P1082|from=Q183}} yields: 84,358,845. Wat do I have to do and where could I read this without having to ask here? --Saehrimnir (talk) 14:11, 27 September 2015 (UTC)

You have to use Module:Wikidata (Q12069631) for better data retrieval methods. --Pasleim (talk) 15:25, 27 September 2015 (UTC)

Units not shown in example property documentation

When the example is taken from the statements of the properties and the property type is quantity with units, the unit isn't shown on the talk page (see for example Property talk:P2067). It shows the quantity without the unit. Mbch331 (talk) 16:08, 27 September 2015 (UTC)

It needs to be updated in the module (mentioned in the section above). Some things to remember:
  1. https://gerrit.wikimedia.org/r/239054 (not deployed yet)
  2. decision whether we want to use labels or quantity symbol (string) (P416) P558 (P558)
  3. correct algorithm (using existing methods)
If no one has picked this up, I will do this as soon as I have tested it in cswiki. Matěj Suchánek (talk) 19:29, 27 September 2015 (UTC)
Hm, I hadn’t heard of using quantity symbol (string) (P416) for unit symbols yet. Is there any discussion for this? ’Cause I don’t think that’s the right property – that seems to be about quantity symbols, and the current format string (“single character”) doesn’t even fit most units (“kg”). —Galaktos (talk) 21:05, 27 September 2015 (UTC)
Unless they keep inventing Unicode symbols like ㎏ and ㎢ ;) --BurritoBazooka (talk) 22:36, 27 September 2015 (UTC)
@Galaktos: Sorry, I meant P558 (P558). Matěj Suchánek (talk) 09:02, 28 September 2015 (UTC)
Ah, thanks, that makes a lot more sense. (Though perhaps we need a new property, since apparently P558 (P558) has an unsuitable data type – see talk page.) —Galaktos (talk) 10:59, 28 September 2015 (UTC)

New constraint for datatype with unit

Is it possible to create a new constraint for numeric datatype with unit to limit the number of items used as unit ? Snipre (talk) 21:29, 27 September 2015 (UTC)

@Ivan A. Krestinin: +1 (something like {{Constraint:No unit}} and {{Constraint:Unit type}}). Matěj Suchánek (talk) 08:59, 28 September 2015 (UTC)
  Done, please see {{Constraint:Units}}. — Ivan A. Krestinin (talk) 16:10, 28 September 2015 (UTC)
The first results should be available at Wikidata:Database reports/Constraint violations/P2067 tomorrow. Although there might be some violations as I don't know all the units for mass that have been used so far with mass (P2067). Mbch331 (talk) 16:29, 28 September 2015 (UTC)

List of units with associated properties

A lot of persons start using units datatype but don't know which items to select for the unit. I proposed to create a list in wikidata:Units to list all units, the quantity measured by the units and the associated properties. This should helps people to find the relation between properties and units. Snipre (talk) 21:58, 27 September 2015 (UTC)

Special:ListProperties/quantity gives a list of 140 properties by data type quantity. Does this help, Snipre? --Atlasowa (talk) 09:04, 28 September 2015 (UTC)
Yes, thanks for the link. Snipre (talk) 09:09, 28 September 2015 (UTC)

Wanted pages

Is someone monitoring Special:WantedPages? It currently lists almost a dozen "missing" pages that are linked to from over 100 other pages. The "most wanted" page (P107 (P107) with 577 links) and 6th most wanted (P132 (P132) with 179 links) are properties that were deleted 5 and 9 months ago, respectively. Is anyone working to remove/replace these properties from items? The third "most wanted" page is $1 (459 links), which is a generic placeholder used in some programming contexts that is not appropriate, AFAIK, in any context on this wiki (e.g., even in templates); the 5th and 15th "most wanted" ($link and $talkpage) are similar. Is there something I don't know about how Wikidata works that would make such "variable" links meaningful/useful? - dcljr (talk) 23:32, 27 September 2015 (UTC)

Hmm. Looks like "$1" has a special meaning in the "Translation:" namespace. - dcljr (talk) 23:34, 27 September 2015 (UTC)
Yes, $1, $link, $talkpage have special meaning in the Translation: namespace. So they can be ignored. Mbch331 (talk) 08:27, 28 September 2015 (UTC)
The page doesn't seem to be of much use for Wikidata. --- Jura 08:29, 28 September 2015 (UTC)
The properties P107 and P132 aren't used on items so there is nothing we can remove. Most of the incoming links are from talk pages. --Pasleim (talk) 08:33, 28 September 2015 (UTC)
If there was a MediaWiki hook to ignore links from some namespaces, it would be awesome. By the way, I have just fixed one entry. Matěj Suchánek (talk) 08:55, 28 September 2015 (UTC)
I restored 1 item because it had incoming links from the main namespace: ISO 639-3 (Q17376887). Would be nice if someone could check the item and it's incoming links. See if the incoming links can be deleted or the item merged. Mbch331 (talk) 08:58, 28 September 2015 (UTC)
I thought I had checked WhatLinksHere for at least the 2 properties I mentioned, but apparently not. Oops. If you click on the "(### links)" link after each item, you can narrow it down to links from certain namespaces (e.g., the main one). Jura is right: the properties I mentioned aren't used in the main namespace. And $1 has a special meaning. So nevermind. - dcljr (talk) 04:29, 29 September 2015 (UTC)

New Tool: Wikidata Timeline

 
Screenshot of Wikidata Timeline, displaying a query for Former Countries

Hello, I've been working on a Wikidata-based webapp called Wikidata Timeline. It lets you visualize a list of items (specified by a Wikidata Query) in the form of a timeline. Here are some samples:

Known to work completely on Chrome 44+. Tested to mostly work on Firefox/IE :) It's stable enough to be useful (for anyone who wants to use it), but still has a decent todo list of features to add. Let me know what you think! You can also report problems (or view documentation, or view the source code) at Github. --Hardwigg (talk) 05:46, 28 September 2015 (UTC)

It doesn't work in Chrome 45 here. Sjoerd de Bruin (talk) 06:48, 28 September 2015 (UTC)
:( Could you tell me how it's not working/what's happening? Operating system might also be useful so I can try to recreate your issue. --Hardwigg (talk) 14:30, 28 September 2015 (UTC)
It was on a school computer, running Windows 8.1. Javascript console shows Uncaught Error: [$injector:modulerr]. Sjoerd de Bruin (talk) 06:48, 29 September 2015 (UTC)

Nice, Hardwigg! I tried to improve the info at Wikidata:Tools/External tools#Wikidata Timeline, do you have a screenshot to add? --Atlasowa (talk) 07:37, 28 September 2015 (UTC)

Thanks Atlasowa! I added a screenshot there (and here above). --Hardwigg (talk) 14:30, 28 September 2015 (UTC)
Nice tool, thanks! And also interesting data. Wouldn't have thought that Wikidata knows more current wars than at any other point in history, and that it doesn't know a single war for almost 500 years between 525 and 967. However, that may change with more items getting filled with more data. --YMS (talk) 08:28, 28 September 2015 (UTC)
I found that interesting as well. Wikipedia says that the Dark Ages are "characterized by a relative scarcity of historical and other written records at least for some areas of Europe, rendering it obscure to historians", so it might be a limitation of the period itself. But then again there are 555 items which are instance of:(subclasses of:War), of which 343 (~60%) lack start time information. --Hardwigg (talk) 14:30, 28 September 2015 (UTC)
It's no more unusual than finding that Amos n' Andy was the only American sitcom until about 1950 (no distinction is made between the radio and TV broadcast), and that I Love Lucy was not a sitcom (does not appear in the chart). --EncycloPetey (talk) 13:30, 29 September 2015 (UTC)

P131 items need cleaning

Many items are linked to located in the administrative territorial entity (P131) but some of them are incorrect (P31 is not subdivision of a country) like this or some of them don't have country (P17). the query is here.Yamaha5 (talk) 07:56, 28 September 2015 (UTC)

User:Babel AutoCreate

This automatic account has recently created many categories that do not make sense. Eg. Category:Cs-3 should be Category:User cs-3. If you look at Special:CentralAuth/Babel AutoCreate, it's clear there are some problems cross-wiki. I suppose it should be blocked and nuked. Does anyone know whom we should report this? (Notifying per if this account is malfunctioning please notify the community.) Matěj Suchánek (talk) 09:49, 28 September 2015 (UTC)

My guess would be Phabricator. Mbch331 (talk) 10:23, 28 September 2015 (UTC)
@Lydia Pintscher (WMDE), Aude, Katie_Filbert (WMDE):. Snipre (talk) 11:27, 28 September 2015 (UTC)
See phab:T112868. It seems it's already been fixed and isn't creating new ones. It's not clear whether there's a script to delete the bad ones it created yet or not. - Nikki (talk) 12:30, 28 September 2015 (UTC)

Mandatory language fields for languages not recognised by Wikidata

I've come across this twice, but I can't remember the first example. It was a tribal language of Colombia I think. The second example is in Demet (Q18411288), where the official title of the work is in Ottoman Turkish, a language that isn't used much any more, to the extent that Wikidata doesn't list it separately. The best I could do is "Turkish", but that is not the same language and in fact incorrect. There are probably many other examples, because languages fall out of use over time, but their data doesn't stop existing.

  • Where can I make a suggestion to have this fixed/changed? I don't think this area is a good place to make suggestions like this.
  • Is this a known issue? Was it discussed before somewhere, maybe by someone like me bringing this up here, or by someone making suggestions during development?

--BurritoBazooka (talk) 23:06, 28 September 2015 (UTC)

That lots of languages are missing should be a known issue. It comes up regularly. I would love to know what the proper way to request a missing language is, because my experience so far is that requests go largely ignored regardless of where they're requested.
I did find phab:T59342 which suggests that Ottoman Turkish is supposed to already be supported (although clearly it's not). It's probably the same problem as phab:T93880 where Northern Thai has just been enabled but still can't be used in monolingual text properties. - Nikki (talk) 01:06, 29 September 2015 (UTC)
Hey :) There is a problem that the frontend and backend currently use two different ways to determine which languages are available. This leads to errors where there should be none. Adrian is working on that. He was on paternity leave for a few month which is why this stalled unfortunately. The ticket for that is phabricator:T78006. The other issue is languages we said should work not working. If such cases exist please reopen the existing phabricator tickets for them so we can figure out what's wrong - maybe just a missing configuration. --Lydia Pintscher (WMDE) (talk) 07:28, 29 September 2015 (UTC)
Nikki proposed a Wikidata:Property_proposal/Generic#hanja property for Korean names etc written in Hanja and I voted against it and said we should use monolingual text properties like official name (P1448) instead, with a language that indicated the name was written in the Hanja script.
Lydia Pintscher (WMDE): Does phab:T97882 cover this or does it need a separate phab item? If number with units can take any (or no) item as a unit why can't monolingual text do the same and have any, or no item as a language. That way we can create new language-plus-script items as required? Joe Filceolaire (talk) 22:45, 29 September 2015 (UTC)
Yes that is enough. The underlying mechanism is different so it's not as trivial as just flipping a switch. But we'll look into it. I understand that more languages need supporting. --Lydia Pintscher (WMDE) (talk) 16:13, 30 September 2015 (UTC)

Wikidata weekly summary #177

 
Here's your quick overview of what has been happening around Wikidata over the last week.
Worked on a small birthday present (One month left!) and party organizing (more info coming in the next days) \o/ Matěj Suchánek (talk) 19:35, 29 September 2015 (UTC)

Let's celebrate Wikidata's birthday next month!

Hey folks :)

Wikidata is turning 3 years old on October 29th. This is a reason to celebrate for all of us. We'll be having a big party in Berlin and I'd love to see as many of you there as possible. You can find out more at Wikidata:Third Birthday/Party

We also have a few more things planned where I'd love to have videos, images and other input from you. More on that on Wikidata:Third Birthday/Party as well.

Cheers --Lydia Pintscher (WMDE) (talk) 14:19, 30 September 2015 (UTC)

Mixed data: deconvolution

Mixed data

Generally, these two terms are exchangeable, but does have differences.

‘Deconvolution’ is more popular, and mathematically more concise.

Consider X=A*S, deconvolution should produce a unique solution that reflect the ground truth.

In contrast, decomposition can be any ‘matrix factorization’ in the form of X=A1*S1 or A2*S2, the solutions are not unique and may not reflect the ground truth.

Deconvolution, refers to process with goal of identify ultimate components, elements. Decomposition, refers to break down the mixtures into parts, but not necessarily the ultimate components.  – The preceding unsigned comment was added by ChunyuLiu (talk • contribs). 18:31, 19 September 2015‎ (UTC)

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat/Archive/2015/09&oldid=1102551798"