Wikidata:Project chat/Archive/2015/10

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

given name (P735) on items about pen names

I'm wondering what the proper value for given name (P735) for items whose primary label is a pen name (Q127843) should be (like Robert Jordan (Q166351) and Robin Hobb (Q234403))? AFAICT from the german and english articles linked at given name (Q202444) neither Robert nor Robin are the correct given names because they're not the names given to these person at (or shortly after) birth. The same question also stands for family name (P734) --Mineo (talk) 06:43, 24 September 2015 (UTC)

For pen names we have pseudonym (P742). Use given name (P735) only for the name given at birth. --Pasleim (talk) 07:51, 24 September 2015 (UTC)
That's what I was hoping for, thanks! --Mineo (talk) 07:58, 25 September 2015 (UTC)
@Pasleim: should we take that literally? (I'm guessing we shouldn't) So legal names which were not assigned at birth (i.e. they changed it later) are not given names? Or can given names also be "given" by other certificates/themselves and so be changed? For example, Chelsea Manning (Q298423)'s given name (P735) is listed to appropriately account for when her name changed from Bradley to Chelsea, but Chelsea isn't her birth name. Also, Jimmy Carter (Q23685)'s given name (P735) is listed as "Jimmy", but I don't think that is even his legal name, and his name at birth is commonly known to be "James". --BurritoBazooka (talk) 08:44, 25 September 2015 (UTC)
@Pasleim, Mineo, BurritoBazooka: I don't think the advice given above is correct. If one were to run a query for "all SF authors with first name Robert", one would certainly want Robert Jordan to be included in the returns. I therefore suggest that Robert Jordan should have a given name (P735) value = "Robert", but maybe also with an agreed qualifier statement, perhaps applies to part (P518) pseudonym (Q61002), so that if somebody really wants to exclude pseudonyms from their query than they can. Jheald (talk) 10:19, 25 September 2015 (UTC)
I can not agree with that. There is no unique defintion of a given name, but defintions you find in literature speak about a name given to a person [1], not chosen by oneself. Only exceptions can be if there was a legal act by which the name was changed. --Pasleim (talk) 16:05, 26 September 2015 (UTC)
I think it is a mistake to put too much weight on precise wording of the label or description attached to an item or property. These should be considered to be current reasonable approximations at most, and are far from guaranteed to be directly equivalent even across languages. So I think it is not appropriate to base an analysis in a legalistic mode on what may or may not be the narrow meaning of the word "given".
More important, it seems to me, is how the property can be useful. In this case it is useful to be able to find authors by their first name, even if that first name relates to a pseudonym. It would be easy enough to add that additional information using a qualifier, to allow people to make the distinction; but would make searching harder without it. In any case, we may not necessarily even know whether an author's name is a pseudonym or not. It would be silly to remove an existing given name (P735) just because we discover the name is a pseudonym. Property birth name (P1477) exists to record the actual name somebody was legally registered with at birth. That is the appropriate place to look if somebody really wants a birth name, rather than somebody's most commonly used personal name distinct from their family name. Jheald (talk) 19:02, 4 October 2015 (UTC)

About Property:P627

I have a question about the Property:P627. The IUCN ids are not stable. In 2013.2 version, the website changed all ids in the bird species. As you may read in es:Wikipedia:Bot/Solicitudes/Archivo/2014 Sem2#UICN, the site manager said about it: "The Ids are not fixed, although we try our best not to change them – the birds had to change because of an internal system issue. You should really be linking using the name of the species rather than the Ids, if you can. We are working on an API, that allows you to link to the species name, and this redirects you to the proper page on the Red List website, if that would help. You can have a look here: , look at the weblink method". Therefore in es:Template:IUCN ids not used, but a redirection that is based on the species name. Is it appropriate to maintain a id that is not stable over time? Metrónomo (talk) 06:29, 27 September 2015 (UTC)

@Metrónomo: Usually the ID will vanish if a taxon is no longer assessed or is lumped/splitted. My bot is updating IUCN conservation status (P141) together with IUCN taxon ID (P627) here for a while. The data provided by IUCN have two problems: 1) they do not refer to all „objective” synonyms and 2) spelling variants are unresolved. This makes it harder to connected them to an item with taxon name (P225). We have around 1000 items with IUCN conservation status (P141) mostly imported from enwiki which have no reference. A lot of them are birds. :( --Succu (talk) 18:46, 2 October 2015 (UTC)

Who is an author of Tree-tool?

Hello. Who is an author of Tree-tool? I need the name, just for a honour. I'm writing the wikidata-manual for our ru-ws community. -- Sergey kudryavtsev (talk) 12:29, 30 September 2015 (UTC)

@Sergey kudryavtsev: It's Magnus Manske I think. — Ayack (talk) 13:17, 30 September 2015 (UTC)
Eh, it's seems that all wikidatа tools created by Magnus! Only WEF tool set is by Vlsergey. -- Sergey kudryavtsev (talk) 03:52, 1 October 2015 (UTC)

Editing Wikidata's data on a client input page

For my bachelor's thesis I will develop a concept to improve the editing on a client. I've set up an input page and would be happy about any comments and suggestions! Incabell (talk) 13:06, 30 September 2015 (UTC)


Hey :) Charlie is working with me. It'd be great if you can give her input and also help spread the word to the Wikipedias so the editors there give their input. --Lydia Pintscher (WMDE) (talk) 13:27, 30 September 2015 (UTC)
Hi Incabell. Welcome to Wikidata. I fixed your link. Joe Filceolaire (talk) 23:44, 5 October 2015 (UTC)

Problem with an element


I have a problem with the item voice actor (Q2405480). In English, it is "voice actor", but it was translated in French as "acteur de voix", or "acteur vocal", which, while being literal translations, make no sense at all. The problem is that in French there is no real equivalent to "voice actor". In general, we refer to people who do dubbing (of foreign productions, of cartoons) by the expressions "doubleur" (dub actor (Q21012853)) or, occasionally, as "acteur de doublage" (dub actor (Q11481802)). I don't know about other languages, but in Italian they are known as "doppiatori". So I changed the french translation of voice actor (Q2405480) as "acteur de voix-off" ("voice-over actor") but that is not satisfactory, because in English "voice actor" is often used for people who do dubbing (for example : Mel Blanc) and not only voice-overs.

Also, I think that dub actor (Q21012853) and dub actor (Q11481802) may need to be merged (in favor of dub actor (Q21012853) which is shorter and more common). But with voice actor (Q2405480) we have three elements which often refer to the same thing, although the third one has a wider sense, which cannot be translated. So, apart from the fact that "acteur de voix" should never be used (it simply does not exist), what should I do ? Thanks for your help. Jean-Jacques Georges (talk) 18:41, 30 September 2015 (UTC)

See Talk:Q2405480. --Kolja21 (talk) 23:33, 30 September 2015 (UTC)
There is, unfortunately, no way to always have good natural sounding labels for everything in every language. Languages just don't work that way. Some will consider two things a single concept while others will consider them two separate concepts and may not have a nice way to refer to them both as a single concept, like here. In those cases, all you can really do is to use a label which is more descriptive (e.g. something like "actor for voice-only roles") or list multiple things (e.g. "voiceover or dub actor"). If you haven't already, asking on the French chat page (Wikidata:Bistro) for suggestions might help. - Nikki (talk) 00:17, 1 October 2015 (UTC)
Nikki, the problem is that some users have created a new infobox (here) which imports raw data from wikidata. As a result we have a lot of weird, badly translated stuff imported into the French wikipedia. I went to see that problem after seeing that the infobox had labelled a French actor who does dubbing "acteur de voix", which makes no sense at all in French. So if the data here impacts Wikipedia directly, what should we do ? Unless you think that the data here should not be imported directly into biographical templates. I also mentioned that issue with Danrok, who seemed to think that. So unless raw data import is prohibited, we have to find a solution to untranslatable - and therefore badly translated - things being imported on various wikipedias... Jean-Jacques Georges (talk) 05:48, 1 October 2015 (UTC)
Surely, this boils down to what is claimed in the original source? For example, the credits in the actual movie/work, and other reliable sources. If someone has been credited as a voice actor in a given movie, then all we do is make that statement in the data. Although that isn't always so straightforward because one movie can have many releases/cuts which are not identical, and at present we generally only have one data item per movie. The occupation property is always going to be problematic because it is an abstract concept, and not exact data in the way that date of birth is for example. Danrok (talk) 15:28, 1 October 2015 (UTC)
Danrok yes but as I told you the problem is that now we have on wikipedia fr biographical templates that automatically import stuff from wikidata. So we have to find solutions of some sort... Unless you confirm me that wikidata's data must not be imported into wikipedia's template. That seems to be your opinion, if I recall, but I'd like to know what the other users here think.
As for voice actor (Q2405480), since it is very often (most of the time actually) used to refer to people who, in French, are known as dub actor (Q11481802), maybe the solution would be to translate voice actor (Q2405480) as dub actor (Q11481802) in French, or to merge both items ? (even though dub actor (Q11481802) has a somewhat more restrictive sense...) Jean-Jacques Georges (talk) 18:46, 1 October 2015 (UTC)
Thanks Jean-Jacques Georges for bring this here. The fact that after French Wikipedia started using our data we found this thorny little problem is not a bug. It's a feature. This is a knotty little problem tied up with some language issues (a french label that is just wrong) but also with some cultural issues (not many actors make a living dubbing movies into English so it isn't such a big deal for English speaking editors) and we wouldn't have known it was there without the french experiment. These problems have now been dragged out into the sunlight.
First thank the French editors for spotting this.
It sounds like we need two different occupations - one for voice actors in original roles and another for voice actors recording translations (including translations of the voice actors in the first group). Does that sound right to the French editors or do these groups overlap so much as to be virtually the same? Do we need a supergroup that these two can be subclasses of? If so what should it be called?
If we need two (or three) items then we need to untangle the various existing items and their sitelinks and align them with these roles/occupations.
Then we need a simple mechanism for the French editors to correct the data for any particular item, as needed.
In short we need the French editors to think like knowledge engineers - they deal with templates so they are already half way there. Does that make sense? Joe Filceolaire (talk) 23:33, 5 October 2015 (UTC)
On the talk page for this item there is a definition regardless of language barriers: Talk:Q2405480#Classification. --Kolja21 (talk) 23:54, 5 October 2015 (UTC)
Joe Filceolaire : I'm afraid that these groups overlap a lot so they are virtually the same. Some actors dub mostly cartoons, other specialize in dubbing TV or theater productions, but others do pretty much everything. If I take the case of Roger Carel, one of the best-known "voice actors" in France (he has been, among many other things, Astérix's official cartoon voice for about 50 years), this guys dubs cartoons (foreign or not), films and TV series, and he has also done a lot of voice-overs. Not to mention the fact that he has also appeared a lot on screen, so he's not only a "voice actor" (but still his voice is very familiar in foreign productions dubbed into French). There is not a specific word in French for someone doing all these things : since dubbing is the most familiar activity, people generally say just "doubleur" (litterally "dubber") or "acteur de doublage" ("dubbing actor"). It's the same for Italy, where dubbing is extremely common.
Indeed, I think there is a cultural factor at work : since in the US or the UK, dubbing is not very common, people have started using expressions like "voice actor", but it's basically the same thing. Something like 90% of the occurences of "voice actor" seem to refer to what the French would call a "doubleur" (i.e., someone who post-synchronizes production). So I'm starting to think that "voice actor" should be simply translated as "doubleur" in French and "doppiatore" in Italian. However, such notions do not include people who do things like audiobooks.12:50, 8 October 2015 (UTC)

Technical properties for films

Where would I propose creating technical properties fields for films. Sadly those categories were deleted from Wikipedia, and the data now only resides as IMDB technical specifications. See technical specifications for the Third Man at IMDB vs what we current information we store for the same film here. --Richard Arthur Norton (1958- ) (talk) 22:42, 30 September 2015 (UTC)

Wikidata:Property proposal is the place for all property proposals. If you're not sure which of the subpages to use, feel free to add them to the unsorted page. - Nikki (talk) 23:33, 30 September 2015 (UTC)
@Richard Arthur Norton (1958- ): Before starting the proposal of new properties, please contact the Wikidata:WikiProject Movies. There are already a bunch of existing properties to describe movies so better be sure of the need of new properties. Snipre (talk) 08:31, 1 October 2015 (UTC)
The subpage is "creative work". We already have "running time" and "aspect ratio". I think there is also a pending proposal for film format. If the data is in infoboxes at Wikipedia, it's fairly simple to import (categories are not needed). --- Jura 08:42, 1 October 2015 (UTC)

Olympic Disciplines

As well documented in the en:Olympic sports article on Wikipedia, the Olympic movement operates a hierarchy of Sport / Discipline / Event, as an example Aquatics / Diving / Mens 3m springboard.

We currently have the classes Olympic sport (Q212434) and Olympic sporting event (Q18536594) - so we are missing the disciplines class in the hierarchy.

I would suggest creating a Olympic discipline, as a subclass of Olympic Sport and with subclass Olympic sport event.

This would however entail that sport (P641) would need to move from Olympic sport to Olympic discipline.

Please share your thoughts on this topic --VicVal (talk) 10:53, 1 October 2015 (UTC)

Not all olympic sports are sectioned into disciplines (in other words: there is only one discipline for these sports; example: Rowing / – / Men's eight). How would these cases be affected by and handled with your proposed changes? Apart from that, I don’t see the point with sport (P641): Could you please describe why changes would be necessary and give examples? —MisterSynergy (talk) 11:55, 1 October 2015 (UTC)
Good point! I got myself lost in my own pity, as Diving typically ends up as being referred to as Swimming or more seldom Aquatics. So question is, if we should introduce the "Discipline" for those Olympic Sports which actually have this structure and then accept that "all" Olympic sports are divided into two different class structures?
For sport (P641) the identification of Diving, Water Polo and Swimming should then be at the Discipline level and Aquatics be defined at the sports level. --VicVal (talk) 08:52, 2 October 2015 (UTC)
Well in general I think it is useful to have an Olympic discipline layer in place, thus I support your idea. Since sport (P641) is used quite messy anyways, I would suggest not to bother about it. Rather try to connect diving (Q7735) with sport (Q349) by using subclass of (P279) as required on Property talk:P641. This needn’t necessarily be a direct connection, you can have something like diving (Q7735) (the Olympics discipline) subclass of (P279) swimming (Q31920) (the Olympic sport) subclass of (P279) water sport (Q61065) subclass of (P279) sport (Q349). Apart from the very first subclass of (P279) statement in this list, everything is already properly set as far as I can overlook it (okay, maybe swimming (Q31920) needs some changes depending on whether we position it on sports or discipline level; look at the WP articles what the actual content is). Useful in this context are also Olympic sport (Q212434), Olympic sporting event (Q18536594). An item “Olympic discipline” does not exist as far as I see, but I think it would be okay to create one and use it with instance of (P31) on the corresponding items according to e.g. en:Olympic sports. If you create the “Olympic discipline” item I would appreciate if you arrange this initial setup. —MisterSynergy (talk) 09:22, 2 October 2015 (UTC)
There is, in my opinion, a need for a taxonomy of sport. This should accommodate the fact that the paralympics, for instance, may have a dozen different events in a discipline. A discipline can can separate veterans and juniors divisions each with full range of events.
'sport (P641)', as I understand it is to refine a claim that some has <occupation:sportsperson>. This seems to correspond to "Olympic discipline" as this is the level at which most persons (as I understand it) specialise. They compete in various events in that discipline but few compete in multiple disciplines. Joe Filceolaire (talk) 23:12, 5 October 2015 (UTC)

Authority control

Hi all, I only recently discovered that the English Wikipedia's authority control template is only pulling data from a tiny subset of person-related properties. It was my understanding that this would resolve eventually to all relevant authority control properties on any item of any type. Does anyone know of a Wikidata project in this area that is working on something like this? I would like to see an Authority control template that pulls useful sourcelinks for various domains non-person-related, such as artworks and books. --Jane023 (talk) 11:08, 1 October 2015 (UTC)

It's really up to the Wikipedia to decide which data it wants to pull. From the Wikidata side, authority control identifiers on artworks or books are just external identifiers (string properties) like those on people. For books, there is a specific WikiProject, but I think its main purpose is to not rely on external identifiers ;) --- Jura 11:38, 1 October 2015 (UTC)

(Bot for) Daily articles & categories

There are many pages like Q21034767 & Q18710838. Could somebody please write a script to fill all of them with properties like I've done in these two? It would be also cool, if there was automatic script filling these properties on creation of such item.


Danny B. 11:09, 1 October 2015 (UTC)

I'm quite unsure if we should really add follows (P155) and followed by (P156) to categories. --Pasleim (talk) 13:36, 1 October 2015 (UTC)
For Q18710838, shouldn't the main subject be just an item "30 September 2015" with P31: e.g. "date" (currently it's "Wikinews:2015/September/30"). The item "30 September 2015" would have "point in time": 2015-09-30. --- Jura 13:51, 1 October 2015 (UTC)

Problem between temperature scales and temperature scale units

Jasper Deng
Egon Willighagen
Denise Slenter
Daniel Mietchen
Emily Temple-Wood
Pablo Busatto (Almondega)
Antony Williams (EPA)
Devon Fyson
Samuel Clark
Tris T7
Robert Giessmann
Cord Wiljes
Jonathan Bisson
Charles Tapley Hoyt
Peter Murray-Rust
  Notified participants of WikiProject Chemistry Tobias1984
Daniel Mietchen
Simon Villeneuve
Toni 001
Marc André Miron
  Notified participants of WikiProject Physics

We have three temperatures scales degree Celsius (Q25267), kelvin (Q11579) and degree Fahrenheit (Q42289) but these items are used for the units too. I think we have to differentiate the temperature scales from their units. What do you think ? Snipre (talk) 20:19, 1 October 2015 (UTC)

Réaumur scale (Q223061)? --Succu (talk) 20:23, 1 October 2015 (UTC)
We can add it to the list. But this doesn't help in the current problem resolution. Some comments ?Snipre (talk) 20:52, 1 October 2015 (UTC)
@Snipre: I currently don't see the conflict of definition, but good to have the discussion now. Is there any other database that seperates those 2 concepts (ping @Emw: for his opinion)? --Tobias1984 (talk) 21:25, 1 October 2015 (UTC)
@Snipre: What exactly is your point? Absolut vs. relative scale? --Succu (talk) 21:47, 1 October 2015 (UTC)
@Succu, Tobias1984: Just have a look at the classification of the items, for example degree Celsius (Q25267) which is defined as "instance of" temperature scale (Q2394680) and should be "instance of" "unit of temperature" according to the general classification for units.
For me there is a problem especially if we consider Richter scale (Q38768): this is a scale without unit. So we shouldn't mix scale and corresponding unit. The scale has some special properties like origin, method of graduation or measurement which are not relevant for the unit. Snipre (talk) 08:35, 2 October 2015 (UTC)
@Snipre: I think it should be subclass of (P279) of temperature scale (Q2394680). And couldn't we just add 2 statements to the p279 property? --Tobias1984 (talk) 21:17, 2 October 2015 (UTC)
making temperatures comparable is tough. The easiest way is to have a bot that goes round and wherever it finds a temperature in Kelvin, Fahrenheit etc. it adds the value in Celcius to that statement. Not very elegant but quite effective. Joe Filceolaire (talk) 23:00, 5 October 2015 (UTC)

What do we have to do with similar units

Jasper Deng
Egon Willighagen
Denise Slenter
Daniel Mietchen
Emily Temple-Wood
Pablo Busatto (Almondega)
Antony Williams (EPA)
Devon Fyson
Samuel Clark
Tris T7
Robert Giessmann
Cord Wiljes
Jonathan Bisson
Charles Tapley Hoyt
Peter Murray-Rust
  Notified participants of WikiProject Chemistry Tobias1984
Daniel Mietchen
Simon Villeneuve
Toni 001
Marc André Miron
  Notified participants of WikiProject Physics

Sometimes there are units with different labels but equivalent concepts. Examples:

  • "Siemens per meter" is equal to "ampere per volt meter"
  • "gram per cubic centimetre" is equal to "gram per milliliter"

Do we have to create different items or do we have to use one item with different aliases ? Please comment in the next talk page Wikidata talk:Units. Thanks Snipre (talk) 20:52, 1 October 2015 (UTC)

We can use one item with aliases but there will be a few cases where separate items are needed. These will generally be where the units are used in different spheres and for measuring different things. The deciding factor: Do we need to make different statements on the different items? Joe Filceolaire (talk) 21:47, 1 October 2015 (UTC)
Well, state that something is "X Siemens per meter" when the source tells "X ampere per volt meter" looks strange. I would prefer to have different items for them, with something like "said to be the same" between them. -- Innocent bystander (talk) 08:48, 2 October 2015 (UTC)
@Innocent bystander: Nothing prevent you to add both terms in the label like "Siemens per meter/ampere per volt meter". I create an example with gram per cubic centimetre (Q15639371). I think we should stay with the principle of "one concept, one item". If we start to create several items for the same concept in order to take account of cultural or particular cases, we will face a big problem like the one with have with human occupation which have different words depending on the gender in some languages. Snipre (talk) 13:11, 2 October 2015 (UTC)
I think we should stay with the concept of not adding anything to our statements, that the source does not tell. -- Innocent bystander (talk) 13:42, 2 October 2015 (UTC)
I disagree. People people are looking for information, not an excerpt collection. There is no need to create separate claims for the same information stated in different ways in different sources. Jc3s5h (talk) 15:26, 2 October 2015 (UTC)
Claims like "Hydrogen has the mass of 1,6605 · 10−27 kg" when the source in reality tells "1,008 u" looks very strange to me. I would then prefer to see a quality-version to the "stated as"-property. -- Innocent bystander (talk) 09:05, 3 October 2015 (UTC)
@Innocent bystander: Can we stay in the scope of the first question ? We never propose to change the value from the sources but to be able to merge same units. If you take a closer look at the proposed examples, you can see that 1 gram per cubic centimetre is equal to 1 gram per milliliter. So there is no conversion for the values. Snipre (talk) 07:24, 5 October 2015 (UTC)
I know, there is no difference in magnitude between these entities. But in some ages and in some sciences, they prefer their own way to write things. A chemist normally do not write tons/m3, and a civil engineer do not talk about g/cm3, even if it is exactly same thing. This since a laboratory-chemist (normally) do not handle tons of a substance and civil engineers do not handle grams. As I said, I can accept such "conversions" if there is a kind "stated as"-property as a qualifier. -- Innocent bystander (talk) 11:23, 5 October 2015 (UTC)

So what's the unit difference between gram (Q41803) and kilogram (Q11570)? --Succu (talk) 22:14, 1 October 2015 (UTC)

I think it would be nice to create a relation between these items, to make it possible to make conversions between units with the help of statements, instead of hardcoding it into the modules and templates. How about a conversion-property that connect all convertible units to the SI-units? -- Innocent bystander (talk) 08:48, 2 October 2015 (UTC)
The number of conversions for all possible derived compound SI units quickly becomes unmanageable. If there is to be anything of this sort, there should be an agreed list of preferred units for each type of measurement, and all the variations should have conversion factors to the preferred unit. Jc3s5h (talk) 15:26, 2 October 2015 (UTC)
I do not propose that we should have relations between Lumen and Kilogram or even between "atomic mass" and "hektogram", but only from the non-SI-standard-units to the standard SI-units like gram to kilogram. That would make it easier to define such things as foot (Q3710) and mile, which can have multiple definitions. -- Innocent bystander (talk) 08:51, 3 October 2015 (UTC)
Maybe we should provide new property for dimension of physical quantity (Q19110), with layout similar to has part (P527), where components are our selected basic SI units and probable some few other. Any unit item included in the set should be decorated with qualifier for exponent (Q2233915) (as number) and optionally unit scale (i.e. something from SI prefix (Q131299)). I don't know how to handle logarithmic scale (Q937378) yet, but maybe someone has brilliant idea for it. Such definition in the unit item allow to create script that convert it to artificial universal unit. Then using said to be the same as (P460) you can find other units, and using reverse conversion from that definition got result in that units. Paweł Ziemian (talk) 15:55, 2 October 2015 (UTC)
We have one item for 'Waiter/waitress' - 'waiter (Q157195)'. If your language has two words and no word that cover both then use both words in the label or some long description like the English label for that item. The same applies to measurement units. There needs to be a real reason that can be expressed in statements to justify separate items rather than just having aliases.
If you want to record the exact words of the source then use "quotation (P1683)'. That is what it is for. Statements are for turning that information into a form that is machine readable and comparable.
We need a property to give the size of every measurement unit in standard SI units - every length unit in metre, every pressure unit in Pascal etc. This is the first step to making every physical measurement in wikidata comparable with every other similar measurements.
Some properties to record which fundamental SI units make up which derived units and in what combinations would probably be nice as well. and that is my opinion. Joe Filceolaire (talk) 22:49, 5 October 2015 (UTC)

Report for people with year of birth but not date of birth

Is there a report for people with year of birth but not date of birth? I would love to fill in those missing dates from the WWI and WWII draft registration for people that resided in the United States. --Richard Arthur Norton (1958- ) (talk) 22:37, 1 October 2015 (UTC)

Here's a query for male Americans born after 1875 and before 1930, that have a day-specific date of death, but only a year-specific date of birth. It finds 1563 of them.
You should find it easy enough to drop any of these conditions, if you want to make the query broader.
Use the 'Download' button to get a tab-separated file (or whatever format is most useful for you). Jheald (talk) 09:54, 2 October 2015 (UTC)
Goof stuff, thanks! Working well, I was able to fill in the first one in just a few seconds. --Richard Arthur Norton (1958- ) (talk) 16:59, 2 October 2015 (UTC)
If anyone else is interested in filling in the missing dates of birth, you can register for free at Familysearch and have access to passport applications and the WWI and WWII draft and the Social Security Death Index as well as several state birth indexes. --Richard Arthur Norton (1958- ) (talk) 18:38, 3 October 2015 (UTC)

Instead of opening "edit", it redirects to "Set a sitelink"

Few minutes ago I clicked on "edit" (upper-right) at least seven times and every single time it redirects me to "Set a sitelink". It happened also with tens of other items in past two days, and it's going worse and worse. I do not want to set a sitelink, just to change or remove item, so I'm losing my precious time because some charlatan programmer can't fix basic things. Who is responsible for it? --Orijentolog (talk) 09:58, 2 October 2015 (UTC)

That means the Javascript isn't fully loaded yet. Could be the amount of gadget and scripts or the speedyness of your conputer. Sjoerd de Bruin (talk) 10:01, 2 October 2015 (UTC)
@Orijentolog: if this isn't working for you after Sjoerddebruin's suggestion, try using a different browser. I like to keep Chrome as a "vanilla" browser, as I have filled Firefox with privacy/CSS plugins. Things tend to work better with less plugins. Also, it is good to approach these things in a level-headed way, looking inwards first instead of trying to blame others from the outset. To help fix this problem (if it might be happening for other reasons than what was already suggested), what browser version are you using? --BurritoBazooka (talk) 14:40, 2 October 2015 (UTC)
Sjoerddebruin & BurritoBazooka, thanks for info. I use Mozilla as primary browser, and I checked Java and it's up-to-date. Now it's faster then in morning, but still a bit slow. It's very strange because after 6,5 years on Wiki and over 125,000 edits, I don't remember anything like this before. So I guess it's my right to be pissed off, a little. :) --Orijentolog (talk) 14:59, 2 October 2015 (UTC)
@Orijentolog: The latest version of Mozilla Firefox is 41.0.1 (I use 40.0.3 and it works fine most of the time), you can check which version you have at "Help" menu > "About Firefox". Also, a technicality: Java is not JavaScript, the latter is only named like that because the language looks kind of similar (a stupid name, in my opinion, it's confused many people). JavaScript is very much reliant on the browser itself (the program runs in the browser) and not Java at all (it does not run as a Java applet). See w:JavaScript#JavaScript_and_Java. Chrome has an entirely different way of doing JavaScript, so pages with lots of JavaScript tends to work better there (in my testing, anyway).
I have often had the same issues as you, but it is only when I have a lot of tabs open, and Firefox becomes slow and doesn't load the Javascript components fast enough. Yesterday at one point it wouldn't load at all (not even in Chrome) unless I was logged out of WikiData, I think that was a problem on Wikimedia's end or with a gadget that was enabled on my account maybe (I suspect LabelLister, if anything). I agree that it should work better/faster, but I think it's just Firefox being slow. --BurritoBazooka (talk) 15:39, 2 October 2015 (UTC)
@BurritoBazooka:, I'm using 41.0.1 version. It could be because of tabs, because I have tens of them opened in the same time (for cleaning wanted categories), but still confuses me because even before many tabs were opened but it worked fine. --Orijentolog (talk) 17:46, 2 October 2015 (UTC)
@Orijentolog:, try to disable Content Translation tool on wikipedia (if it is enabled) and check if it works. There is a bug: phab:T114462. --XXN, 22:00, 3 October 2015 (UTC)

Wikimedia user names as data

My recent proposal for a property for Wikimedia user names was recently archived as "not done". Many of the objections were on procedural grounds, for example that it "May violate WMF privacy policy".

I dispute this - not least because we already store the data in more generic properties (for example, Jimmy Wales (Q181) has website account on (P553)=Wikipedia (Q52); qualified with website username (P554)="Jimmy Wales").

It would be bizarre to not hold this data, when we hold individual's identities on Twitter, Facebook, various authority control systems, and the like.

I think we should reach consensus as a community that it is legitimate to hold this data once it is declared by the individual concerned and/ or is widely publicised in reliable sources. (We should of course, have an overriding policy, preventing outing in all of its forms).

Is this a suitable forum, or should we call hold a formal RfC? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:37, 2 October 2015 (UTC)

@Pigsonthewing: Decisions on the property proposal page should never be thought of as official or community consent. A small well-written RFC might give the issue more attention. --Tobias1984 (talk) 21:23, 2 October 2015 (UTC)

Aspect ratio vs width to height ratio

Q1441762 vs. Q20970434 (Aspect ratio vs. width to height ratio). Should they be merged? One is a general term and the other is used exclusively for images. They are synonyms. --Richard Arthur Norton (1958- ) (talk) 16:55, 2 October 2015 (UTC)

We also have 16:9 Q20970433 and 16:9 Q1383069. The aspect ratio one has films linked to it, but the ratio version has nothing. --Richard Arthur Norton (1958- ) (talk) 00:20, 3 October 2015 (UTC)

And we have also aspect ratio (Q1441762) --ValterVB (talk) 07:45, 3 October 2015 (UTC)
My mistake, I just corrected it to your proper link. Should they be merged as synonyms? --Richard Arthur Norton (1958- ) (talk) 18:35, 3 October 2015 (UTC)
The second pair yes. There is a difference between the two items of the first. --- Jura 19:18, 3 October 2015 (UTC)
Except that if you look at the statements and the sitelinks and the "What links here" on width to height ratio (Q20970434)there is nothing that shows a difference between these. I think the items linking to Q20970434 should be redirected and the items merged.
  • If there is a wikipedia with two separate articles then we need two separate items. We may recommend they merge the articles and then merge the items.
  • If there are statements about one item that are not true about the other item then we need two items.
  • Don't make two separate items just because statements are better in your language if you have two different items. This is a sign that you need longer labels and descriptions in your language making clear that one item covers both language versions (e.g. 'actor/actress' even if the English label is just 'actor' - in current English 'actor' covers both sexes).
Hope this helps. Joe Filceolaire (talk) 21:52, 5 October 2015 (UTC)
It doesn't. The existence of a height-to-width-ratio isn't language dependent, but, in forum chat, we don't need either. Obviously, we can improve on the sitelinking. Personally, I don't understand the statements added to Q20970433, as Q1383069 seems more specific. --- Jura 08:02, 6 October 2015 (UTC)


Some creative works are written in reply to a previous one. How to record it on a data entry?

Looking on current properties, I've found based on (P144), main subject (P921), inspired by (P941) and follows (P155)followed by (P156) are the most closer to my need, but I dont think that they are right for it. Any thoughts? Lugusto (talk) 23:38, 3 October 2015 (UTC)

@555: Propose a new property, at Wikidata:Property proposal/Creative work. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:19, 4 October 2015 (UTC)

Learning how to add Music Artist to Freebase using bot

So lost on this topic of having a bot add me to the freebase site. I understand the freebase website is in read-only so what other way can I get google to recognize the information that would normally just be added to freebase. I do not have a website to be able to add the schema. Please help. The Artist is already added to wikipedia articles and wikidata as new item. What is Wikirecon_bot and how can I learn to get it to create a freebase new artist under music producer?

Wikirecon_bot is a Freebase-thing it seems. You've could add information to the Wikidata-item, but there is no guarantee that Google will pick it up. Sjoerd de Bruin (talk) 07:06, 4 October 2015 (UTC)

Learn how to add datasets that are good enough to be uploaded using a bot like TptBot.

Instructions would be nice

First things first. Make sure your data is "good enough" for wikidata. Our standards are a lot higher than freebase. Until you sort that one you ain't uploading anything. Joe Filceolaire (talk) 22:04, 5 October 2015 (UTC)

Fusion problems

Can someone fusion German de:Kategorie:Short Message Service with English en:Category:SMS-based question answering services (Q8695384) ? 23:50, 4 October 2015 (UTC)

The problem with this one is that "SMS-based question answering services" refer to a very specific type of service provided over the medium of SMS, where questions of users using the service are answered. This does not correspond to de:Kategorie:Short Message Service, which refers to anything related to SMS. They can't be merged because they aren't the same thing. I think a better merge candidate for de:Kategorie:Short Message Service would be en:Category:Text messaging, but I'm not sure, because in English "text messaging" these days can refer to any service where instant messages are sent with phones, whereas "SMS" refers to just one system (of which there are many others like WhatsApp) whereby messages are delivered. Maybe that doesn't matter. I'll leave that to someone else to decide. --BurritoBazooka (talk) 00:49, 5 October 2015 (UTC)

Can someone fusion German de:Kategorie:Vermittlungstechnik (Q9165699) with English en:Category:Telephone exchange equipment 23:54, 4 October 2015 (UTC)

Link to item

Scratching my head: How do I link to a wikidata item from the associated page in some sister project? -geraki talk 07:26, 5 October 2015 (UTC)

A direct link would be [[d:Q12345]] or [[wikidata:Q12345]]. Some projects will have special templates too. - AdamBMorgan (talk) 10:26, 5 October 2015 (UTC)
..unless you mean automatically picking up the data item linked to a page. I think that would require a Lua Module and, again. some projects will already have templates for it. - AdamBMorgan (talk) 10:28, 5 October 2015 (UTC)
Or  ? Also directly on Q12345 --- Jura 10:43, 5 October 2015 (UTC)

I mean (as AdamBMorgan understood correctly) automatically picking up the data item linked to a page, and in fact not a link but just the string/value, for example "Q41" in a template inside en:Greece. I can't find it in the documentation of any "wikidata" modules in various wikipedias. -geraki talk 10:55, 5 October 2015 (UTC)

If the sitelink for en:Greece is defined on Q41#sitelinks-wikipedia, you don't need to call explicitly Q41 on en:Greece.
Sample: To get the information from Q41#P31: you can use {{#property:P31}} and don't need {{#property:P31|from=Q41}} to get the same result. --- Jura 11:05, 5 October 2015 (UTC)
I've remembered the name of the Lua Module. You can use {{#invoke:Wikibase|id}} on any project that has a version of Module:Wikibase (eg. English Wikipedia's version), which is based on mw:Extension:Wikibase Client/Lua, so most should have it. - AdamBMorgan (talk) 11:27, 5 October 2015 (UTC)

Merging items from mgwiki with other items

Somebody created items for the articles from mgwiki without checking for existing items first. Today I merged James Edward Grant (Q19948889) with James Edward Grant (Q119576). I found these duplicates by chance. I have found at least two cases in the last weeks.

Maybe it would be a good idea to run an automated checkup: If an item exists that refers only the a mgwiki article, and another item with the same lable in any language exists (especially useful for biographies), they should be merged. Perhaps a bot could create a list of merging candidates, so that we could check them by hand later. Jonathan Groß (talk) 10:56, 5 October 2015 (UTC)

There is a bot doing this sort of thing. [Project Merge] looks for merge candidates between pairs of languages, looking for pairs of wikidata Items that link to articles with the same name in the respective wikis. There isn't a pair involving mgwiki. It seems you add requests [here]. I've added enwiki mgwiki as a pair as it seems the most useful pairing involving mgwiki, as enwiki is the biggest wikipedia, and I might be able to contribute to merging. You can add others if you want. Silverfish (talk) 18:55, 5 October 2015 (UTC)

Thanks. Jonathan Groß (talk) 11:25, 6 October 2015 (UTC)

Wikidata weekly summary #178

Proposed global ban for Tobias Conradi (aka Tamawashi)

There is currently an open RfC on Meta-Wiki proposing a global ban for Tobias Conradi (aka Tamawashi, see Category:Wikidata sockpuppets of Tamawashi for more names he has used here). The RfC is at meta:Requests for comment/Global ban for Tobias Conradi. - Nikki (talk) 15:55, 5 October 2015 (UTC)

Couple of days ago I blocked two more of their socks, following the CU results on the English Wikipedia.--Ymblanter (talk) 23:24, 5 October 2015 (UTC)

Fusion problems

Can someone fusion en:Category:Methodist denominations (Q9209758) with de:Kategorie:Methodistische Denomination ? SusanSonntag (talk) 17:32, 5 October 2015 (UTC)

AFAICT this is   Done: Q18820408 was merged into Q9209758 by SusanSonntag (move statements) and PLbot (create redirect) —Galaktos (talk) 17:40, 5 October 2015 (UTC)

How to get a wikipedia page name for an item using the API

Hi - this is a really basic question,, but I couldn't find an obvious answer in the API docs. How can I use the API to find the name of a page on (say) en.wikipedia? So e.g. I have[1339,350]and link[enwiki], and as a return value I want "Cambridge" and "Johann Sebastian Bach", both valid page names on Seems like an obvious question, so apologies if I'm just being dense. Thanks HYanWong (talk) 20:08, 5 October 2015 (UTC)

@HYanWong: the API you linked is only for WDQ. I do not think this is possible with WDQ, but I'm not very experienced with all the APIs so I'm not sure. Once you have Q1339 and Q350, you can use and (a different API) and get the English WP article URL using the path ['entities'][ITEM_ID_HERE]['sitelinks']['enwiki']['url']. edit: See my example below of how to do this with wbgetentities. --BurritoBazooka (talk) 20:54, 5 October 2015 (UTC) --BurritoBazooka (talk) 00:21, 6 October 2015 (UTC)
Python example to help some others understand:
import requests #

johann = requests.get("").json()
cambridge = requests.get("").json()


# edit about 2 hours after posting: I figured out how to get multiple items.
# Advantage of this is that we only need to do one request to the web service. Less wait time.

from urllib.parse import quote # makes " " into "%20" etc

items = requests.get("|Q350&format=json").json()
johann_title = quote(items['entities']['Q1339']['sitelinks']['enwiki']['title']) # I wonder why we can only get a title and no URL is available :/
cambridge_title = quote(items['entities']['Q350']['sitelinks']['enwiki']['title'])
url = "{}"

print("johann:", url.format(johann_title)) 
print("cambridge:", url.format(cambridge_title))
I don't know whether there's a way to get multiple items in one JSON request. --BurritoBazooka (talk) 22:20, 5 October 2015 (UTC)

Does anyone have any documentation for Wikidata's EntityData and wbgetentities APIs, for me and others? I can't even remember where I picked up the information about Special:EntityData. I tried to find it for HYangWong but can't so far. --BurritoBazooka (talk) 21:11, 5 October 2015 (UTC)

Ah, I found some auto-generated docs here: -- more links here: --BurritoBazooka (talk) 21:14, 5 October 2015 (UTC)
That's really helpful, especially pursuing the multiple query options (I want to do this for over a million items: maybe perusing the dumps would be better). It's a shame this isn't available in the WDQ interface, which seems very quick even for large queries like this. Thanks again. HYanWong (talk) 07:53, 6 October 2015 (UTC)
Ah, some discussion at I see.HYanWong (talk) 08:14, 6 October 2015 (UTC)

Cause of death

What is the cause of death for people who died in airplane accidents? --Richard Arthur Norton (1958- ) (talk) 04:10, 6 October 2015 (UTC)

Well, "car accidents" are provided as an example for that property, so wouldn't aviation accident (Q744913) make sense? --BurritoBazooka (talk) 05:11, 6 October 2015 (UTC)
Perfect! Thanks. --Richard Arthur Norton (1958- ) (talk) 03:05, 8 October 2015 (UTC)

Infinit loop caused by cyles when executing a SPARQL query


I want to get instances and all super types of a particular entity using the sparql query :

   wd:Q289 wdt:P31/wdt:P279* ?o.

Try it!

a QueryTimeoutException is firing, because the entity "Entity" <> creates a cycle by poiting to itself. Can you make it as a root ? – The preceding unsigned comment was added by (talk • contribs).

I don’t understand what your problem is… you (or at least, the same IP) created the cycle yourself, and that has already been fixed. So where’s the problem? —Galaktos (talk) 11:56, 6 October 2015 (UTC)
Cycles should not be able to cause path queries to get stuck in infinite loops.
In a query like the one above, the engine will keep trying to add another P279 step to the graph until no new nodes are visited that have not already been visited. It will then return the results.
It doesn't matter if some of the paths are loops, because such loops will get to the point where they do not add any new nodes that have not yet been visited, and will then be considered done. Jheald (talk) 12:08, 6 October 2015 (UTC)
And yet the query is timing out. Jheald (talk) 12:11, 6 October 2015 (UTC)

The query ought to be returning the following

Interestingly, the following variant of the query does work:

SELECT ?o ?oLabel WHERE {
   wd:Q289 wdt:P31 ?a .
   ?a wdt:P279* ?o.
   SERVICE wikibase:label {
       bd:serviceParam wikibase:language "en" .

Try it!

It is possible that in the first query, the engine might be implicitly searching ?a wdt:P279* ?o before wd:Q289 wdt:P31 ?a -- resulting in a massively inefficient query. Jheald (talk) 12:29, 6 October 2015 (UTC)

Here is the 'explain' report for the query that works: [2]
and for the query which doesn't: [3]
(See Wikidata:SPARQL query service/query optimization for a little more about such reports, though it doesn't add much).
It does indeed seem that the original query is trying to effectively run ?a wdt:P279* ?o, and as a result is exploding. (Look at the "Units Out" column of the tables at the end, until the pipeline gets killed).
I had come across a similar thing already, when BIND was used to provide a value for a variable on the right-hand side of a path query [4]; but I didn't realise that, on the basis of the above, seemingly problem applies whenever a more-than-one-stage path query ends in a variable. Jheald (talk) 13:02, 6 October 2015 (UTC)
I have updated the bug report previously filed with Blazegraph (BLZG-1543), and also the the note on our own suggestions/issues page. Jheald (talk) 13:22, 6 October 2015 (UTC)

Could be fictional, we're not sure, humans

So the topic of Kasongo Ilunga (Q1735013) came up on the Wikimedia-UK mailing list and I realised we don't deal well with people we suspect are fictional but it's never been proved so now

OK? Joe Filceolaire (talk) 17:03, 6 October 2015 (UTC)

Quite an interesting move for Biblical characters.
So we now have eg
I suspect that may be correct, but I'm not sure I'd want to be the one to be making the call as to who gets to be instance of (P31) human (Q5) and who doesn't.
Incidentally, what is the guidance on real people who appear in major fiction series, possibly with fictionalised histories, families etc -- eg historical novel series, history plays by Shakespeare, "alternative history" novels, etc ?
Should the wives in the novels be included on the items for the real people (where they may be picked up in searches for real wives?)
Or should new items be created for the fictionalised lives, perhaps said to be the same as (P460) the historical personages (or fictional analog of (P1074)) ?
What about biblical figures generally identified with historical personages (like Nebuchadnezzar II (Q12591) ?), but for whom some claims / relationships may only appear in the Biblical text ?
There's some further thinking, or at least guidance, needed here I think. Jheald (talk) 19:48, 6 October 2015 (UTC)
For what it's worth, we currently have ten examples of fictional analog of (P1074) items identified as humans: Jheald (talk) 20:26, 6 October 2015 (UTC)
Plus about twenty fictitious entities said to be the same as (P460) humans: Jheald (talk) 20:36, 6 October 2015 (UTC)
Imho we should follow a pair of items (like "historical figure" / "figure as depicted in the Bible") when we refer to a fiction biography that may include some non accurate stuffs. Considering the Bible is something you can (or not) believe, I think not everyone would consider it as a reliable, or a secondary source. What seconded by historians and historycal science journals should go into the "historical figure" item, and what's said in the Bible should go into the other one. author  TomT0m / talk page 10:57, 7 October 2015 (UTC)
Jheald: There is also based on (P144). There may be a bunch of fictional analogs using that.
TomT0m: It seems like we don't resolve controversies here - we record the controversy and move on. sort of. Joe Filceolaire (talk) 20:42, 7 October 2015 (UTC)
Sixty hits with based on (P144): Quite an interesting list of fictional figures, even if my reaction to a few entries would be 'citation needed' Jheald (talk) 21:19, 7 October 2015 (UTC)

Merge data

I'd like to merge this with this, as they are on the same topic, but I don't know how. Could anyone help me?--李4 (talk) 00:13, 7 October 2015 (UTC)

@李4: There are instructions on how to do this at Help:Merge. There's a section on "automatic merging" (the first section) which is the easiest but just needs a gadget to be enabled. --BurritoBazooka (talk) 01:57, 7 October 2015 (UTC)
Unfortunately, the diethyl peroxide (Q15726032) is C₄H₁₀O₂, but diethyl ether peroxide (Q3487286) is C₄H₁₀O₃, which is really different, unless if contributors of zhwiki article can change contents, there's nothing to be merged here. psst merged ignore me --Liuxinyu970226 (talk) 03:32, 7 October 2015 (UTC)
I think that's a typo; if you check the zhwiki page, then the formula (CH3CH2OOCH2CH3) is really C4H10O2. There is nowhere to stick in a third oxygen atom. --李4 (talk) 03:38, 7 October 2015 (UTC)
Now that you also agree with me, I removed the mistaken formula. --李4 (talk) 03:40, 7 October 2015 (UTC)
@李4: But enwiki contains Diethyl ether hydroperoxide (CH3-CH2-O-CH(OOH)-CH3)... , how do you resolve this, only change it? --Liuxinyu970226 (talk) 03:42, 7 October 2015 (UTC)
Oh, it's my bad. Now how can I undo this?--李4 (talk) 03:54, 7 October 2015 (UTC)
@李4: To undo the merge you can go to the history of each item and click "restore" (screenshot) next to the revision which is not yours (to restore the revision of the item as it was before the merger). --BurritoBazooka (talk) 04:05, 7 October 2015 (UTC)

Property for online version of work?

Do we have a property for an online version (e. g. a scan) of a work?

In my case, one of dewiki’s sources for the article on Christianus Carolus Henricus van der Aa (Q1083011) is available as an online scan, so I’d like to add it as reference to some statements on the item. If I understand correctly, I shouldn’t use reference URL (P854) for this, but instead create an item for the source and then use stated in (P248). Is there a property to link to the scanned version in the new item for the source? —Galaktos (talk) 10:40, 7 October 2015 (UTC)

Try full work available at URL (P953). --Succu (talk) 10:49, 7 October 2015 (UTC)
Exactly what I was looking for, thanks! —Galaktos (talk) 10:53, 7 October 2015 (UTC)
Oh. I see that Trinity College Dublin have recently made the Book of Kells (Q204221) available online. Adding this claim! Joe Filceolaire (talk) 20:47, 7 October 2015 (UTC)

How to list all items without a qualifier for a given property?

In AutoList, is there a way to find all the items who have no qualifier for a given property? For example, all the items with award received (P166): Commander of the Legion of Honour (Q10855212) but without any qualifier for this statement. Thanks in advance. — Ayack (talk) 17:20, 7 October 2015 (UTC)

I don't think so, but if you compare the 2591 items with the statement [5] to those 61 with the statement and any of the qualifier for P166 [6] you should find your list. You can compare the two with pagepile. --- Jura 17:52, 7 October 2015 (UTC)
@Ayack: Can you write a little bit of SPARQL ? see how to make a negation UPD or maybe a better link : , you should be able to write a query "select all statements with a property and not exists a qualifier snak for the statement". Please ping me if you don't get to a working solution :) author  TomT0m / talk page 17:58, 7 October 2015 (UTC)
Here's a query in SPARQL: -- open the link, then hit 'execute' to run. (Apologies to User:TomT0m if by posting I'm spoiling his teachable moment). You could then paste the query into the SPARQL box in PagePile to get the results into Magnus's tool workflow, if you haven't already done so via Jura's method. Jheald (talk) 18:36, 7 October 2015 (UTC)
The difference between the two must the qualifiers that are being used, but not recommended: Wikidata:Database_reports/Constraint_violations/P166#Properties_statistics --- Jura 18:42, 7 October 2015 (UTC)
Here's the count of qualifiers that are used: -- start time (P580): 79, point in time (P585): 61, date of death (P570): 1.
You had excluded P580 and P570, which are both considered violations. Jheald (talk) 18:57, 7 October 2015 (UTC)
eg: François Nizard Charles Joseph d'Hénin (Q3085468) (etc) for start time (P580); and Alfred-Henry-François Mimaut (Q14918086) for date of death (P570) -- which appears to be keyboard error. Jheald (talk) 19:02, 7 October 2015 (UTC)
I changed the constraint for P166, so going forward start time (P580) should be fine. --- Jura 19:10, 7 October 2015 (UTC)
Seems that for P166 one should be looking for statements that don't have specific qualifiers. Not sure if Ayack was actually looking for P166 .. --- Jura 19:23, 7 October 2015 (UTC)

@Jura1, TomT0m, Jheald: Thanks for all your tips. I'll try to adapt them to my needs. I will get back to you if I have another question. — Ayack (talk) 08:37, 8 October 2015 (UTC)

Editing an item about yourself

Hi, are you allowed to create or edit an item about yourself (assuming your notable)? thanks --Plarishome (talk) 06:32, 8 October 2015 (UTC)

Notable means you have at least one Wikipedia article. If this is the case pls feel free to create and edit the item.--Ymblanter (talk) 06:59, 8 October 2015 (UTC)
One major discussion took place here. And the conclusion was maybe something like: Yes, you are allowed to edit about yourself, but be carefull. -- Innocent bystander (talk) 07:02, 8 October 2015 (UTC)
Seems to become increasingly popular, but I don't think it's a good idea. --- Jura 07:11, 8 October 2015 (UTC)

Mobile users are now redirected to mobile view by default

Hey folks :)

As previously announced we've now flipped the switch. Users on mobile devices are being redirected to the mobile view on Wikidata. This will give them a better view of the data on small devices. If needed they can switch to the desktop site. There is still work to do until Wikidata is truly great on mobile but this is a good step. Thanks everyone who helped.

Cheers --Lydia Pintscher (WMDE) (talk) 09:44, 8 October 2015 (UTC)

An easy improvement for the mobile view might be to do a separate section for external identifiers. (currently within statements). --- Jura 10:09, 8 October 2015 (UTC)
Yes totally. We'll be separating them both on desktop and mobile. Ticket for that is at phabricator:T95287. --Lydia Pintscher (WMDE) (talk) 10:25, 8 October 2015 (UTC)
Interesting read, but the plan to do that is probably even older than the ticket labeled "Epic". Given that it's an epic, I suppose you don't want another round of comments ;)
A short term solution for mobile view with the most frequent ones might not be too complicated to implement. --- Jura 10:41, 8 October 2015 (UTC)
Hehe yeah. Wrt doing it on mobile only: If it would get us closer to having it solved on desktop as well I'd say let's do it. However when we dissected this in detail earlier this week it seems there is no benefit and we should just do it properly for everything. --Lydia Pintscher (WMDE) (talk) 10:56, 8 October 2015 (UTC)

'/header' and '/footer' links

Is it safe to delete Template:Infobox periodic table group/footer (Q15060928) & Template:Infobox periodic table group/header (Q15075201) which have each six sitelinks? Wikidata:Notability/Exclusion criteria say any /header or /footer pages. Matěj Suchánek (talk) 15:00, 8 October 2015 (UTC)

Firearm companies

What statement and property can I use to tie all the firearm companies together? --Richard Arthur Norton (1958- ) (talk) 00:45, 9 October 2015 (UTC)

I reckon instance of (P31): company (Q783794) and product or material produced (P1056): firearm (Q12796). --BurritoBazooka (talk) 03:10, 9 October 2015 (UTC)

Problem with Qualifers

Here the qulaifiers for the adjacent station: Amersham should be -

  • Connecting line : Metropolitan Line ; Direction : Amersham Station
  • Connecting line : London to Aylesbury line: Aylesbury.

Currently the qualifiers are collapsed. Is there a way to do "sub"-"qualifiers?

What I want to say is "The adjacent station on the Metropolitan line is via the Metropolitan line towards Amersham, Amersham, and via the adjacent London to Aylesbury line towards Aylesbury, is Amersham.

Some more complex examples would be Chalfont and Latimer.

"The adjacent station to Chalfont and Latimer is via the Metropolitan line towards' Chesham, Chesham , towards Amersham, Amersham, and via the London to Aylesbury line towards Aylesbury is Amersham.

"The adjacent station to Baker Street is :

  • 'via the Metropolitan line from platforms 1,3,4 towards Harrow on the Hill, Uxbridge, Amersham, Watford, Chesham is Finchley Road
  • via the Metropolitan line from platforms 2 towards Aldgate is Great Portland Street
  • via the Circle line from platfoms 5 towards Edgware Road is Great Portland Street
  • via the Circle line from platfoms 6 towards Hammersmith is Edgware Road.
  • via the Bakerloo line from platforms [placeholder] towards Harrow and Wealdstone is Marylebone
  • via the Bakerloo line from platforms [placeholder] towards Elephant and castle is Regent's Park.


For some mainline stations this might get rather complicated indeed. Sfan00 IMG (talk) 10:12, 9 October 2015 (UTC)

The other issue the above exposed is the apparent lack of a property for "passanger service depature point", which in the above example is the platform number, but could in other contexts be the departure gate, terminal bay, bus stop code etc..

"platform" as a property currently relates to something else.

Anyone here familiar with the UK NAPTAN/NPTG dataset? Analysing how it defines stuff might be usefule here, and as I understood it OpenStreet map was also slowly importing it meaning that at least part of it was under an "open" license...

It may be possible from NAPTAn to have in Wikidata, every railway station in the UK mianland, (not to mention EVERY bus-stop, frerry berth etc...). Do you like Big Data? Sfan00 IMG (talk) 11:04, 9 October 2015 (UTC)

Converting sitelink titles in wikidata dump to urls

Hi - I'm working with the wikidata JSON dump files, which don't give URL to other wikis (e.g. enwiki) directly, but instead list a page title with spaces rather than underscores, and unicode characters listed as e.g. \u0413. What is the standard way to turn these into a normalised mediawiki url? For example, is there some basic regular expression I can use that is documented somewhere? (NB: I'm actually trying to match the page names against the page view stats provided at, which have titles like "Alue_Meudem,_Lh%C3%B4k_Suk%C3%B4n,_Ac%C3%A8h_Bar%C3%B4h", but I'm assuming that once the titles are in normalised form I can simple uri-escape them to get a matching string). Thanks if anyone can help. HYanWong (talk) 21:49, 9 October 2015 (UTC)

Sorry - I solved it - the problem was in conversion of the JSON files to unicode, not in the dump itself. I'll try to document this somewhere. HYanWong (talk) 00:52, 10 October 2015 (UTC)

Translingual labels?

Some labels, like that for Earth (Q2) (♁, ⊕), or United States of America (Q30) (🇺🇸 - if you don't have the proper Unicode, this is a pair of regional indicator symbols), do not conventionally belong to any particular language but are indeed classified as "English" or "German", for example. More practically, languages and countries have their ISO codes classified as labels in languages (like "de" for German or Germany), but ISO codes are not designed to be only interpretable in a particular language. Wiktionary solved this by having a "Translingual" section for terms which do not fit in any language, see for example wiktionary:x#Translingual. I think there should be a separate space to handle labels which do not fit in any language. What do you think? Has this been discussed elsewhere? Thank you. --BurritoBazooka (talk) 19:15, 9 October 2015 (UTC)

There are string properties for some symbols. Maybe a new one for unicode symbols could be made. --- Jura 19:23, 9 October 2015 (UTC)
The immediate practical problem I see with what I said above is that people who might not be editing Wikidata in English don't have the convenience of looking up ISO codes and the like (like "en" and "zh" or whatever) while adding statements, unless those ISO codes are also listed as labels in their own languages. The listing of ISO codes in multiple languages is needless - they don't belong to any language, they belong to all of them. --BurritoBazooka (talk) 19:30, 9 October 2015 (UTC)
string property are not linked to any iso code and displayed whatever language one is editing in. It's different from monolingual strings. --- Jura 19:45, 9 October 2015 (UTC)
Would that make lookups like this possible in any language without adding that ISO code to each one? Your mention of "properties" makes me think that we'd need a change in how the item lookup thing works in order to also search those string property values, not just label values. --BurritoBazooka (talk) 19:51, 9 October 2015 (UTC)
I did a bit of testing. Apparently when using Wikidata in other languages and looking up labels, English labels are preferred in the search if labels in your own language can't be found. I think this is not proper handling of translingual labels/aliases, because by definition they don't prefer English. --BurritoBazooka (talk) 20:02, 9 October 2015 (UTC)
That's the language fallback kicking in. It's a little bit more complicated than just preferring English (although the fallbacks end up at English eventually). There's a diagram at File:MediaWiki_fallback_chains.svg and you can see which order will be used for you by going to Special:MyLanguageFallbackChain. As far as I'm aware, it's not intended as a solution to translingual values, but as a way to make the site more usable, particularly for speakers of less well represented languages, because it's really hard to use the site when all the links on a page say (no label) and none of the search results have a label either. - Nikki (talk) 20:34, 9 October 2015 (UTC)
Shouldn't string values in properties eventually be searchable ? --- Jura 20:45, 9 October 2015 (UTC)
I think that's phab:T58626. - Nikki (talk) 23:45, 10 October 2015 (UTC)

Mexico wins

Mexico won gold medal at London's Olympic Games 2012  – The preceding unsigned comment was added by (talk • contribs) at 23:32, 10 October 2015 UTC‎ (UTC).

  Done Mexico national under-23 football team (Q1926260) --BurritoBazooka (talk) 23:41, 10 October 2015 (UTC)
I'm thinking about adding the rest of the medallists in this manner. Does anyone know a machine-readable database for getting this information? edit: Found one. The ones I've found would need a parser written for, and I figured it would be worth my time to ask here first just in case anyone knows. --BurritoBazooka (talk) 00:12, 11 October 2015 (UTC)

Wikidata:WikiProject Economics

Wikidata:WikiProject Economics Needs more participants and the tables need updating for all the quanity-datatypes that were created recently (See: Wikidata:Property_proposal/Pending/2#total_debt). Please sign up and help with the waterfall of quantitaive property creations we are currently experiencing. --Tobias1984 (talk) 15:01, 4 October 2015 (UTC)

Posting to keep this visible a little longer. --Tobias1984 (talk) 11:33, 11 October 2015 (UTC)

minimum number / unequal precision

Trying to update SS Princess Alice (Q1544903) the sources say the number of deaths is "over 650" so a value of 650 is not accurate but I can't see how to indicate that 650 is a minimum value?

Sources also say that between 69 and 170 people were rescued, so a possibility is to say 650 +101/-0 but it doesn't seem possible to enter unequal precision? Thryduulf (talk: local | en.wp | en.wikt) 10:15, 11 October 2015 (UTC)

From what I know, unequal precision can be added through the api, but not in the UI. -- Innocent bystander (talk) 16:39, 11 October 2015 (UTC)


Can somebody give me a hint of how I in Lua identify the datatype of a random Property?

mw.wikibase.getEntityObject( "PX" ) I guess, but thereafter? -- Innocent bystander (talk) 15:25, 11 October 2015 (UTC)

Try {{#invoke:Wikidata|Dump|id=P105}} on enwiki sandbox to see data representation in Lua. The id=P105 is random entity identifier to show. Indexed parameters are names or indexes in internal tables. Each parameter for next level deep. I see there {{#invoke:Wikidata|Dump|id=P105|datatype}} with value "wikibase-item". Paweł Ziemian (talk) 15:42, 11 October 2015 (UTC)
simply entity.datatype after entity = mw.wikibase.getObject('P123'). Matěj Suchánek (talk) 16:06, 11 October 2015 (UTC)
Thank you! That solved my problems. I made some tests with Module:Cite. It looks like it can be extended to support any property with string-datatype which have a formatter URL (P1630). -- Innocent bystander (talk) 16:38, 11 October 2015 (UTC)

Search items by description

Is there any (easy) way to search for some string in the description field of items only in one particular language? I'm looking for something like Special:ItemDisambiguation, except it had to be not searching on labels, but on descriptions. Thanks in advance. Paucabot (talk) 14:23, 6 October 2015 (UTC)

Not sure. You could use --- Jura 14:27, 6 October 2015 (UTC)
Actually, might be easier --- Jura 14:37, 6 October 2015 (UTC)
It ought to be possible in SPARQL. The SPARQL service (I think) has got fast indexing and range extraction on predicate values. However, the following query times out:
PREFIX schema: <>

SELECT ?q ?desc WHERE {
    ?q schema:description ?desc .
    FILTER (STRSTARTS(?desc, 'Langue'))     # Could also use CONTAINS()
    filter (lang(?desc) = "fr") .

Try it!

Maybe somebody can make it work? Jheald (talk) 14:52, 6 October 2015 (UTC)

Try Autolist to see why it doesn't ;) I think you might want to try "langue" instead of "Langue". --- Jura 15:04, 6 October 2015 (UTC)

Thanks, yes, that would indeed be exactly the problem if the query were returning 0 hits.
Also SPARQL has a function LCASE, so one could try FILTER (STRSTARTS(LCASE(?desc), 'langue'))
But the real trouble here is that the job is getting killed while it's still trying to gather every instance of ?q schema:description ?desc, rather than just extracting the cases it needs from an index. Jheald (talk) 15:14, 6 October 2015 (UTC)
Thanks for your replies, but I have not been able to get it to work. I'm not very skullfull with queries and databases. Actually, I was searching for categories with wrong catalan descrition wikimedia-kategori made by bot. Paucabot (talk) 17:06, 8 October 2015 (UTC)
About 270k: (slow).--- Jura 17:30, 8 October 2015 (UTC)
Thanks, @Jura: I thought they were less. I'll try to ask someone to fix it with a bot. Paucabot (talk) 18:10, 8 October 2015 (UTC)
@Jheald: right now, the query search does not support full text search (which is required to make it efficient). We are looking into how exactly integrate our existing search infrastructure (i.e. ElasticSearch) with Wikidata data and the query service. That's not done yet, so right now the text searches using the query service are not efficient yet. You may try to use the regular search instead. Or wait a bit until we figure out how to integrate it. :) --Smalyshev (WMF) (talk) 08:01, 12 October 2015 (UTC)
@Smalyshev (WMF): I can see why a special indexing approach could help support SPARQL text searches involving CONTAINS(), to find text anywhere in a string; but surely the standard B+ tree that (as I understand it) Blazegraph uses for all predicate + value combinations should be enough to support searches involving STRSTARTS(), for text at the start of a string? Does Blazegraph really leave us with no indexing on text values, if we choose not to activate a full prefix-tree type index of the sort that would support CONTAINS() ? Jheald (talk) 08:40, 12 October 2015 (UTC)

measured physical quantity (P111) and measurement scale (P1880)

The page for measured physical quantity (P111) says that it is inverse property (P1696) measurement scale (P1880). But this is not right - measured physical quantity (P111) relates physical quantities and units, while measurement scale (P1880) relates physical things and scales. E.g. example in measurement scale (P1880) is earthquake (Q7944) => Richter scale (Q38768), but the inverse would not be an example of measured physical quantity (P111). So I think the inverse property (P1696) there is wrong. --Laboramus (talk) 08:33, 12 October 2015 (UTC)

Wikidata weekly summary #179


I recently updated to put in some distance data.

I doubt Wikidata has trainfans, but I did want feedback if the taggging approach used here was Ok before suggesting it was more widely used..

Reamaining issues :

  • Dataset for the distances given Eu databse rights
  • Wikidata Quantity type doesn't as far as I know support multi-unit ( I'm using Km here), but for other rail-disatnces I'd have to use miles and chains, which would have to be converted to a single figure first ( which is cumbersome, NB All Mainline rail distances in the UK are miles and chains not km.)

Thanks ;) Sfan00 IMG (talk) 16:01, 8 October 2015 (UTC)

Wikidata:Units is a good place to discuss unit problems. Snipre (talk) 16:59, 8 October 2015 (UTC)
Yes, we have rail fans on Wikidata: User:Thryduulf, for instance. Perhaps we should have a trains project here on WIkidata? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:29, 8 October 2015 (UTC)
Assuming you can get the data out of Network rail or TfL i don't see any objection to a UK rail project here :)

(On that front I think Network Rail were considering releasing some data under "Open" terms...) Sfan00 IMG (talk) 21:37, 8 October 2015 (UTC)

I wouldn't limit it to just the UK, lots of things will apply to trains/rail more generally and a broader project is likely to attract more people too (I've added a few stations and light rail lines, but not UK ones, so I would add a more general trains wikiproject to my watchlist but probably not a UK-specific one). - Nikki (talk) 13:04, 13 October 2015 (UTC)
In my opinion 'length'length (P2043) is not the same as distance and should not be used to express the distance to the next station. I guess a more fitting property like 'distance' is not available yet? Michiel1972 (talk) 12:22, 13 October 2015 (UTC)
This came up on IRC, where I said length sounds really weird, but that I couldn't find any better property. If someone wants to propose a distance property, I would support it. - Nikki (talk) 13:04, 13 October 2015 (UTC)


I have some historical data that I'm trying to align with modern data, but it's only identified using USG/GNS UFIs. I see that we have GNIS ID (P590) for GNIS ids, and that helps me with my US data, but not with other countries. Does anyone have any suggestions for how I can align them? Thanks, Bovlb (talk) 03:19, 10 October 2015 (UTC)

Can you convert GNIS ids to Lat/Long? Sfan00 IMG (talk) 16:17, 10 October 2015 (UTC)
The GEOnet Names Server (Q1194038) includes Lat/Long for each entry for a GNS-UFI, however mostly only with an accuracy of 1 arc minute. With that and a query to find items around a coordinate it should be possible to find some matches. It would be a good idea however to propose a new property so we can store the GNS:UFI directly with each item, though it will be a big task to actually add all the statements then. Ahoerstemeier (talk) 08:39, 12 October 2015 (UTC)
Thanks. I guess I could match them up by some combination of lat/long (within a mile or so), fuzzy name match, and some sort of type match. If I do manage to do that, it would be good to have somewhere to dump the information.
I see that some of the GNS ids are recorded in English Wikipedia using Template:GEOnet, Template:GEOnet2 and Template:GEOnet3. Maybe there's some way to import them. Bovlb (talk) 23:59, 12 October 2015 (UTC)
I proposed a new property. Thanks for the suggestion. Bovlb (talk) 16:06, 13 October 2015 (UTC)

volume (P478) won't allow renaming (heading <-> alias)

With the recent script change to not allowing a duplicate of property and alias, there is an issue. Someone has recently changed the title of P478 from volume to volume (book) which needs to be named back to "volume". It does not allow it given the error to say that it current exists. I am unaware whether it is just as the leading word existing in another alias or whether it was a memory of a previous version. Assistance requested.  — billinghurst sDrewth 03:44, 12 October 2015 (UTC)

I fixed it, I think the label was taken by volume as quantity (P2234) which I have now renamed. --BurritoBazooka (talk) 04:09, 12 October 2015 (UTC) edited to add: I think whatever script you are referring to, User:billinghurst, actually prevents any two properties from having the same name in any particular language. --BurritoBazooka (talk) 04:13, 12 October 2015 (UTC)
I think it would be better if volume (P478) also got an unambiguous label, like “volume of work”. —Galaktos (talk) 09:33, 13 October 2015 (UTC)
"volume number" might work too. - Nikki (talk) 12:05, 13 October 2015 (UTC)
"volume for quantity" doesn't make any sense to me :/ - Nikki (talk) 12:05, 13 October 2015 (UTC)

Feed list of articles to

Is it possible to use tool on specified list of articles? I am able to generate list of wikidata entries without P18 property for entries in my region - where I have local knowledge allowing better verification. Also, I have no interest to spend time on adding images of (for example) television presenters but I am willing to spend time to improve data used in my project Mateusz Konieczny (talk) 15:27, 13 October 2015 (UTC)

See also Mateusz Konieczny (talk) 15:29, 13 October 2015 (UTC)
@Mateusz Konieczny: Try --- Jura 16:06, 13 October 2015 (UTC)
@Jura1: - thanks, it works perfectly! Mateusz Konieczny (talk) 18:48, 13 October 2015 (UTC)

Mental illness

At Mabel Boll Q6721414 I have added medical condition (P1050): mental disorder (Q12135) since she died at a mental institution. There are two references in her Wikipedia article. Two people keep deleting it without engaging on the talk page. There are clearly no BLP concerns. By deleting it we are saying there is a stigma to mental illness and it should never be mentioned (like they do in Japan for cancer). It is no different than saying she had diabetes. If she was there for Alzheimer's disease and that was listed as "medical condition (P1050)", no one would question it, but Alzheimer's is a mental illness. If you do not like "mental disorder" then get rid of it from Wikidata and come up with a pleasant euphemism like "brain disorder". --Richard Arthur Norton (1958- ) (talk) 21:24, 4 October 2015 (UTC)

Try adding point in time (P585) as a qualifier. People might be a little more comfortable it was clearer that the data was specific to a particular point in time, rather than a lifelong condition. Jheald (talk) 21:50, 4 October 2015 (UTC)
and a reference or two. Joe Filceolaire (talk) 21:29, 5 October 2015 (UTC)
Why is Alzheimer a mental illness ? It has all the hallmarks of an ordinary disease ? Thanks, GerardM (talk) 11:32, 14 October 2015 (UTC)


Two questions.

  1. Is it already possible to add units with quantity?
  2. Have coordinates, that should be imported to Wikidata. If I add them with QuickStatements, the precision is too good. Ideas on putting the right precision? --Edgars2007 (talk) 10:21, 12 October 2015 (UTC)

@Edgars2007: When you add a number, then a pop-up will allow you to choose an appropriate unit (See: Wikidata:Units). - About the coodinates: I hope I understood your question: Wikidata is not able to cut away zeros to indicate precision. So the value you added has to be read like a mathematical number and not like a measurement that turned out to have a lot of zeros. But I think work is being done to improve this. --Tobias1984 (talk) 10:32, 12 October 2015 (UTC)

Oh, wasn't clear enough. Is it already possible to add units with quantity with QuickStatements? --Edgars2007 (talk) 10:34, 12 October 2015 (UTC)
Maybe @Magnus Manske: knows. --Tobias1984 (talk) 07:25, 13 October 2015 (UTC)
Not yet. --Magnus Manske (talk) 09:55, 13 October 2015 (UTC)
@Magnus Manske: OK, but what about coordinates? And are units in your plans? --Edgars2007 (talk) 19:36, 13 October 2015 (UTC)
Just to note I would really like this too - I have used QuickStatements a lot (especially the past few days), it's great, but allowing it to add quantity data would be even better. I have about 3000 nuclear isotope half-lives I'd like to get in from the nndc database; I don't want to have to do those by hand! Though maybe a bot would be the right thing here, I'm looking into that too. ArthurPSmith (talk) 15:19, 13 October 2015 (UTC)
You can add unit-less quantities and coordinates, as detailed in the manual. Quantities with units are on the to-do-list. The very, VERY long to-do-list. --Magnus Manske (talk) 22:02, 13 October 2015 (UTC)
Magnus, I really know, how to add coordinates and quantities. The problem with coords is this (which I suppose is in the tool, not in Wikidata or elsewhere). When I add them with Quickstatements as "@57.85/24.35", I get too precise coordinates — "57°50'60.000"N, 24°20'60.000"E" (see this version). I then have to manually set up precision. --Edgars2007 (talk) 11:58, 14 October 2015 (UTC)

Best way to store Commons Category ?

In most entries the related Commons Category is stored as P373, for example in
Palazzo Abatellis Q1255652
In other entries there is the Commons Category stored in the "Other sites" section, for example in
San Giuseppe dei Teatini (Palermo) Q649685
Which way is the better/actual one? -- 06:38, 14 October 2015 (UTC)
This is disputed. I still advocate the sole usage of Commons category (P373); unless, of course, the item is about a category itself, in which case it should be linked directly. Jared Preston (talk) 06:43, 14 October 2015 (UTC)
If I understand your answer right, the second example San Giuseppe dei Teatini (Q649685) should be changed? Do You have an example, where a wikidata item is about a commons category itself? -- 08:25, 14 October 2015 (UTC)
Yes, you understand correctly. I meant, for example, Category:Germany (Q1410828). The item is solely about categories, therefore it is OK to link to commons:Category:Germany. Jared Preston (talk) 08:44, 14 October 2015 (UTC)
Latest discussion appears to be here: Wikidata:Requests for comment/Category commons P373 and "Other sites". Note that if only Commons category (P373) is used then Commons does not pick up the site links for its "In Wikipedia" sidebar section (which is how that project uses site links). - AdamBMorgan (talk) 09:22, 14 October 2015 (UTC)
Although if Commons category (P373) is present, then the script wdcat.js can be used to show a link to Reasonator. Jheald (talk) 09:28, 14 October 2015 (UTC)

StrepHit IEG proposal: last call for support

Dear all,

This is a last call for supporting the StrepHit IEG proposal before the formal review period (October 20th):


StrepHit is a Natural Language Processing pipeline that harvests structured data from raw text and produces Wikidata statements with reference URLs from reliable sources.
We have already received lots of precious endorsements for which we are grateful, but your voice is crucial and we are missing yours!
If you like the idea, please consider clicking on the endorse blue button on the project page.

Looking forward to your updates. Cheers,

--Hjfocs (talk) 13:27, 14 October 2015 (UTC)

Sitelinks and missing P569

At sitelinks and P569, a short status on how many items for people ( instance of (P31) = human (Q5) ) currently have date of birth (P569).--- Jura 13:28, 13 October 2015 (UTC)

Nice Jura! Is "all" just a sum or does it take into account that some items are linked to multiple languages?
Maybe nice to make a list of items without date of birth sorted by the number of sitelinks descending? Fixing a couple of those should have a nice impact. Multichill (talk) 09:24, 14 October 2015 (UTC)
There is obviously some overlap, but I found there is less in items about people, than I had initially thought (People_charts#Resources).
It might even be that items with overlap are more likely to have P569 already defined.
I will do such a list, but personally, I find the unique ones more important. --- Jura 09:49, 14 October 2015 (UTC)
Here we go: User:Jura1/test569 --- Jura 10:29, 15 October 2015 (UTC)
And a selection with sitelinks to nl: User:Jura1/test569/nl. --- Jura 10:47, 15 October 2015 (UTC)

Plans for Quantities with lower/upper border?

The Quantity data type stores a value, lower bound, and upper bound. However, this is only exposed as a single variance – i. e., you can only enter values where the value is right between lower and upper bound, e. g, 50±2 or 1000±10. This has led to property proposals for qualifiers “upper bound” and “lower bound” to store asymmetric ranges of accepted values. I think this would be a mistake, since it seems to me the Quantity data type is intended to provide this already – but it doesn’t do that. Are there any plans for supporting explicit value ranges in quantities (e. g., 50 (47–52) or 1000 (992–1009))? —Galaktos (talk) 09:21, 14 October 2015 (UTC)

Can't we fix different values for bounds using the API ? But even if you can put different values, you still have the problem of the distribution function. You have to specify that information to be able to use correctly the uncertainty data. This topic was already mentioned to the dev team but this is not the main priority now. Before starting with the creation of new properties about uncertainty a complete reflexion should be done about what kind of data are necessary in order to have a correct description. Snipre (talk) 11:37, 14 October 2015 (UTC)
There are two different things. An interval and an uncertainty. You need to be clear which you are referring to.
An interval needs two properties like start time (P580) and end time (P582). For most quantities we are interested in the length of the interval so that can be described by a length property or similar. If you come across a case where we actually need to specify the start point and the end point then you will need to make a Wikidata:Property proposal for these.
For the uncertainty associated with a measurement you can specify the +/- range. There is work being done to improve how uncertainties are represented but it is likely to take months before we get the fruit of that work. Hope that helps. Joe Filceolaire (talk) 22:31, 14 October 2015 (UTC)

I would like to add a logo to a company inside Wikidata but i don't have an image in the commons.

Good evening.

A client i'm following is having a problem on Google, because Google is using wikidata to fill the knowledge graph and at the moment there are two blocks, one related to the brand (Unipolsai) and another one related to an old brand (the new one is the result of a merger).

I've tried to fill all the relevant informations (in particular the old names of the companies) inside wikidata:

I have some doubts regarding the company logo, i've also asked on IRC and it seems that in order to be referenced, the logo has to be part of the commons.

As you may imagine, the logo is copyrighted and has been published inside wikipedia according to the exemption doctrine policy.

What would be the correct way to add proper "link" to the logo from wikidata that Google my understand ?  – The preceding unsigned comment was added by (talk • contribs).

Wikidata is not the main source of the Google Knowledge Graph. There is no other way to use the logo inside Wikimedia. Sjoerd de Bruin (talk) 16:16, 14 October 2015 (UTC)
@Sjoerddebruin: Do we know anything about this process? It seems like it would be good to have an information page (e.g. Wikidata:Wikidata and search engines), for the many people trying to come here to fix up a company profile. - Also: Is there maybe a Google page we can refer to, or is it just the "Feedback" button that is visible below the box? --Tobias1984 (talk) 16:26, 14 October 2015 (UTC)
Top secret information, sadly. Only the feedback button helps, I think. Sjoerd de Bruin (talk) 16:36, 14 October 2015 (UTC)

Help:FAQ/Freebase, see especially: (6) By adding to Wikidata, I have a free ticket into Google's Knowledge Graph, right? - No. --Atlasowa (talk) 17:15, 14 October 2015 (UTC)

@Atlasowa: Thanks. That is actually a nice summary of a lot of questions, but a little hidden under Help:FAQ/Freebase. Doesn't seem like a Freebase question, that people want to add some data about their company. --Tobias1984 (talk) 19:35, 14 October 2015 (UTC)
  • Post a link to the new logo here. Although trademarked, it may be ineligible for copyright and can use a PD-Textlogo license and be loaded to Wikimedia Commons. If it is a graphic it can be loaded into the Wikipedia article under a fair-use license. I am not sure how long it will take before Google replaces it and stops displaying the old one, but once added to the article it will be ready when Google updates. --Richard Arthur Norton (1958- ) (talk) 23:26, 14 October 2015 (UTC)
Thanks @Richard Arthur Norton (1958- ):! The logo is here  – The preceding unsigned comment was added by (talk • contribs). You can sign your posts using --~~~~. --Tobias1984 (talk) 11:32, 15 October 2015 (UTC)

Doubt about merging two elements

duo (Q15618652) and duo (Q10648343) are supposed to be the same and have to be merged or there is any reason to keep them as they are? I was going to merge them until I saw that duo (Q10648343) is subclass of (P279) of duo (Q15618652) and make me think that this can be intentional and maybe instead of merging theme the descriptions should be modified to make clearer the difference between them. -- Agabi10 (talk) 21:27, 8 October 2015 (UTC)

It looks to me like there might be some subtlety in the exact meaning of the various sitelinks, but I don't speak any of the languages, so I can't say for sure what it is. Popcorndude (talk) 01:03, 9 October 2015 (UTC)
Maybe User:Andreasmperu has the answer, he reverted the last said to be the same as (P460) statement on this issue. --BurritoBazooka (talk) 03:14, 9 October 2015 (UTC)
It's worth going into Reasonator and looking at what subclasses the two items have. There are subclasses of "two people" that may not be appropriate for "duo". ("Duo" to me implies a slightly closer connection -- eg two people performing together, or being very closely associated together in some common activity).
But looking at the sitelinks, they may be worth concentrating on "duo"; and some of the "instances of" could, I think, be refined from "two people" to "duo". Jheald (talk) 08:14, 9 October 2015 (UTC)
I think sibling duo (Q14073567) and subclasses need a bit of clean up: Not all twins act together in the sense of "duo". However one could argue that twins who are not a duo are not notable here (but what if they are notable here in an hypothetic case where they are victims of a famous murder?). -- Gymel (talk) 10:37, 9 October 2015 (UTC)
I'd say it does not matter as "twin" is unquestionably a notable concept. author  TomT0m / talk page 10:41, 9 October 2015 (UTC)
@Agabi10, BurritoBazooka: Unfortunately Andreasmperu reverts corrections without explanation and is not answering on his user page. (BTW: He is also fast in merging without looking at talk pages or labels other than English, see Talk:Q3511312#merge.) --Kolja21 (talk) 23:37, 14 October 2015 (UTC)
I will try to bring this section to his attention on his enwiki talk page. @Andreasmperu: --BurritoBazooka (talk) 00:10, 15 October 2015 (UTC)
duo (Q15618652) is about a group of humans (Q16334295) composed of two members, whereas a duo (Q10648343) is a group of two people that work together, so it is a subclass of duo (Q15618652). Btw, I have not been that active for the past weeks, and I will probably will not be able to edit regularly for the next two weeks or so. @Kolja21: Undoing an edit is not the same as reverting it: reversions are to be used only for vandalisms. Also, I am "she" (quite unusual here, I know). Andreasm háblame / just talk to me 18:00, 15 October 2015 (UTC)
Undoing an edit is not the same as reverting it: reversions are to be used only for vandalisms. – are you referring to rollback? AFAIK “revert” usually refers to undo, not rollback. —Galaktos (talk) 19:33, 15 October 2015 (UTC)

Founded by vs Founder of

We have "Founded by" for organizations, but what is the equivalent to be added to the biography to link to the organization? What would be "Founder of"? I tried looking but could not be found, just job title like president and CEO for organizations. --Richard Arthur Norton (1958- ) (talk) 23:32, 14 October 2015 (UTC)

With a few exceptions, we tend not to do reciprocal properties. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:19, 15 October 2015 (UTC)

website account on vs individual properties

We have website account on (P553) with qualifier website username (P554) and we also have different properties for different social networks like for example P2035 (P2035), Facebook ID (P2013) or Twitter username (P2002). I would like to know which way is the preferred and why we are maintaining both of them. It's confusing for the users and hard to prevent duplicated information. -- Agabi10 (talk) 21:20, 15 October 2015 (UTC)

Where a dedicated property exists it is preferred.
We don't want to have to create dedicated properties for every single social network, no matter how small, so we also have the generic property website account on (P553) for use with other social networks. In those cases use website username (P554) as a qualifier to indicate the username on that network.
Note that the Property pages for each of the specialised properties should each have a <subproperty of:website account on> claim so that when we finally get a "search (including subproperties)" function we can search on "P553 including subproperties" and get all the different social networks with one search. Joe Filceolaire (talk) 22:14, 15 October 2015 (UTC)
So if the preferred one is the dedicated one and both of them should exist to prevent creating properties for all the elements... Should we add a conflicts constraint to show in the report if the property website account on (P553) itself is used with any of the social networks which have their own property? And maybe also requesting a bot to change the current violations of the constraint to their own property? -- Agabi10 (talk) 22:42, 15 October 2015 (UTC)
The conflict constraint already exists, see Property talk:P2002 and Property talk:P2013 --Pasleim (talk) 10:14, 16 October 2015 (UTC)
If you are interested in Youtube, you could oppose my proposal at Property_proposal/Generic or participate in the three property proposal discussions in a meaningful way. --- Jura 12:19, 16 October 2015 (UTC)
I think we should create as many specific properties as necessary and avoid website username (P554) whenever possible. We now have statements on properties, we will soon have a dedicated datatype for external identifiers. Specific properties are decidedly more usable and more maintainable than website username (P554), even for small websites. --Zolo (talk) 12:34, 16 October 2015 (UTC)

@Pasleim: That's not what I mean, well, not exactly. I was talking about that constraints but used in website account on (P553). The constraints you talk about check if there is duplicated info, the ones I am saying checks that website account on (P553) is not used for specific sites. -- Agabi10 (talk) 13:58, 16 October 2015 (UTC)

Best practice in statments of article (Q191067) itens

Hy I would like to know the best way to give bibliography information article (Q191067) items.

I update Epigrams on Programming (Q5382930) to show volume, issue, page and date as qualifiers to published in (P1433) and just notice that El periodismo taurino de 1898 (Q17516879) state all this information as direct statements on the item itself.

What is the best way to input this data: as qualifiers or claims? Or both?

Any help is appreciated. Carlos Porto (talk) 02:22, 16 October 2015 (UTC)

I think the way you've been using is better, since it's technically possible for one article to be published in more than one work. --BurritoBazooka (talk) 05:09, 16 October 2015 (UTC) ETA: I usually use the statement published in (P1433) and then add volume, issue, page, etc information as references, not qualifiers. I'm also not sure what the 'standard' here is. --BurritoBazooka (talk) 09:52, 16 October 2015 (UTC)
If the text is re-published online, shouldn't there be an additional item for the online edition? --- Jura 12:06, 16 October 2015 (UTC)
what methodology would provide cleaner code to query for the information? As qualifiers or references? Carlos Porto (talk) 17:54, 16 October 2015 (UTC)

Help me understand WDQ syntax, or a bug in WDQ?

Hi, the following WDQ query:


returns the items The Assassination And Death Of Abraham Lincoln (Q19832974) and Dupuy de Lôme Letter (Q5244605), even though neither of those items have a country of citizenship (P27) value of Britannia (Q3240725) or Kosovo (Q1231).

On the other hand, this query, which I expect to behave the same:

 claim[27:3240725] or claim[27:1231]

returns nothing. Both Autolister and Wikidata's own API give the same results, so I think the problem is with Wikidata's own API because that's what Autolister uses. I'm confused, because if I understand the documentation correctly, both should yield nothing since neither of those items even has a country of citizenship (P27) statement. What am I reading wrong? Or is there a bug? Thanks. --BurritoBazooka (talk) 05:04, 16 October 2015 (UTC)

try claim[27:3240725,27:1231]. Your query returns all items with with country of citizenship (P27)=Britannia (Q3240725) or P1231 (P1231). --Pasleim (talk) 10:11, 16 October 2015 (UTC)
The first sample should be claim[27:3240725,27:1231] to be equivalent. Otherwise you get P1231. --- Jura 10:12, 16 October 2015 (UTC)
Well .. seems it was answered twice at the same time ;) --- Jura 10:13, 16 October 2015 (UTC)

Primary Sources tool showing additional MusicBrainz work IDs

Symphony No. 3 (Q223533) already has a MusicBrainz work ID (P435) statement, but the Primary Sources tool is suggesting nine more statements with different IDs. The MusicBrainz web page for all of them looks completely identical. Should I accept these statements? Is it usual for MusicBrainz to have multiple IDs for the same work? —Galaktos (talk) 21:20, 15 October 2015 (UTC)

The general principal is that we keep all the IDs since all correctly link to MusicBrainz pages. When MusicBrainz finally sort themselves out and redirect nine of these to one canonical page, then we still keep all ten IDs but mark the ones that MusicBrainz has redirected with rank "Deprecated". This is so that years from now some strange database which never got around to updating can still find the relevant wikidata item. At least that is how I understand it. Joe Filceolaire (talk) 22:26, 15 October 2015 (UTC)
...When Wikidata finally sorts ourselves out and activates that rank feature ;) --BurritoBazooka (talk) 05:21, 16 October 2015 (UTC)
Alright, thanks. —Galaktos (talk) 10:44, 16 October 2015 (UTC)
@BurritoBazooka: I don't get your point. The rank feature is already activated. --Casper Tinan (talk) 14:03, 16 October 2015 (UTC)
@Casper Tinan: Ah, I'm wrong then. Last time I tried it (which was about a month ago when I first started using Wikidata), the rank buttons did nothing in my browser. Now they do. --BurritoBazooka (talk) 21:58, 16 October 2015 (UTC)
On phab:T115112 someone thinks that marking them as deprecated would be the wrong semantics. Personally I don't see any point in manually checking and adding redirecting IDs, as long as we have the non-redirecting ID, the redirects can very easily be imported by a bot once we've figured out the right way to store (and display) them, which is why I asked on Wikidata_talk:Primary_sources_tool#Redirecting_MusicBrainz_IDs for them to filtered out. - Nikki (talk) 14:21, 16 October 2015 (UTC)
Also, the IDs do already all point to the same page, the canonical ID is always used (and therefore the page content is always identical) regardless of which ID is used in the URL. There's just no HTTP redirect to change the URL. - Nikki (talk) 14:31, 16 October 2015 (UTC)

Remember the bot-import of a quarter million dutch streets? Now guess what...

Wikidata is really fucked up. Now also poisoning VIAF.

Meanwhile at occitan Wikipedia they are still looking for help to clean up the mess by botcreated wikidata-stubs from 2 years ago, which led to half (~40.000) of ocWP articles as "Pages_with_script_errors" after P107 deletion. FUBAR. --Atlasowa (talk) 20:47, 16 October 2015 (UTC)

Isn't it VIAF's algorithm who's bad? It's easy to blame the street items. Sjoerd de Bruin (talk) 20:50, 16 October 2015 (UTC)
Indeed. It's not just Dutch streets which VIAF has created incorrect matches for (the one I remember was something in Jamaica which VIAF linked to something in the UK). If other people do stupid things like creating matchings to Wikidata based on nothing more than the two things having the same name, it can hardly be blamed on us. - Nikki (talk) 22:16, 16 October 2015 (UTC)
There's more examples at Wikidata_talk:WikiProject_Authority_control#Locations. - Nikki (talk) 22:24, 16 October 2015 (UTC)


choreography (Q180856) and choreography (Q3764113). --Richard Arthur Norton (1958- ) (talk) 21:24, 16 October 2015 (UTC)

They can't be merged because they both have links to enwiki, eswiki and tewiki. It seems that the first one is more general than the second, and the second is already marked as a subclass of the first one. - Nikki (talk) 21:38, 16 October 2015 (UTC)

Wrong disambiguation items

These items has wrong disambiguation links and wrong claims. please help to clean themYamaha5 (talk) 11:18, 16 October 2015 (UTC)

The list is even a bit longer, because you made a typo. I've edited your link to correct your typo. Mbch331 (talk) 11:26, 16 October 2015 (UTC)
No, your query is too strict. For some disambigs it makes sense to have such statements. So the clean-up is not needed. --Infovarius (talk) 12:08, 16 October 2015 (UTC)
I checked some items and there is much to do, but it is not a bot job. Items are linked both, to Disambiguation pages and article pages. Or only to articles. --Diwas (talk) 19:48, 17 October 2015 (UTC)
@Infovarius, Diwas: I didn;t say we should removed or edit blindly or by bot. by sure users should check links and claims Yamaha5 (talk) 13:05, 18 October 2015 (UTC)
to sort items by local wiki we can use link[xxwiki] like this (I done [fawiki]) Yamaha5 (talk) 13:59, 18 October 2015 (UTC)

Formatter URLs broken?

formatter URL (P1630) seems to be broken… for all kinds of identifiers, I only see the value of the identifier in statements, not a proper link to the formatter url with the value filled in. Is anyone else experiencing this? —Galaktos (talk) 17:07, 17 October 2015 (UTC)

Can you give an example? I checked a couple of random entries and it seem to work fine for me. --Laboramus (talk) 19:00, 17 October 2015 (UTC)
Does refreshing the page help? - Nikki (talk) 19:37, 17 October 2015 (UTC)
@Laboramus: Example: Mona Lisa (Q12418). Atlas ID (P1212), Joconde work ID (P347), GND ID (P227), VIAF ID (P214), Bibliothèque nationale de France ID (P268), P727 (P727), Freebase ID (P646), Library of Congress authority ID (P244), IdRef ID (P269), Libraries Australia ID (P409) all have a formatter URL (in fact, Bibliothèque nationale de France ID (P268) has two – not sure what that’s supposed to mean :D ). None of the statements are links for me.
@Nikki: No, neither refresh nor purge solve the problem for me.
I take it that neither of you are experiencing this, so I’ll try fiddling with my add-ons, settings etc., to figure out what might be causing this… —Galaktos (talk) 21:32, 17 October 2015 (UTC)
The links work in a private window, i. e. when I’m not logged in. Phew, not a browser add-on problem :) let’s play with my settings next. —Galaktos (talk) 21:36, 17 October 2015 (UTC)
@Galaktos, Nikki: It also happens to me, but not always, it's not even in specific pages. Sometimes it solves refreshing once and sometimes I refresh 4 or 5 times and it still is not a link. I don't know what can it be the problem and I am not able to replicate at this moment, but for me it happens everyday at some moment, specially with IMDB and Steam identifiers. (But that doesn't mean that they are the most commons ones to failure, they are the ones I check most...) -- Agabi10 (talk) 21:38, 17 October 2015 (UTC)
I can’t figure it out. Disabled all userscripts, still broken. Disabled gadgets (some at a time), changed nothing. Disabled beta features, same. Not sure what else could cause it. —Galaktos (talk) 21:55, 17 October 2015 (UTC)
@Agabi10: I'm not currently having a problem with the links, so I can't say for sure whether it's the same problem or not, but I've been having similar issues with other scripts for a few months and I just created phab:T115794 for the issues I'm having. - Nikki (talk) 21:56, 17 October 2015 (UTC)
@Galaktos: I'm not currently having a problem with the links, no, but I am having problems with other scripts (see my reply to Agabi10 above). - Nikki (talk) 21:56, 17 October 2015 (UTC)
FWIW, I also frequently find that the formatter URLs don't display when the page is first loaded (especially if it's loaded after they've just been added). Reloading usually but not always solves it. Never traced a cause... Andrew Gray (talk) 22:14, 17 October 2015 (UTC)

More than one history?

I would like to get some attention to a Q about history of topic (P2184) here! -- Innocent bystander (talk) 14:26, 18 October 2015 (UTC)

Gadget/script for local search

I'm wondering if here on Wikidata exists some gadget or user script that creates one more search box for searching only local wikidata-pages, like the ones from Wikipedias, also with search suggestions feature enabled? --XXN, 23:14, 13 October 2015 (UTC)

Killed in action

We have "military casualty classification" but what would the "cause of death" and "manner of death" be for people "killed in action". I have been looking at Category:American military personnel killed in World War I and wondering how best to classify them for both "cause of death" and "manner of death". Those fields are empty. I can add "military casualty classification" and fill it with "killed in action" but the other two fields will be empty. --Richard Arthur Norton (1958- ) (talk) 17:58, 16 October 2015 (UTC)

A claim needs to be added to military casualty classification (P1347) to mark this as a 'subproperty of (P1647)' either cause of death (P509) or manner of death (P1196). That way, when we have a 'search including subproperties' function we will be able to find these cases even if the item only has a "military casualty classification" claim. Joe Filceolaire (talk) 10:48, 19 October 2015 (UTC)

"outline of ..."

There is a bunch of articles who are entitled oultine of something on enwiki. Do we have guidelines or ideas here on how to deal with it ? <Aspect of: something> ? author  TomT0m / talk page 13:29, 17 October 2015 (UTC)

@TomT0m: There are enough of them that it is probably worth creating a sui-generis class for them.
As to the best way to describe them, one approach might be use is a list of (P360) 'topics'/'concepts', 'related to' subject X. Jheald (talk) 14:33, 18 October 2015 (UTC)
Jheald: This assumes that these items have no meaning as 'real world' items and should go in our separate "wikimedia only" classification.
In practice most list articles represent 'class' items and should use the 'subclass of' property to describe them and these can be linked to by properties like 'instance of', 'family name', 'office held' etc.
The 'Outline' articles are different however in that they are not a class of similar things; they are rather a list of parts related to an item. This would indicate that maybe we should use 'part of' to link other items to the 'Outline of' items. This brings us to another issue. 'Outline of' articles are, in nearly every case, semantically duplicates of other articles. By this I mean that we have articles on 'Botswana' and 'Outline of Botswana'. For anything that might have the claim <part of:Outline of Botswana> it is probably more useful to add the claim <Country:Botswana> instead.
Which brings me back to say Jheald is probably right. 'Outline of Botswana' should just have the claim <outline of:Botswana> (where 'outline of' is a new property) because the 'Outline of..' items are, semantically, duplicates so they should be coralled off in a 'wikimedia stuff' section. TomT0m I hope that helps. Joe Filceolaire (talk) 10:34, 19 October 2015 (UTC)

Names without diacritics - a.k.a. entry or something else?

I notice that is if you search for Michaela Pavlícková at you will get a "not found" but if you search for Michaela Pavlickova, you will get a hit:Michaela Pavlickova.

Obviously, they've adopted the convention to drop diacritics. Does this mean that "Michaela Pavlickova" Should be listed as an entry in wikidata in the "also known as" field or is it handled differently or not at all?--Sphilbrick (talk) 16:00, 17 October 2015 (UTC)

The search function on Wikidata should already handle this… you can find Michaela Pavlíčková (Q2052957) by searching for either “Michaela Pavlíčková” or “Michaela Pavlickova”. —Galaktos (talk) 17:08, 17 October 2015 (UTC)
Thanks.I take it this means we should not enter the name without the diacritics in the a.k.a. field.--Sphilbrick (talk) 17:55, 17 October 2015 (UTC)
@Sphilbrick: No, the aliases should stay and the name without diacritics should be put in the alias field. This is because even if the software can find the names even if they have diacritics when suggesting fields it will suggest first the most similar ones. Putting as an alias the name without diacritics makes that even if it exists something less connected with the same name without diacritics the most used ones appears at the top. For example it has been problems in Spanish with people adding medico (medicate) as occupation instead of médico (doctor) because medico appeared first and médico was hidden under more elements starting with the character sequence "medico". With the alias now appears the first one because it is one of the most used ones. -- Agabi10 (talk) 01:10, 18 October 2015 (UTC)
Does the software separate what is a diacritics and what is not? When I search for "Osten", I still get results for "Östen". In Swedish, Ö is not O with diacritics, it's a completly different letter. It's like answering "No, we did not found 'Bell', may I suggest 'Well' instead?" But many search engines are far far worse than the one we have here. -- Innocent bystander (talk) 11:54, 19 October 2015 (UTC)

Wikidata weekly summary #180

University/Hochschule and User:Infovarius

I'm not sure what to do. User:Infovarius has twice reverted edits on university that I made as the superclass higher education institution made no sense (see the wikipedia links for Hochschule, which refer specifically to a German/Scandinavian type of institution, and one which is explicitly noted does NOT include even many German universities). I tried to engage User:Infovarius here - User_talk:Infovarius#University_is_NOT_a_subclass_of_Hochschule and got no response. Suggestions on what to do? Is this an issue with higher education institution meaning different things in different languages? ArthurPSmith (talk) 15:04, 28 September 2015 (UTC)

Russian terminology:
ru:университет (university, Q3918) is subclass of ru:высшее учебное заведение (higher education institution, Q38723);
ru:высшее учебное заведение (higher education institution, Q38723) is subclass of ru:образовательное учреждение (educational institution, Q2385804)
Ivan A. Krestinin (talk) 16:27, 28 September 2015 (UTC)
Ok then - "higher education institution" makes sense as a label - but it is clearly a different thing from the German term "Hochschule". Also there doesn't seem to be an English wikipedia page that corresponds to "higher education institution" exactly. Anyway, I suggest we need to disconnect the two meanings from different languages under Q38723. I'm willing to do this but maybe somebody who is familiar with more of the languages linked would be better? ArthurPSmith (talk) 17:15, 28 September 2015 (UTC)
Related discussions: Talk:Q38723Ivan A. Krestinin (talk) 17:37, 28 September 2015 (UTC)
@Infovarius: reverting someone twice and not responding to talk page messages? Not cool. Please be kind to other users.
@ArthurPSmith: this is one of those fun modeling jobs. We have multiple countries and multiple languages all looking the same, but actually a bit different. Let's take Hogeschool van Amsterdam (Q945017) and University of Amsterdam (Q214341). The first one is what we call in Dutch a "Hogeschool" (like Hochschule) and the second one a university. In the Netherlands a university is definitely not a subclass of a "Hogeschool". To fix this we'll probably have to create new well defined classes and interconnect these. Take a look at en:Bologna Process for quite a bit of background information. Probably first step is to split higher education institution (Q38723) into the more generic "higher education institution" and the more specific Hochschule. Multichill (talk) 17:47, 28 September 2015 (UTC)
In Swedish (sv), you probably have to split the term "Högskola" in how it is used today, in the 20th century and earlier. It is a term is constant change. -- Innocent bystander (talk) 17:51, 28 September 2015 (UTC)
In Danish I would say "higher education institution" can be a superclass for "universitet" (University) and "professionshøjskole" (Vocational university/Fachhochschule). vocational university (Q17028020) and higher education institution (Q38723) is unfortunately a mess in the Danish part of Wikidata at the moment. — Finn Årup Nielsen (fnielsen) (talk) 17:55, 28 September 2015 (UTC)
We could just take the mayor of a place in the Netherlands (Q13423499) approach and create separate items for each type of school in each country and put these items in a tree structure. Multichill (talk) 18:12, 28 September 2015 (UTC)
Well, I hope this is ok - I've created Hochschule (Q21028957) and linked all the language wikipedia links that seemed right. That leaves behind higher education institution (Q38723) which probably should have some replacement links added and revisions of labels in other languages. ArthurPSmith (talk) 19:28, 28 September 2015 (UTC)
I am sorry for long non-responding: it's quite hard for me to track all lines of wiki (and offline)-activity... And two reverts, if they separated by several days, may look to me as very different activities too... But I like the solution that was made by Arthur by now. --Infovarius (talk) 21:32, 29 September 2015 (UTC)

Dear colleagues, thanks to ArthurPSmith to invite me here and to Infovarius to attract my attention by reverting my edits. I am afraid of misunderstanding of the German term as this discussion is not visited yet by anybody from the German Wikipedia. So let me contribute to find the best solution.

Please look at the definition in the German article de:Hochschule, saying: Hochschule ist der Oberbegriff für eine Einrichtung des tertiären Bildungsbereichs - literally: Hochschule is the generic term for an institution of tertiary education). The German term Hochschule and the other group of linked articles (including ru:высшее учебное заведение and bg:висше училище, etc.) represent the superclass for all institutions of higher education. These articles (de, ru, bg) are main articles in their eponymous categories, all of them linked to the superclass en:Category:Universities and colleges.
The problem has arisen because the linked articles in German, English, Italian wikipedias, named Hochschule, describe this German term, while the other articles, linked to them (fi, fr, it, nl, nn, no, sv), refer to translations, bearing narrower meanings - e.g. fr:haute école is not a generic term, as it represents only a type of institution of tertiary education. Simultaneously the Bulgarian term (висше училище) is an exact translation of Hochschule, has the same meaning of generic term, but its article (like those linked to it) stays alone, not linked to the German one!
Deleted later! The French haute école is the superclass term, comprising all the higher education institutions. --Elkost (talk) 11:00, 19 October 2015 (UTC)
The main principle in interlanguage links is to connect terms by meaning, not by sounding/writing - e.g. fi:Korkeakoulu and ee:Kõrgkool from very close languages (Finnish and Eesti), are here in different groups as they seem similar but have narrower and wider meanings accordingly. Therefore the right place for de:Hochschule is not in the group of the homonymous Germanic and translated terms (fi, fr) with narrower meaning, but in the group of generic terms for all institutions of higher education - bg:висше училище, etc. --Elkost (talk) 14:32, 10 October 2015 (UTC)
@Elkost: @Infovarius: @Multichili: - but Elkost, the enwiki article en:Hochschule only describes the German term as you translate it because you just edited it to state that. It previously indicated rather than being "the" German generic term, it was "a" generic term. And the enwiki article still has a section titled "Specialised term" which describes a meaning that is quite different from the generic one. The dewiki article de:Hochschule doesn't say quite the same thing, but it does include this diagram which shows several types of post-secondary institutions that are NOT considered to fall under the term Hochschule. So at the least it is a German word that describes something generic associated with post-secondary education, similar to the word "College" in English, but it does not encompass all post-secondary education, and should not be considered a stand-in for the true generic term "higher education institution". At least that's my view. No English-speaking person would understand why the term "Hochschule" would appear in a generic hierarchy of education above University or College. ArthurPSmith (talk) 13:53, 12 October 2015 (UTC)
A note to correct my spelling of @Multichill: - but also to more directly address the comments of Elkost - the specific example fr:haute école looks to me to be describing exactly the meaning of the term Hochschule as translated into French, as it is talking about these types of institutions in specific countries. It does not refer to any French institutions (you may be thinking of fr:Grande école which is a specific type of French institution).
Oh, yes, I've meant grande école, speaking about haute école! Thanks for the notice! --Elkost (talk) 11:00, 19 October 2015 (UTC)

Dear friends, please pay attention to what I have mentioned above: These articles (de, ru, bg) are main articles in their eponymous categories, all of them linked to the superclass en:Category:Universities and colleges. This means that the articles (and categories) of all kinds of higher education institutions (incl. universities - Kategorie:Universität, etc.) are together in the superclass Kategorie:Hochschule, which is linked to en:Category:Universities and colleges. This link indicates quite clear that for German speakers Hochschule is equivalent to Universities and colleges as generic term for all types of higher schools, including the universities.

The most fast and easy way is to look at de:Kategorie:Hochschule - it includes as subclasses (sometimes overlapping each other) the categories Universität, Medizinische Hochschule, Christliche Hochschule, etc.
The diagram concerned in the article de:Hochschule shows some types of higher education institutions in Germany (incl. Universität, etc.) with heading Hochschule (in bigger letters and spacing) as their generic term. You can see more types in section "Formen der Hochschulen" (i.e. Formes of Hochschule), including once again Universität.
Obviously the "Specialised term" section in en:Hochschule does not present correctly the matter, opposing Hochschule to Universität as independent types of higher schools (in fact they correlate as general to particular). The schools concerned are called Hochschule because, not being Universität and without such specific common name, they still remain Hochschule (i.e. higher education institution). --Elkost (talk) 13:21, 19 October 2015 (UTC)
But "Fachschulen" and "Berufsakademien" are higher-education (tertiary) institutions but NOT considered "Hochschule"; therefore even by the logic of the dewiki page, Hochschule cannot be the generic term covering all higher education institutions, it covers only a subset. Whether or not it includes universities is not the issue, the question is whether it should be linked to the generic term for all such institutions, or be a subclass. It seems clear to me it should be listed as a subclass, not as a synonym. ArthurPSmith (talk) 13:46, 19 October 2015 (UTC)
Tertiary (post-secondary) education is for adults, including: 1. higher education; 2. various types of training in schools, institutes, centres, etc. "Fachschulen" and "Berufsakademien" in Germany are considered only as tertiary, but not higher-education institutions. Therefore, just by the logic of de:Wiki, Hochschule is the generic term covering all higher education institutions.
Additionally: look at the definition in the first sentence of de:Universität, saying: Universitäten... sind Hochschulen mit Promotionsrecht, sogenannte wissenschaftliche Hochschulen... - literally: "Universities... are higher schools (Hochschulen) with doctoral degrees, so-called research higher schools (Hochschulen)..."
So for German-speaking people Hochschule is the generic term (above Universität) for higher schools, as well as for English-speaking people universities and colleges is the same generic term (above university). --Elkost (talk) 17:16, 19 October 2015 (UTC)
@Elkost: should we take this discussion somewhere else? There should be a solution that makes sense. Looking at the category pages you mention it seems to me that maybe we need to address Category:Universities and colleges (Q5741962) as well? "Universities and colleges" isn't necessarily synonymous with "Higher education institute" in the first place so maybe some reorganization there is needed anyway? ArthurPSmith (talk) 16:47, 19 October 2015 (UTC)
@ArthurPSmith: We have already explained our standpoints. Let us hear others. --Elkost (talk) 17:27, 19 October 2015 (UTC)
@Elkost: well nobody else seems to be commenting. Did you read the comments above from @Multichill: and @Innocent bystander: that described how the related terms in the Netherlands and Sweden are not as generic as you seem to think? As to your latest comment - "higher education institution" in English does not exclude vocational institutions, it is essentially synonymous with post-secondary or tertiary education - see en:Higher education. Perhaps the solution we are looking for would be to link Hochschule to "Universities and colleges" as a subclass of Higher education institution? But I don't think even that is quite right because the categories don't seem to overlap precisely between the different languages. ArthurPSmith (talk) 18:29, 19 October 2015 (UTC)
Well, the categories in Sweden do not even overlap within our own nation. The meaning of the terms have changed by time. And since the latest changes, happend less that two decades ago, nothing is safe. -- Innocent bystander (talk) 18:41, 19 October 2015 (UTC)
@ArthurPSmith: Did you read carefully my comments? I have never said that the related terms in the Netherlands and Sweden are generic, but just the opposite, look at this from above: ... the other articles, linked to them (fi, nl, nn, no, sv), refer to translations, bearing narrower meanings... (narrower than the generic term - in German, Bulgarian, etc.) --Elkost (talk) 11:49, 20 October 2015 (UTC)

I tried in vain to explain briefly definitions and correlations from de:Wiki, grounding also on my past experience as a lecturer in a great Bulgarian university where by duty I had to examine and compare higher education systems of countries (mainly in Europe). Therefore I strongly recommend to everybody interested to consult an other expert in higher education, speaking well both German and English. Sorry, I have no spare time to spend any more for this tough discussion - in time when I am starting a new, great business abroad. Anyway I highly appreciate your volunteer wiki-labour. All the best to everybody! --Elkost (talk) 11:49, 20 October 2015 (UTC)

Count the items of a property

Hello. One suggestion: can we have an automatic number under a property which will show how many items are in the property? If the property has for example more than 10. See [9], property "participant". I have to count all items one by one to find how many are there to check if I have put the right number of items. Xaris333 (talk) 23:44, 18 October 2015 (UTC)

I tried adding $("div.wikibase-statementgroupview.listview-item").each(function(){ var c = $(this).find(".wikibase-statementview").size(); if (c > 10) { $(this).find(".wikibase-statementgroupview-property-label").append("<br><span style='color:#777;font-size:smaller'>("+c+" statements)</span>") } }); to my common.js and that seems to do what you want. - Nikki (talk) 23:41, 19 October 2015 (UTC)

Nikki its working!! Many thanks!! Xaris333 (talk) 12:44, 20 October 2015 (UTC)

@Xaris333: Great :) If it doesn't work on a page, try reloading the page. I keep having problems with gadgets/common.js not working every time and refreshing usually helps (I also mentioned that in the #Formatter_URLs_broken.3F section above). If I find a way to make it always work, I'll let you know. - Nikki (talk) 16:27, 20 October 2015 (UTC)

Same items in a property

Just to learn: why we can put the same item twice or more in a property? Are there situtations that need this? Xaris333 (talk) 23:46, 18 October 2015 (UTC)

Example: someone could hold an office more than once (position held (P39)), with different start time (P580) / end time (P582) qualifiers. Like here. —Galaktos (talk) 23:50, 18 October 2015 (UTC)
I see. But what about here. Why can I have to instance of (P31) the same item twice human (Q5)? Xaris333 (talk) 23:56, 18 October 2015 (UTC)
The question is: why did you add it twice? A bot eliminates such statements once in a while. --- Jura 00:39, 19 October 2015 (UTC)
By mistake for example. Xaris333 (talk) 08:53, 19 October 2015 (UTC)
Yes, also I think it is much better to put such statement twice then to put the qualifiers twice (which is unfortunately sometimes done) because in the latter case there is no matching between start/end pairs. Of course, humans can guess that if the qualifiers are something like 1920,1925 and 1921,1926 then it's probably 1920-1921 and 1925-1926 but it would make machine processing more complicated. So I think having the repeated statement with e.g. two position held (P39)) is much better. --Laboramus (talk) 07:37, 19 October 2015 (UTC)
I think Jura was pointing to the double human (Q5) in Angela Merkel (Q567) by Xaris333. In case of multiple times holding an office, there is no problem in adding it twice, with different values for the qualifiers. Mbch331 (talk) 08:38, 19 October 2015 (UTC)
  • I like that she has references to prove that she is human. Her opposition candidates probably demanded that. --Richard Arthur Norton (1958- ) (talk) 21:30, 19 October 2015 (UTC)

Error on merging

I tried to merge protoplanetary disk (Q20940243) with protoplanetary disk (Q505781) but it shows error. may it is a bugYamaha5 (talk) 19:28, 19 October 2015 (UTC)

Yes, see phab:T115892. You knew it was one if you followed the link given with the error. Sjoerd de Bruin (talk) 19:29, 19 October 2015 (UTC)
And I wanted to do some merge work :) Good old Murphy's law (Q188908). --Edgars2007 (talk) 16:43, 20 October 2015 (UTC)
@Yamaha5: This has been resolved, you should be able to merge items with sitelinks again. Hoo man (talk) 23:33, 20 October 2015 (UTC)

Wrong SUDOC authority links

Right now, a bot is adding IdRef ID (P269) by taking the links from Virtual International Authority File (Q54919). So far so good - however both of the Thai provinces which were added so far the link was wrong. E.g. for Chiang Mai (Q233588), the bot-added value in fact belongs to a different administrative unit (Chiang Mai Provincial Administrative Organization (Q21154966)). I have set the wrong value as deprecated and added the correct one, that should make our database consistent. Can we notify VIAF anyhow about such mistakes, or hope they will notice when importing a new Wikidata dump? Ahoerstemeier (talk) 22:22, 20 October 2015 (UTC)

They will notice changes in VIAF ID (P214). The question is however, if (or when) we notice the wrong VIAF number in our items and can correct them. Mass-importing other Authority numbers from VIAF (or sources mirroring data from VIAF) IMHO is highly problematic. IIRC the recemt actions of User:ShonagonBot have never been discussed or approved at Wikidata:Requests for permissions/Bot. -- Gymel (talk) 04:38, 21 October 2015 (UTC)

sr/sh labels

Which script are they meant to be in? Latin or Cyrillic? Is this defined somewhere? --- Jura 17:32, 14 October 2015 (UTC)

Maybe this answers it:

Serbo-Croatian (Q9301) has:

  • sh Srpskohrvatski jezik
  • sr српскохрватски језик
  • sr-ec српскохрватски језик
  • sr-el srpskohrvatski jezik

and Serbian (Q9299)

  • sh Srpski jezik
  • sr српски језик
  • sr-ec српски језик

Serbian Wikipedia (Q200386) and Serbo-Croatian Wikipedia (Q58679) have both scripts. --- Jura 17:51, 14 October 2015 (UTC)

Cyrillic seems to be used as the default script for Serbian (the Serbian Wikipedia usually uses Cyrillic for page names, and the logo is only in Cyrillic), which I guess would imply that the sr and sr-ec labels should always be the same (not sure it makes sense to have both if that's true). As for Serbo-Croatian... I have no idea. The Serbo-Croatian Wikipedia is a mixture of both Latin and Cyrillic (without any automatic conversion between the two like the Serbian Wikipedia has), but (from what I've seen) it uses Latin more often than Cyrillic. - Nikki (talk) 18:00, 14 October 2015 (UTC)
Nikki is more or less right. Since sr-el is always automated (from sr-ec), I don't know why it can't be here. I guess it's a technical thing. I know there was something wrong with adding "Serbian" in statements/qualifiers before... Jared Preston (talk) 18:17, 14 October 2015 (UTC)
Thanks for your help. I created items to store sr-ec and sr-el. --- Jura 13:15, 21 October 2015 (UTC)

"known alive for" / floruit

is floruit (P1317) really needed ? We can use birth and death date with a precision of a century, seems simpler to query and more economic. author  TomT0m / talk page 17:32, 20 October 2015 (UTC)

There is a proposal to delete 'Floruit' now that we have work period (start) (P2031) and work period (end) (P2032) but we still need a property to record the information we do have. I would be completely opposed to replacing this accurate sourced info with guesses for birth and death dates but there is nothing to stop you from adding guesses for birth and death dates as well as the 'work period' dates. Joe Filceolaire (talk) 20:10, 20 October 2015 (UTC)
@Filceolaire: The description says "Date when the person was known to be active or alive, when birth or death not documented" ???? You seems to imply that it's tied to the activity period when the description says it's more tied to the birth/death dates. This is unclear anyway. author  TomT0m / talk page 20:27, 20 October 2015 (UTC)
TomT0m: this is what sources give us. Sometimes they give us a work period, with a beginning and an end (work period (start) (P2031)/work period (end) (P2032)). Sometimes they just give a single point in time (derived from a single work) when we know the artist was active (and therefore also alive): floruit (P1317).
It's also maybe worth noticing that the precision of floruit is often (though not always) a specific year -- though it can also be used to indicate a particular century. Jheald (talk) 09:56, 21 October 2015 (UTC)
@Jheald: When we have a few informations, we can as well give only the century as the date of birth, is'nt that enough ? Then the "floruit" is redundant by definition ... Should not we have a guideline that "date of birth" is to give first ? It makes easier to query and more regular then. author  TomT0m / talk page 10:02, 21 October 2015 (UTC)
No. If we know that somebody painted a picture in 1885, it is valuable to record that precise fact, which is why biographical dictionaries do it (if no other information is available), rather than just generally state that the artist was alive in the 19th century. For one thing, it helps establish that the artist cannot be the same person as someone who died in 1860. Jheald (talk) 10:09, 21 October 2015 (UTC)
Also, if somebody created a work in 1820, we can't even say for sure which century they were born in -- 19th century or 18th century? floruit (P1317) precisely reflects the facts we have -- and matches what is stated in the source. Jheald (talk) 10:13, 21 October 2015 (UTC)
@Jheald: We got to work on the display of such information, because Wikidata data model and internal store can totally record informations such as a precise date with an arbitray interval of validity as far as I know. It's the display of information right now who is not good enough right now (I promised to work on this during an IRC office hour /o/ ) People thinks it's "19th" century but that's not correct, you can totally enter "1832" with a century precision. And the 1832 won't be lost.
This does not change the fact that we should have a guideline
I guess another question is : would not it be best to have a more generic property "known alive in" to separate the concerns ? "Floruit" could be a subproperty of it if it's more specific to working periods. author  TomT0m / talk page 10:46, 21 October 2015 (UTC)
In my understanding floruit (P1317) is precisely that generic property you are asking for, not giving a very rough estimate of their dates of birth and death but showing exactly the moment or period a person appears in the sources we have for that time and place. --HHill (talk) 11:40, 21 October 2015 (UTC)
Not really, see floruit (Q36424), if we know the person was a child in a year, it does not mean she was active artistically on that period. This is the ambiguity I want to point on, the description is not that precise. author  TomT0m / talk page 11:44, 21 October 2015 (UTC)
Then we will have to adjust said description. --HHill (talk) 13:37, 21 October 2015 (UTC)
floruit, taken literally, means "he/she flourished" (at a particular point in time). So it does rather imply that the person in question was at the height of their powers at the point in time given, rather than a mewling infant.
For the case where we want to establish that a child was alive, eg at the time they were mentioned in a document, or that their father died, the way to go may be to use latest date (P1326) as a qualifier on the date of birth, to note a definite terminus ante quem. Jheald (talk) 15:22, 21 October 2015 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────Agree with Jheald, American Heritage Dictionary 3rd edition defines floruit as "The period during which a person, school, or movement was most active or flourishing." The current Wikidata description is wrong to indicate it includes the period when the person was merely alive, and must be changed. Jc3s5h (talk) 16:32, 21 October 2015 (UTC)

"You do not have the permissions..."

I noticed that German Wikipedia's article de:Xunfu (although very short and without references), currently not interwiki-linked, seems to be about the same topic as en:Grand coordinator and provincial governor and the other articles interwiki-linked from Q56911. Well, when I try to add the interwiki the usual way from the German Wikipedia's article, with the "Add links" tool, at first everything proceeds normally: As expected, when I choose English Wikipedia's "Grand coordinator and provincial governor" article, there's a message that there's already a Wikidata item and whether I want to link this article to it. But in the next step, I get an "Error $1" and the message "You do not have the permissions needed to carry out this action". Well, I don't see why not, the Wikidata item doesn't seem to be protected... I suppose I could simply manually add de:Xunfu to Q56911, but before I attempt this, I thought I'd report the issue here; would be interested in the reason. Gestumblindi (talk) 19:39, 20 October 2015 (UTC)

de:Xunfu seems to be linked to Provincial governor (of the Ming and Qing dynasties) (Q1326009), so they'll have to be merged after that gets working again. Popcorndude (talk) 20:19, 20 October 2015 (UTC)
Problem probably is phab:T115892. We have a fix for it now and are in the process of deploying it. Cheers. Aude (talk) 21:53, 20 October 2015 (UTC)
and due to the bug, we temporarily disabled merge rights phab:T115994 for everyone, but these should be restored shortly. Aude (talk) 21:55, 20 October 2015 (UTC)
I've just restored the rights as the problem has been resolved. Cheers, Hoo man (talk) 23:32, 20 October 2015 (UTC)
Thank you all; successfully merged :-) Gestumblindi (talk) 21:08, 21 October 2015 (UTC)


How do I change and create new references when I edit a text? How do I "save" them until later?  – The preceding unsigned comment was added by Balcowiki (talk • contribs).

@Balcowiki: Hi! Are you talking about adding a reference to a claim? --Tobias1984 (talk) 19:02, 21 October 2015 (UTC)

Enabled sitelinks for meta, mediawiki and species

Hey everyone :)

We've just enabled sitelinks (Phase 1) for meta, mediawiki and species. Sitelinks to these sites can be added via the "Other sites" section.

Please note that it's not yet possible to access data from Wikidata on these wikis using the {{#property:…}} parser function or Lua.

Cheers, Hoo man (talk) 00:47, 21 October 2015 (UTC)

@Hoo man: For what it's worth, it works. —Justin (koavf)TCM 02:51, 21 October 2015 (UTC)
But why the Compact Links has been enabled on mw and species, but not on Meta? --Liuxinyu970226 (talk) 07:25, 22 October 2015 (UTC)

Request to delete - It's need to be deleted since I did it by mistake. Thanks --Cyfraw (talk) 09:59, 22 October 2015 (UTC)

I've redirected it to OnePlus (Q16499972). If you find (or accidentally add) a duplicate, see Help:Merge for how you can merge the two. For deletions, we have Wikidata:Requests for deletions. - Nikki (talk) 10:52, 22 October 2015 (UTC)

Lost data

How to best restore fertilization (Q127245) which could not be merged because of that bug in merging? Using rollback/undo/restore is not possible as some of the sitelinks were re-added elsewhere. Bad news is also that there may be many such items whose data is simply lost. If needed, I will create the list on-wiki where users could delete reviewed/fixed items. Matěj Suchánek (talk) 16:29, 22 October 2015 (UTC)

The easiest thing I can think of is to remove the sitelinks from the target item (either restore an older version of the item, or remove them manually if there have been other edits to the item), restore the item that needs merging and then merge the two together. Creating an on-wiki list would be good, Quarry doesn't have clickable links. :) - Nikki (talk) 17:40, 22 October 2015 (UTC)
See Wikidata:Database reports/Lost data. Matěj Suchánek (talk) 18:36, 22 October 2015 (UTC)

Getting link

How can I get fawiki link of these items? Yamaha5 (talk) 16:28, 19 October 2015 (UTC)

Yamaha5: Try --- Jura 10:16, 22 October 2015 (UTC)
@Jura1: How can i get links of some items? i tested and it doesn't workYamaha5 (talk) 18:39, 22 October 2015 (UTC)
Yamaha5: I had tried a similar sample and it had worked for me, but it seems I can't get it to work now either.
Alternate route:
  Done --- Jura 19:15, 22 October 2015 (UTC)

string for founding document reference?


I was looking into adding a bunch of "Staatscommissies" (State committees, Q14740334), which is an ad hoc Crown-appointed committee in the Netherlands. They have been around for 200 years now, some existing a few months, some existing over 100 years. An example is Staatscommissie-De Wilde. Each staatscommissie has been founded 'through' a "Koninklijk Besluit" (Royal Order, no Wikidata item), of which dozens are signed each day, and they typically have a single purpose: in this case, to appoint a committee, its members and scope. A Koninklijk Besluit is also relevant for appointment of officials in various government functions (mayors, high ranking soldiers, high ranking civil servants etc), so maybe there's a wider use.

My basic question is, how do I add a Koninklijk Besluit to the Staatscommissie item? Their numbering changed a bit over time, but until not too long ago, they were identified with a date and follow-number, i.e. "K.B. 24 januari 1936, N°.41". Later, they would be numbered with the year and a follow-number (i.e. 13.000434). How can I add this information to the item? Ideally, I would like to include it as some kind of 'identifier' of the committee, in a default way so that later, when archives are digitised more, they could be linked to. In other items, this Koninklijk Besluit could be used to substantiate an appointment as source, even when a url is not available.

Thank you for your help, I hope this is the right place (the Help page suggested that I ask it at RfC, but that seems highly inappropriate). Effeietsanders (talk) 18:25, 19 October 2015 (UTC)

I think the correct property is main regulatory text (P92), which then can link to the item with the corresponding Royal Order. And you can also use the same Royal Order item as a reference, e.g. the value for inception (P571) then uses as a reference stated in (P248). I am doing something similar for the Thai administrative units and the corresponding Thai royal decree (Q13017629), Ministerial regulation in Thailand (Q6406128) etc. Ahoerstemeier (talk) 13:40, 20 October 2015 (UTC)
I don't think I understand. How can I add 'Koninklijk Besluit 24 januari 1936, No.41' to this item, using P92? Because the only way to use P92 as far as I can see, is by creating an item for the Royal Order. When you consider the KB as an identifier, it would be silly to create items for each and every one of them. Effeietsanders (talk) 18:50, 20 October 2015 (UTC)
Then use inception (P571) to record the creation date and add the Koninklijk Besluit as a reference for this date. Are there urls for these? You will still have to look carefully at the reference properties you choose as many of these link to items rather than to strings. Think again; Is it that silly to have an item for the Besluit as well as an item for the Commission? Are there statements you can make about the Besluit that are not true about the Commission? Is Dutch Wikisource eventually going to add all the Besluits (which will mean they all have items)? Hope this helps. Joe Filceolaire (talk) 20:36, 20 October 2015 (UTC)
I indeed use inception (P571), and thought about adding the KB as reference. However, there are no URL's for those, they are often older than the internet, and not even digitized. Is there another way to add this as a source? It would still be very suboptimal, because of the value of the KB as some kind of unique identifier of the committee.
The only things you could add as statements to the KB's that I can think of, would be generic: that they are of the type Koninklijk Besluit, by who they have been signed, and where. Information not available to me, by the way (and quite irrelevant aside from the date). Please note that we're talking about an individual Royal Command/KB for each decision, which runs into the dozens per day of the past 200 years easily (the "no.41" refers to the 41st KB signed that particular day). It would be more likely that at some point in time, each 'Staatscourant' which includes these KB's would be digitized, and those would become items. But that is a whole different story... Effeietsanders (talk) 21:33, 20 October 2015 (UTC)
OK Effeietsanders. How about this.
That might work. <..> = main claim. (...) = qualifiers.
Help:Sources lists lots of properties that may be appropriate but it does strongly suggest a separate item for the reference text. For magazine and newspaper articles I link to the item for the newspaper and put the rest of the info as qualifiers in the reference. Is there an 'official journal' that publishes these 'Besluit'? Maybe just do the reference as if there is, like I suggested just above. Joe Filceolaire (talk) 01:02, 22 October 2015 (UTC)
With a few exceptions, the Royal Orders are published (in the Staatscourant), but there's at least the category of 'secret royal orders' (which doesn't work with a date, but a different type of numbering per year) and there's the more recent numbering system, which also combines year and a follow-number. I don't think all staatscourant's are easily online accessible to track the right publication, and I don't think there's a logic that could be applied to find it without reading through them. Your workaround above would at least work for the non-secret royal orders that are not recent. It would be human readable at least. The main downside I still see, is that the main identifier to differentiate between committees would be hidden in the qualifier of a statement, rather than be a statement itself. I start to wonder if 'Royal Order' would be a good property for general use. It could be used also as qualifier for appointments of all kind of people or other decisions. Or is that too specific? Effeietsanders (talk) 21:20, 22 October 2015 (UTC)

point in time (P585) and start time (P580)

There are a number of properties where point in time (P585) is used as a replacement for start time (P580)/end time (P582) when, for example, start and end are with year precision and both years are equal. Example: Duisburg (Q2100) for head of government (P6) has entry for 2012 with point in time (P585).

I think this usage is incorrect, as point in time (P585) should mean something that was true at one point (even if we don't know where in the year this point was) - e.g. it is good for population, GDP, etc. values that are measured at certain point, or an even that happened just once such as birth. However, for prolonged things - such as duration of the employment, or tenure as an official, etc. - it should still use combination of start time (P580)/end time (P582) even if both values would be the same year. This allows for easier querying and more logical structure, as each qualifier would have well-defined meaning.

I would like to fix all instances of point in time (P585) intermixed with start time (P580)/end time (P582) using a bot. But before that, I'd like to hear if anybody has objections to that or arguments of why the existing situation may be better. --Laboramus (talk) 18:59, 17 October 2015 (UTC)

Fully agree. For Duisburg (Q2100) the usage was completely wrong, as Sören Link (Q1453221) became Mayor in 2012, and still is the incumbent. So it wasn't a case of start and end having the same value, but simply using the wrong property. I have fixed this one case already. But when using the bot, be careful, replacing a point in time (P585) with a pair of start time (P580) and end time (P582) with same value could introduce wrong data. Ahoerstemeier (talk) 22:33, 17 October 2015 (UTC)
I also agree with the reasoning here, but I’m not convinced that this is the right kind of job for a bot to fix. @Laboramus: how do you plan on detecting errors? Have a list of properties, like head of government (P6), that should always have start time (P580)+end time (P582) instead of point in time (P585)? (Of course, that leaves the problem that, as Ahoerstemeier points out, in some cases only a start time (P580), not end time (P582), is correct… and there are probably cases where it’s the other way around too, because start time (P580) is unknown.) —Galaktos (talk) 22:37, 17 October 2015 (UTC)
Yes, I plan to have a list of properties which I will manually curate to see it makes sense. There are 135 properties now that use both qualifiers, I can go through them in a couple of hours and see if it makes sense for each to use point in time (P585) together with start time (P580), If it does not, I will add it to the list to be fixed. Then I'll run a couple of queries and see how many mistakes there are. If there are a lot, I'll let the bot handle them, if there's just small number I maybe just hand-fix them. But I suspect there is more than I can fix manually since I've seen such thing happening many times. --Laboramus (talk) 01:06, 18 October 2015 (UTC)
I've only discovered nine properties where significant mixing is going on and which should use start time (P580) IMO: head of government (P6) position held (P39) member of sports team (P54) educated at (P69) employer (P108) owned by (P127) manufacturer (P176) member of (P463) residence (P551) I'll see how best to sort these out. --Laboramus (talk) 03:43, 18 October 2015 (UTC)
  Support Matěj Suchánek (talk) 07:50, 18 October 2015 (UTC)
Swedish urban areas is an example which you can interpret in two different ways. Either that they were urban areas the exact date the Census took place, and we do not know the status of the place between the Census dates. Another way to interpret them is to say that when a Census tells an area was an urban area in one date, they stay that way until the next Census-report is published. On svwp we have mainly adopted the second way to interpret them. "Stockholm was an urban area 2010-12-31, and no Census has said anything else since then, so Stockholm is an urban area". But in the 1960-census, a special kind of urban area was described: urban area with a population of 150-199 (Q20738855). This kind of urban area was not confirmed in the 1965-census, and I have therefor here not used P585, but rather P580 to describe them. In 1990, a new kind of urban area was introduced, with a population down to 50. -- Innocent bystander (talk) 08:42, 18 October 2015 (UTC)
One of the problems with using both is that such a statement is very hard to query and automatically reason about - if I try to see who was a head of government in Elbonia in the summer of 1950, query that uses just start/end is much easier than one that has also to account for "point in time". Also, "point in time" is not exactly right there - someone is rarely a head of government for one instant, usually they are for a prolonged period. So stating it as a "point" is a bit misleading IMHO. point in time (P585) is also used several times when some office has yearly rotation - which is even more wrong as in this case we know they stayed in the office for the whole year, not just for one point within a year. Of course, there are legitimate cases of using both - e.g. somebody can be a member of a group for a long time or join the group just for one event - and in this case they should stay as is. That's why I am checking the properties to see it on case-by-case basis before changing. --Laboramus (talk) 22:33, 18 October 2015 (UTC)
If you have a source that says that the head of government of Elbonia on July 7, 1950 was .., you can't add a start date to your statement. You can only add point in time (P585). --- Jura 09:28, 19 October 2015 (UTC)
You can always, that why unknown value Help is made for. Or a plausible value with a large unprecision range. author  TomT0m / talk page 09:41, 19 October 2015 (UTC)
Would you have a sample for your suggestion? --- Jura 09:48, 19 October 2015 (UTC)
@Jura1: No sample especially, it's a generic Wikibase feature, when you know for sure there is a value but nobody knows which one, just use unkown value. If you want an approximative date, choose date with a century precision, would do the trick also. Which kind of sample would you want exactly ? author  TomT0m / talk page 11:22, 19 October 2015 (UTC)
A sample that illustrates the theoretical option you mention is appropriate and useful in the context of the property being discussed. --- Jura 11:52, 19 October 2015 (UTC)
I agree that an start time (P580) statement with unknown value Help can be added in this cases. But I don't think using a particular value with a large unprecision range is a good idea. It's semantically not correct. Also, it can be confusing for the users, since it implies there is some source connected to this particular date and range; which is not - both date and range would be editor's pure guess... If there was an unprecission definition like "before July 7, 1950", it could be used, but to used instead of it "July 6, 1650 ± 300 years" is not OK. Even if technically it can work.--Shlomo (talk) 07:24, 23 October 2015 (UTC)
@Shlomo: "since it implies there is some source connected to this particular date and range" Cannot parse. 1) Undo 2) Try again 3) Ignore 4) The fourth option. Seriously, this is an obvious contradiction, if there is no source there can't be a source linked to the statement. This as the value of a claim, by definition, not that of a statement. On the other hand it allows technically for the item to show up in more queries sorted approximatively, like a 3rd century person show up around the 3rd century in order, which is better than not showing at all or showing mixes with a 21st century person. This is a good compromise beetween absolute certainty by sourcability requirement and lack of informations which would imply no information at all or hardly queryable so useless information. I think. This is a close problem to the problem of computed or inferred statements by raw sources (if we know population and surface, we can compute approxiamte density, should we add density statements on Wikidata based on that even if we don't have the a source?) and how to deal with them in Wikidata actually. author  TomT0m / talk page 09:47, 23 October 2015 (UTC)
  • I can't react to the first part of your comment until "... not that of a statement", because it doesn't make sense for me, sorry.
  • The ordering issue can be solved using point in time (P585). Actually, it can be done more precisely that way, since if we know (e.g.), somebody was a mayor in 1910, the "large unprecision range" method would most probably include him into lists of 19th century politicians, while in fact we have no clue if he was one or not.
  • As for the density: if we have source for surface and population related to the same point of time, why not? A pure mathematical calculation is no research. The problem begins, when we have data about population and data about surface from different sources, with different credibility or/and different approach (e.g. including or discluding disputed territories), related to different point of time (this is usually not a big problem for density, since the surface is relatively stable, but can be a problem for other "combined" statements, like km of highways per capita). In that case, I'd recommend to have only the starting (=sourced) data (km of highways and population) here and let the users (scripts, modules...) decide if they want to make the calculations.--Shlomo (talk) 12:05, 23 October 2015 (UTC)
  • Another example where all three properties can make sense is award received (P166). Most of the awards are awarded for some time (specified, unspecified, or forever - use end time (P582) with no value Help then), but there can be also single-event aword (a ceremony in honor to the awarded person, public announcement, nonrecuring financial donation etc.), where point in time (P585) could be more appropriate.--Shlomo (talk) 07:24, 23 October 2015 (UTC)

Units usage

I've made a quick review of current unit usage:

As a result, I've discovered some unit usage that I think is questionable - e.g. using human (Q5) as unit for number of employees (see Biocoop (Q2904039)), or usage of episode (Q1983062) as a unit for number of episodes (P1113) on Bob's Burgers (Q392396). Do we really need to add such units there? I think in those cases when we're just counting items and not measuring, the units should not be added, as they do not add more information (we would already know that series length is measured in episodes, since all series are measured this way) and are not used consistently.

Thoughts? --Laboramus (talk) 22:51, 19 October 2015 (UTC)

  • In the case of the employees using human (Q5) as unit doesn't make any sense, but it maybe can make more sense if we are specifying the number of male and female employees. In the case of the series we can count it's length in more ways than in episodes, the episodes one is the most used one, but what about seasons? If only the most common usage of the measurement is used I agree that it shouldn't be put specifically, at least if it can be considered general knowledge. The thing is that it should be analyzed case by case if it is really necessary for that specific usage and limiting it by software is not something that we should not do in any of the examples. -- Agabi10 (talk) 23:07, 19 October 2015 (UTC)
    I would say length in seasons and in episodes should be different properties, because it measures different things. Otherwise it's like having area and populaion of the country in the same property - while those are somewhat related, they are different things. In the same way, number of seasons and number of episodes are different things and querying them both in the same property would only lead to a massive confusion. Of course, technical it is possible to put all quantity properties into one property and distinguish them by units, but it's not a good idea. I would say we should use units either when it's clear they are necessary - such as "the length of this road is 23... of what? meters? miles? kilometers?" or when we want to distinguish separate groups, but not when the property should always (or nearly always - maybe some fantasy company might employ non-humans) be counting the same thing. --Laboramus (talk) 03:23, 20 October 2015 (UTC)
    I proposed the
    ⟨ series ⟩ has part (P527)   ⟨ season ⟩
    quantity (P1114)   ⟨  5 ⟩
    scheme or similar with sequence length. author  TomT0m / talk page 11:13, 20 October 2015 (UTC)
  • @Laboramus: I disagree with the number of episodes and the number of seasons both don't fit in number of episodes (P1113), I don't disagree that it would be better to create "number of episodes" and "number of seasons" as properties, but I don't think that using the unities to specify which one you are measuring in this case would be incorrect (at least while we keep using this property). Specifying it we can check the unity and know if it is talking about the episodes or the seasons. If we don't specify we guess that the most common usage is the episode number, but we can't be sure that it is going to be the episode number what we are getting. And I don't think it is fair to compare it to putting all the numeric values into the same property. -- Agabi10 (talk) 03:38, 20 October 2015 (UTC)
Population/census data is never just a total number of humans. There is always a breakdown by race/sex/age etc. en:1800 United States Census for instance breaks down the count into 12 different groups such as 'free white females aged 26-45' and 'other free persons'. I think it makes sense for the item for 'Tennessee' to have claims
    • <population:19227 - 'humans'>
    • <population:6992 - 'free white females aged 26 - 45'>
    • <population:309 - 'other free persons'>
This saves us from using awkward qualifiers to convey the same information. I think there is a case for using this type of unit in most numbers to make it clear what it is a number of. Joe Filceolaire (talk) 20:24, 20 October 2015 (UTC)
That sounds excellent until you then try to write a query that finds 10 most populous cities, and you discover that now population can be in "people", "humans", "homo sapiens", "males" + "females", "humanoids" and another 1000 ways depending on creativity of whoever composed the entry. That makes efficient processing of the data very frustrating. --Laboramus (talk) 19:00, 22 October 2015 (UTC)
And when we add such things as 4,231 New Yorke, we suddenly have a unit that is impossible to translate to some languages. -- Innocent bystander (talk) 10:30, 23 October 2015 (UTC)

WDQ question

I want to create a query in WDQ (so I can use it in autolist to check where I can remove P513 (P513) (which is a depricated property).) that lists all items that contain P1477. I tried CLAIM[1477], that returns exactly 8 items that all have set birth name (P1477) to "unknown value". P1477 has values in monolingual text. I know there are items with values for P1477 as I've been adding them. Mbch331 (talk) 13:04, 23 October 2015 (UTC)

Probably not, but SPARQL can help: [10] --- Jura 13:16, 23 October 2015 (UTC)

Modelling methodology for classes

I am trying to find out the basic modelling methodology for classes used in Wikidata. I am trying to find this out because the company I work for is interested in using Wikidata as a source of background knowledge. To do that effectively, we need to know how groups of things are modelled, for example so what we can determine which diseases are in Wikidata or which colors are in Wikidata. This information is also useful when designing a new domain and is even useful when adding new items.

As far as I can tell, there is no easily accessible information on the basic class modelling methodology used in Wikidata. There are several related pages, including

and some pages that have been last modified in 2013. None of these appear to policies or guidelines and none of them are easy to find.

The kind of thing that I was looking for would talk about how to model regular classes (like Human) and higher-order classes (like disease and color), and discuss what should and should not be done. In particular, it would warn against the multi-level nature of disease (which has malaria as both an instance and a subclass) and color (which has some colors as instances and others as subclasses). It appears that there are quite a few domains that hav this sort of problem. The page that is closest to providing the answers is Help:Modelling but that page is much more general and doesn't have enough information on modelling of classes.

Is this information actually easily available and I haven't noticed it? Should there be a policy or guideline about this? How can I best contribute to the development of such a policy or guideline? Peter F. Patel-Schneider (talk) 18:19, 21 October 2015 (UTC) [The timestamp is a bit off.]

  •   Comment Taxons are fairly well structured. See WikiProject Taxonomy. --- Jura 16:23, 21 October 2015 (UTC)
    I had never an answer from taxonomists about taxons beeing or not clade (Q713623). So I would not say this is an example of perfection as this is an important property of a taxon. author  TomT0m / talk page 16:32, 21 October 2015 (UTC)
    Did you have a specific problem? --- Jura 16:42, 21 October 2015 (UTC)
    With classification of organisms ? If I had only one ! Well, all humans items in the project are instances of a "common name" class, wtf, and non are instances of the "homo sapiens" taxons. Non taxonomic groups of animals are as well "common names", whatever that means. Succu hates me and reverts me whenever I try to translate the WikiProject page. Impossible to talk with them to make them use the community decided scheme Peter is trying to talk here whenever it does not change a lot for them and provide a consistent classification scheme totally compatible with Help:Classification. author  TomT0m / talk page 16:55, 21 October 2015 (UTC)
  •   Comment @Peter F. Patel-Schneider: I did some work on reorganizing the Basic membership properties page here: User:ArthurPSmith/Test/Help:Basic_membership_properties - which you seem to have noticed - but didn't feel brave enough to take it live yet. Also it was waiting on a rewrite of the Glossary page that was in progress here: User:Filceolaire/Draft:Glossary. There are definitely a lot of inconsistencies and overlap in the use of 'instance of' vs 'subclass' at the moment, and it's not being enforced perhaps because the guidelines aren't clear enough (though I think they are actually pretty clear). ArthurPSmith (talk) 17:52, 21 October 2015 (UTC)
    @ArthurPSmith: I did indeed look at the first page you mentioned, but I had not found the second. The information in User:Filceolaire/Draft:Glossary about instance and class is relevant, so thanks for the pointer. The page User:ArthurPSmith/Test/Help:Basic_membership_properties has information about instances and classes, which is useful, but doesn't talk about how to do the actual modelling. However, where are the guidelines on the use of classes? As far as I can tell even Help:Basic membership properties is not (yet) a guideline to be followed and anyway it doesn't talk about how to do the actual modelling of classes. I don't think that I have found any pages on classes that are guidelines that are supposed to be followed, but maybe I don't understand the status of some of the pages that I have noticed. Peter F. Patel-Schneider (talk) 18:19, 21 October 2015 (UTC)
    Do not look to deep into the status of pages. This project is young and setting to firm guidelines can damage the development of the project. We currently have an RFC about if Color should use instance or class. An interesting question is raised: "What is a Color?". When we find the answer to that question, it's probably easy to see which should be used. In some subjects that question is simple, in others it isn't. -- Innocent bystander (talk) 18:51, 21 October 2015 (UTC)
    I would say instead that the problems with color are the result of not having firm enough guidelines. Why not try for some guidance so that these problems don't arise over and over? The guideline would say something like it is important to figure out what belongs to a class, and what does not belong to the class, and write this down somewhere so that the use of the class has some consistency. Even if this advice is not always sufficient, creating and populating classes without worrying about this fundamental aspect of classes generally ends up with information that is hard to modify and hard to use. Peter F. Patel-Schneider (talk) 20:24, 21 October 2015 (UTC)
    OK, perhaps the best way forward is to work on part of the Help:Modelling page so that it contains good information about modelling classes. Peter F. Patel-Schneider (talk) 20:24, 21 October 2015 (UTC)
    Peter F. Patel-Schneider Before going forward and be lost in the examples we need to set up the general parameter of the classification (in our case this can be defined as an ontology):
    • what's about inheritance between classes ? if A is a subclass of B, can all properties and values of B be applied to item A ?
    do we need to copy all properties and values of item B to item A or can we assume that inheritance as automatic and if we want to have all information of item A we need to collect all properties/values from upper classes ?
    • what's about granularity ? Have the lowest level items in the classification to be a instance of ?
    • can an item be an instance of and a subclass of at the same time ?
    Before starting a new description of examples we need to set up a wikiproject in order to list the general characteristics of a classification, then to choose the ones we want for WD classification after checking different cases and only starting that point we can create a general page summarizing all cases (if this is possible) like Help:Modelling. Snipre (talk) 13:00, 22 October 2015 (UTC)
    Snipre hum, you're looking for existing WikiProject like WikiProject Ontology and WikiProject Reasoning, let's not split efforts more than this. Help:Modelling already exists, I spent some time to unit usecases and to try to find solution for these usecase, let's not waste this. There is also RfCs like Are_colors_instance-of_or_subclass-of_color where several persons have commented and where we can see the overall opinion of community at that point. author  TomT0m / talk page 13:06, 22 October 2015 (UTC)
Thanks for the link of WikiProject Ontology: I didn't know the existence of this project. So I correct my previous comment. Sorry but for me Help:Modelling is useless because it treats cases and don't provide the principles. This page doesn't answer the questions I listed; this pages is full of examples which are a problem because each contributor has to first generalize the classification scheme from the examples to be able then to adapt it to its particular use. Typically in a general presentation we should have only items A, B, C and all explanations based on these anonymous items.
Then specific RfCs are the typical worst management of classification: we are defining rules for each field and later we will have conflicts when trying to implement general tools or processes on the data. Your conclusion "...where we can see the overall opinion of community at that point" is totally wrong because most of the contributors will skip this kind of discussion due lack of knowledge in colour problems. Again this leads to the need of generalizing all the cases to focus on the principles and not one a case. Snipre (talk) 14:15, 22 October 2015 (UTC)
@Snipre: I tried to come up with some agreement with a few RfC myself on Classification, this did not work and there never was a clear outcome. What's interesting in Tobias' current one is that it's a case we failed to come to an agreement on which we can test generic principles like the one exposed in Help:Classification. Then he leaves an open question on "is this generalizable" ? To me, as I argument in the RfC, yes, we can classify colors with the Help:Classification principle, this have been an unsolved case yet, and yes, this is a proof that this just works. No matter how we must stop hypercritical thinking, stop overthinking how things should happen and finally close this question opened since the very beginning of the project, we have enough material and experience now. author  TomT0m / talk page 16:25, 22 October 2015 (UTC)
  • @Snipre: To take your questions in reverse order, the answers to the last two seem to be straightforward: (c) Yes and (b) No; (a) I find more interesting -- can you give any examples of any properties that might work this way, in any circumstances? If so, the best way would seem to be to add statements to the properties -- eg 'also applies to all subclasses' (and instances?) or 'also applies to all superclasses'. Jheald (talk) 14:54, 22 October 2015 (UTC)~
    @Jheald: Nothing is straightforward: you said because you have your opinion about how the classification should be. But can you explain what are the consequences of that choice on the classification ? Snipre (talk) 13:04, 23 October 2015 (UTC)
It's certainly true that all properties that apply to a class because they apply to all instances of the class must therefore apply to all subclasses (since the instances of the subclasses are implicitly instances of the superclass, for which the property is defined to hold); similarly for properties like properties for this type (P1963) which characterise what can be said about instances.
But I'm not sure (i) whether the same inference can be made for other sorts of properties, which may be defined only for classes, not for properties; and (ii) which properties we allow to be applied to classes, if they apply to all instances of the class.
This may be something that needs to be noted property-by-property (and perhaps already is, in the constraint conditions). Jheald (talk) 15:25, 22 October 2015 (UTC)
instance of (P31) being an example of a property which (I think) applies equally to subclasses of a class, but not to instances of it. Jheald (talk) 15:38, 22 October 2015 (UTC)
To help consideration, here is the report of the properties most often applied to classes; and, since it seems to be somewhat dominated by genes, here is one excluding classes with found in taxon (P703). Something that jumps out is that there may be subtlety in thinking how identifiers apply to subclasses (since they may often have their own more specific values). Other properties too may take more specific values for subclasses -- which in some sense defines a nesting for those values. We should probably try to make sure that we're capturing that nesting in some appropriate way, if the property is item-valued. Jheald (talk) 16:00, 22 October 2015 (UTC)
I don't think that instance of (P31) applies to subclasses of a class. Consider cars. The class of Honda cars can be considered to belong to the metaclass of car makes. However, the class of Honda Accords does not. Instead it belongs to the metaclass of car models. The main "properties" of classes that are "properties" of their subclasses are things that are true of all instances of the class. For example, one might say that the instance-manufacturer (yes, this is a rather crazy property) of Honda cars is Honda (the company), this would also be true of Honda Accords. There are lots of subtleties to be considered here, and already quite a few places in Wikidata that are going to cause problems down the line (e.g., using part of (P361) on classes to signal that instances have a part of a particular kind). Peter F. Patel-Schneider (talk) 17:32, 22 October 2015 (UTC)
@Peter F. Patel-Schneider: Interesting. Here's a query for the most commonly instanced metaclasses, together with a sample class, and (if there is one) a sample subclass. It might be interesting to create a "metasubclass of" property (or "has metasubclass") to capture what metaclasses are possible for subclasses of instances of the original metaclass. Jheald (talk) 18:41, 22 October 2015 (UTC)
And a tweaked version, also showing the metasubclass
(This query does seem to tend to show up rather a number of apparent classification inconsistencies, typically; although there is also the potential for some deceptive crosstalk, because the subclass or the superclass had more than one P31 metaclass, something I haven't checked for.)
It seems possible to me that we might already have the right property to capture the "metasubclass" relation, in the shape of part of (P361). Though as you indicate, this may be spoilt if part of (P361) is instead being (mis?)used between classes to indicate that instances should have this relationship. Jheald (talk) 19:04, 22 October 2015 (UTC)
An attempt to tweak the query further, to prefer if possible metasubclasses that are different from the metaclass; but it times out. Maybe someone can see a way to make it more efficient? Jheald (talk) 19:38, 22 October 2015 (UTC)
If at first you don't succeed... A list of all metaclasses instanced on subclasses of instances of the 250 most instanced metaclasses.
It's quite a rough list, but all the same not without interest I think. Some of the 'metasubclasses' are obviously crosstalk; some are genuine; and some appear to indicate some clean-up needed, or at least the presence of items needing some classification refinement. Jheald (talk) 20:07, 22 October 2015 (UTC)
@Snipre: What I am proposing to work on is indeed the general methodology for setting up classes and their interrelationships, e.g., what stances to take on things like metaclasses and subclassing and instances. This is actually a precursor to setting up the ontology, which is involved with creating the actual classes and their actual relationships. One does have to consider real needs when doing this, and provide examples, but the idea is to produce some general guidelines or policies. If Help:Modelling is not the right place for this, then I'm certainly willing to use some other vehicle. I don't think that WikiProject Ontology is the right place, for the above reason, but I suppose that that project could be expanded to include knowledge-structuring discussion. Peter F. Patel-Schneider (talk) 14:59, 22 October 2015 (UTC)
@Peter F. Patel-Schneider: As you pointed we have already plenty of examples, the problem is that nobody tries to find the general rules from that bunch of cases. But even looking for a general scheme from current cases is a lost of time because we should first ask the good question (what kind of classification we want, what characteristics we want,...) and then find the principles by applying the desires with the real cases.
As TomT0m said people are not really interested in classification problem because nobody present them the possible classification and the advantages/disadvantages of each one. We need a real effort to propose information about classification instead of working in small group and presenting long pages that nobody understand. That's why a wikiproject is the correct place because to can start with a group of contributors and avoid the current process where one or two contributors write a long document and just ask at the end the agreement of the rest of the community.
Then for you can't start the discussion of the classes if you don't know what kind of ontology you want. You need to fix the framework before starting to work on elements because elements definition has an impact on ontology definition. Normally by defining the ontology you limit the possibilities in the definition of the elements like classes. Snipre (talk) 11:48, 23 October 2015 (UTC)
@Snipre: If a Wikiproject is the right way to go, then I'm fine with that. If WikiProject Ontology is suitable as the project to use, I'm fine with that. I do think, however, that what is needed (at the end certainly and even probably at the beginning) is some document setting out the desires and a way to achieve these desires, with examples of what is currently working right and what is currently not working right. It will need community involvement and community buy-in. I'm willing to put work into such a document.
One thing that is probably going to have to be worked out is the meaning of various terms. To me, an ontology is about the set of classes (object, person, disease, etc.) and their properties, and is not a framework stating principles of how classes are interrelated (no metaclasses allowed, for example). That is a modelling methodology, or perhaps an ontological framework. Peter F. Patel-Schneider (talk) 20:50, 23 October 2015 (UTC)
  •   Comment It seems to me that the big $64,000 property of subclass of (P279) is that it is transitive: if A is a subclass of B, then all instances of A and its subclasses should also be instances of B. The corollary of this is that if there are instances of A or its subclasses that are not instances of B, then P279 is not appropriate to describe the relationship between A and B -- though, depending on the circumstances, instance of (P31) might be.
I think that's pretty much the limit of what we currently aspire to.
The above test really applies most sharply to items that already have sub-items, whether by P31 or P279.
For items that don't yet have sub-items, I think the honest answer is to expect Wikidata to be somewhat fluid. If the item is clearly a concrete individual thing, like a particular person or a particular place, then it will only have P31 properties and won't be a subclass of anything. But for more abstract items, it seems to me - at least for the time being - that the attitude has mostly been one of "Meh. Does it matter?".
If such an abstract item doesn't have any sub-items at the moment, then it might very well currently have a number of P31 relationships to classes it relates to. If sub-items later get defined, then those P31s may get revised to P279s. But, *so long as no such sub-items exist* (eg as was the case for the colours mentioned?), I think the general feeling has been to see the question of P31 or P279 as a pretty low-stakes issue. Small misdemeanor, not many dead.
Yes, in other ontologies such sub-items may very well exist; and if there is a potential for creating them, that probably indicates that the relationship ought really to be a P279 rather than a P31. But for the moment, for leaf nodes, the realistic position is not to expect Wikidata to have been too careful in this area (and therefore to allow for that, if record-matching). Jheald (talk) 14:10, 22 October 2015 (UTC)

Classification/taxonomy presentation tool

Relevant to the above discussion, the following is the announcement of a new pretty-pretty classification taxonomy viewer from the Wikidata list earlier today:

Particularly relevant is that it comes pre-loaded with apparently dubious graphs to investigate -- in particular, relevant to the above, cases (under the tab "Relations Errors") where classes have been made an instance of (P31) a class that they are implicitly also a subclass of through a chain of subclass of (P279) relationships, for example:

This is very often a sign that the instance of (P31) is wrong, and ought to go to something quite different; or is simply a mistake.

For more, see the thread linked above. Jheald (talk) 20:33, 22 October 2015 (UTC)

Rewrite for the Glossary

I have done a rewrite for Wikidata:Glossary and it has been sitting at User:Filceolaire/Draft:Glossary for some time while I have not ben able to work on it. I think it is time to replace the existing version with mine.

  • Does anyone object?
  • Can anyone help?
  • What should I do next?
Joe Filceolaire (talk) 00:04, 22 October 2015 (UTC)
What about 5 complete, 4 almost complete and more incomplete translations? It is worth losing them all? --Jklamo (talk) 12:05, 22 October 2015 (UTC)
Jklamo: I'm afraid it is. That is how bad the current Glossary is. Joe Filceolaire (talk) 15:52, 23 October 2015 (UTC)
Those incomplete translations are not lost. The whole point of adding the translation tags is to keep translations of terms that are still in the glossary, so translators only have to update rather than starting from scratch.--Snaevar (talk) 15:22, 24 October 2015 (UTC)


So we have 2015 Tour de France (Q15220420) with instance of (P31)=Tour de France (Q33881), which I suppose is OK. But how to connect for example 2015–16 NBA season (Q19862716) with National Basketball Association (Q155223) or 87th Academy Awards (Q16773560) with Academy Awards (Q19020)? --Edgars2007 (talk) 13:46, 23 October 2015 (UTC)

@Edgars2007: I'd use organizer (P664)  . author  TomT0m / talk page 14:07, 23 October 2015 (UTC)
Oh, and conferred by (P1027)   for the second example. author  TomT0m / talk page 14:08, 23 October 2015 (UTC)
Actually I think that conferred by (P1027) for Academy Awards is Academy of Motion Picture Arts and Sciences (Q212329). And that was general question, 2015–16 NBA season (Q19862716) and 87th Academy Awards (Q16773560) was just an examples. organizer (P664) wouldn't be the best choise for such items. I think there are cases when organizer isn't the league, but some other organization. --Edgars2007 (talk) 19:16, 23 October 2015 (UTC)
@Edgars2007: And 4 (Q126163) with (most probably) New York City Subway (Q7733). --Edgars2007 (talk) 19:23, 23 October 2015 (UTC)
@Edgars2007: For that one, I's say as both are concrete objects. A general question is not really a set of examples ;) There is (imperfects) introductions to some generic properties in Help:BMP or Help:Classification which expose generic principles. Help:Modelling is an in construction help page to expose and be a point to collect more generically some principles about how to model time, classification, sequences on Wikidata. It needs a lot of work however. For more specific fields like awards there should be in principle like WikiProject Awards do not exist (yet?) Will you be its creator? click here. will provide guidelines. author  TomT0m / talk page 09:41, 24 October 2015 (UTC)

Trouble with WiDaR

I'm having rouble logging into WiDaR, to use the Wikidata Game, using my netbook, a low-spec but new Win8 machine, running Firefox, which can struggle with script-heavy pages. It loads the OAuth special page, but I don't get the pop-up allowing me to "accept". I have previously been able to log in, with this machine. This happens even after a fresh restart, with no other tabs open, and AdBlockPlus disabled. Any work-arounds, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:19, 23 October 2015 (UTC)

I notice the same thing. What works for me is when I have the OAuth page open to log out and login again. Then I do get the pop-up. Mbch331 (talk) 07:38, 24 October 2015 (UTC)
Thank you. I just tried that, twice, but it didn't work for me. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:20, 24 October 2015 (UTC)

P1477 birth name

Please change the name of the aliases "name of birth" and "full name of birth" to "name at birth" and "full name at birth". --Richard Arthur Norton (1958- ) (talk) 01:58, 23 October 2015 (UTC)

Done. Ajraddatz (talk) 02:10, 25 October 2015 (UTC)

Spanish name

For "Julián Oñate y Juárez" and other Spanish names didn't we have a field to contain the "Juárez" portion? Give_name=Julián and familyname=Oñate. I can't find the name of the field anymore, and most records do not use the field. Anyone remember? --Richard Arthur Norton (1958- ) (talk) 16:43, 23 October 2015 (UTC)

What you're looking for is: second family name in Spanish name (P1950). Jared Preston (talk) 17:13, 23 October 2015 (UTC)
Exactly! Thanks. --Richard Arthur Norton (1958- ) (talk) 01:40, 24 October 2015 (UTC)
I still say 'Juarez' is a 'family name'. Joe Filceolaire (talk) 03:32, 25 October 2015 (UTC)

Number of unique articles

Hi, two questions.

  1. Does anyone know how many unique articles are there across Wikipedias? The total number of articles is 36,313,926, but how many of them are duplicates? Say, there are 5 duplicate articles on Picos de Urbión Q3752138. There are 14,960,757 data items on Wikidata, but a lot of them are admin stuff -- categories, templates, project pages, dab pages, etc. Does anyone know how many unique articles/topics are across Wikipedias? Not looking for precision, just a rough idea.
  2. Is there an easy way to run a report of Wikidata items that say have 5 interwiki links but not language X link? Thinking this would be helpful in identifying missing topics on English Wikipedia.

Thanks! Renata3 (talk) 16:17, 25 October 2015 (UTC)

re: #1: There is a summary chart at Wikidata:Statistics. For 25% of the items, we don't know what it is. Specifically for bios, there is People_charts#Resources.
re: #2: There are
--- Jura 16:38, 25 October 2015 (UTC)

Wikidata weekly summary #181

Flow problems

Really, I don't get a notification by email if someone writes on my talk page? At least twice I've missed a message on my talk page because of that. My preferences related to notifications are: Talk page message (web, email), Flow (web). If I put "Flow (web, email)" I will be spammed by Flow messages left elsewhere (on those pages that are on my watchlist). Is the only choice really that I have to stop following those other pages? Because AFAIK there's no possibility to convert your talk page back to normal wikitext. --Stryn (talk) 18:50, 25 October 2015 (UTC)

I think enabling Flow on this wiki was a bad idea; eventually the novelty will die off, and then we're stuck with user talk pages written in a goofy schema that we can't convert back to normal, just like LiquidThreads is elsewhere. --Rschen7754 21:49, 25 October 2015 (UTC)
Flow notifications work fine for me. Joe Filceolaire (talk) 00:27, 26 October 2015 (UTC)


Are there any constraints like apoapsis (P2243)>semi-major axis of an orbit (P2233)>periapsis (P2244)? -- Innocent bystander (talk) 07:32, 26 October 2015 (UTC)

Can you provide a better explanation about what you want ? Thank Snipre (talk) 08:15, 26 October 2015 (UTC)
The value of P2243 is always larger (or equal in rare cases) than that of P2233 and P2233 is always larger than P2244. At least P2243 and P2244 are easy to mix when you add them manually. -- Innocent bystander (talk) 08:20, 26 October 2015 (UTC)
From what I know there is no corresponding constraints. And creating properties for this kind of relations is difficult: you assume that values are from the same source. Because all values can be true even if they don't respect your constraint due to the fact that different methods were used and they provide correct values in their methodology but wrong when compared to other methodologies.
From source A, apoapsis (P2243)=3 > semi-major axis of an orbit (P2233)=2 > periapsis (P2244)=1. Constraints are ok
From source B, apoapsis (P2243)=2 > semi-major axis of an orbit (P2233)=0.8 > periapsis (P2244)=0.4. Constraints are ok
But constraints not ok when comparing values from sources A and B.
Your constraint should include some source matching before comparison. Snipre (talk) 10:22, 26 October 2015 (UTC)
They normally should be with the same source, yes absolutly. They also should have exactly the same epoch. (A property still missing since we do not have a fitting datatype yet.) I would prefer to see properties about orbital elements (Q272626) grouped together, so they can share source and epoch. I do not see any model for that yet. -- Innocent bystander (talk) 10:53, 26 October 2015 (UTC)
There is a structure to group properties. Look at significant event (P793). Snipre (talk) 13:01, 26 October 2015 (UTC)
Another issue is that these are quantity properties with units - if the units entered for the different properties are different how do we manage the comparison? I suppose one could constrain related properties from the same source to always use the same units (that would be expected)?. ArthurPSmith (talk) 14:25, 26 October 2015 (UTC)
Propose properties to describe constraints on the Wikidata:Property proposal/Property metadata page. I don't think the constraint bots are using these properties to define constraints but the plan is for them to do so in future. Specifying that the constraints should be applied to orbital details form the same catalog could be dificult, even with qualifiers. As celestial objects rarely change their orbits I think the simplest is to say the constraints should be applied to the preferred value which should be the current value (where the orbit has changed at some point) or the most accurate value.
There is a Wikidata:Property_proposal/Generic#conversion_to_SI_base_unit and Wikidata:Property_proposal/Generic#conversion_to_standard_unit under consideration at the moment to define all measurement units in the corresponding SI units. This should allow a bot to compare measurements in different units by converting all the units to the same SI unit. Please comment and vote.
Hope this helps. Joe Filceolaire (talk) 16:14, 26 October 2015 (UTC)

looking for best ID properties (protected areas)

Which properties can be used for IDs in the following case? (i think there is no such property yet, so i'm looking for your ideas how to handle this best.)

  • subjects/relevant items: several thousands of nature/conservation protected areas in Germany (ca. 8500 nature reserves plus an even higher number for other types: protected landscape areas, natural monuments…)
  • type of IDs: numbers or number/letter strings assigned by state institutions on a per-state basis (Germany consists of 16 federal states), or on a per-district basis
    • exemplary ID schemes: No. 01234, 7654-032, 4.321, NSG-00102.03, AB-0123
    • uses on Wikipedia: typically in infoxes like here ("Kennung: EN-010"), in lists like here (first column), and directly in article texts like here (second sentence)
    • pros and cons: these IDs are important for area identification, they are regularly used and referenced in German publications (including scientific papers, official lists and databases); they are always unique in relation to the state/district and to the specific type of protected area (like No. 1625 for type "nature reserve" in "state X"), but one number may be used twice or more if you compare all types of protected areas in all federal states or districts
  • note 1: we have WDPA ID (P809) on an iternational basis, but it does not include all types of protected areas; WDPA also has some problems not relevant here (eg. IDs not really stable/may change once in a while, database incomplete/relies on delayed updates by country institutions)
  • note 2: on principle, these considerations may also be of interest for many other countries and/or for other geographical features, but i don't know their official ID schemes (for France see already existing protected areas INPN Code (P1848)) and I'm unsure about potential pitfalls

At the moment i see four options:

  • a) specific properties for every possible combination (type of protected area per reference area = state/district): focusses on really unique IDs, but means many new properties (>100 or rather >1000 combinations in Germany)
  • b) one property for all: sort of generic "authority-assigned protected area ID" (either in Germany only, or worldwide) which is suitable for various frames of reference, but means not using unique IDs
  • c) a mix between a) and b): one ID property for each type of protected area in Germany (ca. 10 types) like "authority-assigned nature reserve ID in Germany": a bit more specific, but still not using unique IDs in all cases (due to similar ID schemes in some federal states/districts)
  • d) an even more generic "authority-assigned ID", which could be used for a very wide variety of subjects worldwide (street registries, cultural heritage etc.) depending on what is defined as "authority"

I think a) and d) are not good options: a) will need too many new properties, and d) is too random/discretionary to be useful. But i'm not sure about b) and c).
Because i did not request any new property yet, and before i provoke a mess on property proposal subpages, i would really appreciate your help here (objections and suggestions, links to related former discussions etc).
Holger1959 (talk) 09:41, 26 October 2015 (UTC)

b) or c) is the way to go. We already have similar properties: catalog code (P528) with mandatory qualifier catalog (P972), inventory number (P217) with mandatory qualifier collection (P195) and website account on (P553) with mandatory qualifier website username (P554) --Pasleim (talk) 12:17, 26 October 2015 (UTC)
This was discussed a while back in relation to other things. We even had an RFC two years ago - (Wikidata:Requests for comment/How to classify items: lots of specific type properties or a few generic ones?). The conclusion was that a specific property for a catalog code had advantages but that it probably was not worth doing for all the smaller catalogs, especially if the data isn't online. The RFC also concluded that where specific properties are created these should be marked as 'subproperties' of more generic properties and we should pressure the devs to create a "search including subproperties" function so we can search a bunch of properties at the same time and don't need to know exactly which property was used in each case.
Since then we have created a lot of specific properties. We have also created subproperty of (P1647) and formatter URL (P1630) to describe these properties.
The first two proposals on the Wikidata:Properties for deletion page today are proposals to delete specific properties (for languages and for railway station codes respectively) in favour of generic properties. These discussions started in january/february and don't seem to have arrived at any consensus yet.
Hope this helps. Joe Filceolaire (talk) 15:57, 26 October 2015 (UTC)

Describing software features

Can WikiData be used to describe something like that ? I know there is almost no required properties, but is it possible to create them easily to describe such details? --Ilya (talk) 23:25, 25 October 2015 (UTC)

  • There are lots of feature comparison tables on english wikipedia, many of them software related. In principal there is no reason not to include the data in these tables in wikidata. This would involve creating a property corresponding to each column in these tables. Then you depend on some Lua whiz recreating the table using wikidata info. Once this is done the same table can be added to other language wikipedias.
The first step is creating the properties for each of the table columns. Many of the properties needed for this already exist. A bunch of camera related properties were created recently for instance. Your best way to proceed is probably to look at a bunch of these tables then go to Wikidata:Property proposal and propose a bunch of properties related to a particular domain. Have regard to other domains as well and where possible propose properties that can be used for other stuff too. Hope that helps. Joe Filceolaire (talk) 00:26, 26 October 2015 (UTC)
  • Please don't start to proposal a bunch of new properties but try to create first a list of properties within a wikiproject and once a first discussion was made, then you can start the properties proposals. There is already a lot of properties available and before starting to create duplicates we need to have a clear overview of what we have and what we want. This can help a lot the approval process. Snipre (talk) 08:14, 26 October 2015 (UTC)
    yes I'm going to create an overview with several islands of technologies — Java SE and EE, Spring, JBoss, Eclipse, Jetbrains, Netbeans, Typesafe, ELK (ElasticSearch/Logstash/Kibana) and MediaWiki. This would need some automation tools, so I would create a Scala wrapper around Wikidata-Toolkit somewhere around spring 2016 for this --Ilya (talk) 20:42, 26 October 2015 (UTC)

Improving WikiProjects on Wikidata

Hello everyone,

Earlier this year, I embarked on a project, WikiProject X, to improve WikiProjects on the English Wikipedia. What we found was that WikiProjects can be a helpful way to organize editors interested in the same subjects, but have issues with sustainability because of how much effort a WikiProject typically takes to maintain. We have approached this by developing a new WikiProject design that presents information more clearly, and by developing automated tools that provide work lists for WikiProject participants. You can learn more at w:Wikipedia:WikiProject X.

For the next phase of this project, I would like to bring Wikidata into the mix. While there are many Wikidata editing tools at our disposal, there are not many avenues for collaboration. I know that when I am interested in a project, I am not quite sure who I can reach out to. Or if I am working on a Wikidata item, I want to know what the "best practices" are for that item, including the properties it should have. Effective, on-wiki collaboration tools can help us figure out the specific needs of Wikidata in a given area and then work together on addressing those needs.

Please review my proposal on Meta. I am eager to hear from you all about the tools Wikidata needs. Harej (talk) 09:45, 26 October 2015 (UTC)

It reads there for Wikidata:

    • Core
      • Identification of needs of Wikidata community and the development of at least one tool around these needs
      • Use of WPX UI design and tools on at least 5 WikiProjects
      • Creation or improvement of at least 500 Wikidata items through WikiProject tools
    • Stretch
      • Development of workflow for identifying item models and the development of five such models. An item model consists of all the features an "ideal" Wikidata item of a given type has. Example model: all biographical items should have date of birth, place of birth, occupation, etc.
      • Clarification of the process for the item showcase, and an increase in the number of showcase items to 50.

Not sure if this the best way to spend 30k on Wikidata. --- Jura 17:03, 26 October 2015 (UTC)

Jura, what would you recommend? I have ideas, but what's important is addressing Wikidata's actual needs; the goals can be adjusted accordingly. Harej (talk) 17:33, 26 October 2015 (UTC)
One interesting tool from my point of view: some automatic statistics board with some timeline in ordre to follow the development of some item classes. Snipre (talk) 19:43, 26 October 2015 (UTC)
That would be helpful.
To interest users at Wikipedia lists like Wikidata:WikiProject Association football/Wanted footballers/en might be a way. Ideally, they would be set up there.
To find showcase items, a dynamic tool might be the better way to go. In general, I think we are more interested in lists of items than individual items.
It might be worth looking around Wikidata and see what works. --- Jura 08:23, 27 October 2015 (UTC)

Notability by Freebase?

With Freebase moving to Wikidata, are Freebase items automatically notable by Wikidata’s standards? Otherwise, we’d lose data in the transition… —Galaktos (talk) 21:31, 26 October 2015 (UTC)

No, Wikidata keeps its own notability critera, and so, all Freebase topics are not notable according to Wikidata's standards. So, yes, we will lose data in the transition. Currently the Primary Sources tool only provides data from Freebase about items that are already in Wikidata. Tpt (talk) 09:08, 27 October 2015 (UTC)

Item creator

I get the message "Blocked user agent". What does this mean? Jonathan Groß (talk) 16:47, 27 October 2015 (UTC)

All pages under are currently down. --Stryn (talk) 16:49, 27 October 2015 (UTC)
What a pity. I hope they're working on it as we speak. Jonathan Groß (talk) 17:01, 27 October 2015 (UTC)
All the tools should work again :-) --Stryn (talk) 17:04, 27 October 2015 (UTC)


Is the Surname field restricted to humans, or should the Surname field be used for entries that are biographies too? We have hundreds of DNB biographies that do not have corresponding human entries such as Q19020700. --Richard Arthur Norton (1958- ) (talk) 20:56, 26 October 2015 (UTC)

Ash Crow
Harmonia Amanda
Чаховіч Уладзіслаў
Place Clichy
Jon Harald Søby
Sight Contamination
Aya Reyad
Tris T7
Klaas van Buiten
Bruno Biondi
Tokyo Akademia
ZI Jony
Da Dapper Don
Data Gamer
Luca favorido
The Sir of Data Analytics
  Notified participants of WikiProject Names Holger1959 (talk) 21:13, 26 October 2015 (UTC)
I think they should be restricted to persons. In the example you give, it is not a surname, just a part of the title... -Ash Crow (talk) 21:35, 26 October 2015 (UTC)
Normally, every DNB entry would have an associated item with P31:Q5.
There is a Wikidata:WikiProject DNB. Maybe User:Filceolaire can give you an idea on how he plans to go about. --- Jura 09:23, 27 October 2015 (UTC)
@Charles_Matthews: --Magnus Manske (talk) 13:23, 27 October 2015 (UTC)

Now MacDougall, John (DNB00) (Q19020700) does correspond to John M'Dougal (Q18731787), and the main subject (P921) link has been filled in. This is what should happen, and it is only a type of duplication of effort to add surnames to biographical articles that do have "main subjects" in this way, rather than on the items they are about.

In principle all the DNB item statements of this kind will be filled in, some day. Someone has to put in the main subjects, and in the relatively few cases where the corresponding item for a person doesn't yet exist, it should be created and tagged with the ODNB identifier. This just takes some time.

It isn't yet possible to quantify the DNB items missing "main subject". But that is because the initial creation didn't tag them all with "part of" "DNB edition", which would have been helpful (with hindsight). That should be a relatively easy automated task, starting from the Wikisource end. And after that, with the relevant items findable by query, the matching work could proceed more quickly. Charles Matthews (talk) 13:50, 27 October 2015 (UTC)

@Charles Matthews:
it should be very easy to do a list with Autolist, using the wikisource category - ;) --Hsarrazin (talk) 15:26, 27 October 2015 (UTC)

That is what I expected, but there is a caveat. If you look at s:Category:DNB biographies you can see the suffixes (DNB00), (DNB01) and (DNB12) all mixed together. Those are respectively parts of Dictionary of National Biography, 1885–1900 (Q15987216), Dictionary of National Biography, first supplement (Q16014700) and Dictionary of National Biography, second supplement (Q16014697). Ideally the items here should be "part of" the correct edition/supplement. For example, the volume numbers restart with each supplement.

Just having everything "part of" Dictionary of National Biography (Q1210343) works in one way, but there would be more work to do to take proper account of the suffix.

So a method that filtered by the Wikisource header templates {{DNB00}} and so on, e.g. with PagePile, would be more pleasing. Charles Matthews (talk) 16:19, 27 October 2015 (UTC)

No Richard. These are DNB items related to the biographies and contain details of the biographies (author, published in etc.) In every case there should be a separate item for the subject of the biography and that item should have the details of the subject, such as the surname etc.
Our first instinct was to just link the DNB biographies to the items for the subject. The problem with that was that many persons have more than one biography in en.wikisource (DNB and Britannica and Catholic encyclopedia for example) and wikidata does not allow us to have multiple links to en.wikisource from one item so we chose to do it this way with each biography getting it's own item. Hope that helps. Joe Filceolaire (talk) 19:33, 27 October 2015 (UTC)

We need "Subject:John Doe" at Wikisource. I lobbied for it there, I think they let me make one page to stop me from pushing for it globally. They only allow "Author:John Doe", so if you have three biographies about "John Food Smith III", who is not an author, you cannot tie them together. They also remove links to Wikidata if they fall within the text. That is a shame, John Tipperton Smith named in someones biography is just a name unless it is linked to a person in Wikipedia or Wikidata. The problem is too few people at Wikisource, so there are no dissenting opinions. Funny, there are only a few people here compared to Wikipedia, but we all seem open to experimenting with new properties. When are we going to become closed and dogmatic? --Richard Arthur Norton (1958- ) (talk) 22:29, 27 October 2015 (UTC)

Elo ratings

When I update Property:P1087 in items as for example Q608600 a "+1" gets automoatically added to the elo. What is that about? Looks wrong. --Gereon K. (talk) 14:09, 27 October 2015 (UTC)

It's about Phab:T105623. --Stryn (talk) 15:44, 27 October 2015 (UTC)
Ie. append "+-0" at the end of the number you add to avoid this. Matěj Suchánek (talk) 17:47, 27 October 2015 (UTC)
Works for me. Thanks for the workaround. --Gereon K. (talk) 22:14, 27 October 2015 (UTC)

What to include in a P131 administrative subdivision hierarchy for England ?

I've been doing some work recently trying to get a fix on where we currently stand with administrative subdivisions for England: see

There's a fair amount needing to be done; but the key preliminary issue that it would be really good to get some advice on is:

  • What should, and what should not, be included in the P131 tree ?

In particular,

  • should it tend to represent only actually active administrative areas ? (ie the very minimum).
  • or is it appropriate to try to fit in every structure that can be cleanly slotted into a nested structure, regardless of whether its significance is domestic, European, administrative or purely statistical ?

As an example of the latter, the EU defines statistical units (NUTS 2 and NUTS 3 areas) that do not exactly correspond to the administrative areas, but instead sometimes to groupings of them, that nevertheless nest cleanly into the next highest administrative level. Is it appropriate to include these in the P131 tree, or should they be excluded as not actually having any administration associated with them?

As a second issue, what would be the best way to deal with the so-called "ceremonial counties"? These are the area entities that English people probably most closely associate themselves with (so that we would definitely like to be able to field a question, who was born in this county), and they almost fit into a nested geographical structure, but there are a couple of aberrations: (See en:Subdivisions of England for the full gory details, with maps and everything).

  • The authorities of North Lincolnshire and North-East Lincolnshire are in the ceremonial county of Lincolnshire, the bulk of which is in the East Midlands region; however, these two authorities are in the Yorkshire and the Humber region.
  • The authorities of Stockton (mostly), Redcar & Cleveland, and Middlesborough are in the ceremonial county of North Yorkshire, the majority of which are in the region of Yorkshire and the Humber; however these three authorities are in the North-East region.

Should one use P131 for as much as one cleanly can? Or would it be much better to simply bite the bullet, and simply take the ceremonial counties out of the located in the administrative territorial entity (P131) tree altogether, and instead create a new sui-generis property for them? (On the basis that if P131, and therefore a P131 hierarchical path query can't be used for some of them, and therefore it should give a clean failure and no results for any of them, so that users don't get lulled into a false understanding by something that might work for most of them but not for all of them).

Advice would be very appreciated. (Particularly as we might work on some of this at a Wikidata training/workshop tomorrow (Saturday) in London). Jheald (talk) 17:45, 23 October 2015 (UTC)

Normally P131 would include current and former administrative layers. Given number of historic layers with a relevancy difficult to assess by non UK people, you might to focus on current ones only. Within the current ones, for localities or even buildings, ideally one would use stable and relevant ones. Try to avoid an issue some of France currently has: a marginally relevant outdated intermediary layer in P131.--- Jura 18:05, 23 October 2015 (UTC)
Thanks, Jura. Though to be honest, I am not sure I understand the difficulty of extra intermediates. So long as the nature of each item is properly identified -- ie what sort of thing it is, then surely it is easy enough for SPQARQL or Lua or Autolist to just keep going up the chain until it finds an object of the kind it wants -- and discard anything like NUTS layers that it's not interested in. The English structure is now so varied from place to place that that seems to me the only way to extract an item of the kind you want. On the other hand, including the NUTS items means that one would no longer be guaranteed that P131 would return the immediately superior administrative area for a particular area -- instead it might return a NUTS area that does not have such significance. Does that matter? Does it conflict unacceptably with people's established expectations? That's I guess what I have come here to ask this board.
One thing that does seem very clear to me is that the nesting has to be perfect, otherwise it won't be possible to iterate P131 to get to a single unique value for a particular level, and querying will break. So it seems to me that if one does want to use historical items earlier than a given point, they probably need to be segregated into a completely separate hierarchy, if P131 is to be used. Which is why there is already a problem with the "ceremonial" counties, which will only get worse the more historical one gets.
Again, it comes back to my fundamental question: is P131 intended to express a purely geographical relationship, that one area nests inside another -- or is it expected that it is also expressing a political relationship, that the larger area is a political entity that encompasses the smaller. This is reflected in questions of anachronism. Is it appropriate to indicate that an area that was abolished in 1973 is P131 a current region which didn't exist until 1994 ? If the relationship signalled by P131 is intended to be purely geographical then there is no problem. But if P131 is intended to express the structure of a particular political system, then the pairing would be a nonsense. Jheald (talk) 20:56, 23 October 2015 (UTC)
We need to bear in mind that people view and edit items here, without using SPARQL/Lua or whatever. So it does help to know directly that <somevillage> is in this department and not in some other. --- Jura 07:39, 24 October 2015 (UTC)
BTW, there is no requirement to have a single value in P131. --- Jura 08:10, 24 October 2015 (UTC)
I'm trying to stay out of this (despite being English) because English subdivisions are one of the most ridiculously overcomplicated ways of dividing up a country I've come across and it confuses me and makes my head hurt. As far as I'm concerned, you're either very brave or completely crazy! :P However, I did want to make a few points:
  • There is part of (P361) which could be used for anything which gets rejected from P131. A new property might not be necessary.
  • Don't forget that things can be linked to multiple items, e.g. North Lincolnshire could be linked to the Lincolnshire ceremonial county and Yorkshire and Humber region, while the current Lincolnshire county could be linked to the Lincolnshire ceremonial county and East Midlands region, and the Lincolnshire ceremonial county could be linked directly to England. That sounds like it would solve the problem of the ceremonial counties and regions not being strictly hierarchical without having to take anything out of the P131 tree.
  • I don't think P131 is designed for purely geographic relationships, but I think the important thing to consider is what is the most useful way to model the data? If taking ceremonial counties out of the P131 tree causes more issues than it solves, it's probably better to leave it in there.
  • I would definitely recommend not suddenly making a big change without giving people time to think about it and comment on it (since you mentioned maybe working on it tomorrow). It would be silly to change it in a rush and then potentially realise it was actually better the way it was. If the things lower down in the hierarchy are neatly linked to the thing directly above (and not to something several steps higher instead... like the 6,500 villages which are linked directly to England), it shouldn't be too hard to change the higher up parts later on if we want to remove ceremonial counties or regions or whatever.
  • It'd be a good idea to add a link to here on the property's talk page. There might be people watching the property page who don't regularly read project chat.
- Nikki (talk) 01:33, 24 October 2015 (UTC)
The above well describes how it looks like in Sweden (Q34) too. But I also have problems with that it is not fully clear if a county is in a province or if a province is in a county. In two cases, they perfectly match each other, but the rest is a perfect chaos. Stockholm municipality is located both in Södermanland province and Uppland province, but only in Stockholm county. Uppland and Södermanland province is also partly located in other counties. Those counties are located in other provinces that are located in other counties etc until you discover that you are located within Oslo metropolitan area or Murmansk oblast. -- Innocent bystander (talk) 08:10, 24 October 2015 (UTC)
Use 'replaced by'/'replaced' claims with 'applies to part' and 'start time'/'end time' qualifiers to indicate the dates when a division had legal significance and what divisions came before/after..
Ceremonial counties in most cases will be <instance of:county> with and end date and a 'replaced by' qualifier. They will also be <instance of ceremonial county> with a start date.
Where a division coincides with a NUTS2 division it can be marked as <instance of:Nuts2 division> as well as <instance of :English region (or whatever)>. I think NUTS divisions do coincide with administrative territorial divisions.
At least that is my opinion. Joe Filceolaire (talk) 03:30, 25 October 2015 (UTC)
Well, add "instance of: Nuts2 division" for Stockholm county would make it very complicated to add statements to such an item. Has head of government (P6) any relation with any nuts-region for example? -- Innocent bystander (talk) 13:54, 25 October 2015 (UTC)
On the other hand a region of England (Q48091) doesn't have a head of government (P6) either (apart from London), but I don't think anyone is proposing that they shouldn't be included; nor does a metropolitan county (Q769628) now have any associated elected body, but they are still quite significant administrative subdivisions for various services. Jheald (talk) 14:23, 25 October 2015 (UTC)
I am not saying that we should include or exclude any type of administrative entity in P131, based on if there is any elected body in it or not. What I prefer not to see, is statements like P6:John Doe, and we cannot conclude if John is "head of government" in a County or in a Nuts-region. Stockholm County and Stockholm County counsil have exactly the same geographic shape, but they have two different "heads of government", the Nuts2-region with the same shape have (of course) no head of goverment at all. -- Innocent bystander (talk) 16:33, 25 October 2015 (UTC)

Follow-up (Administrative subdivisions of England)

  • @Jura1: You write that "there is no requirement to have a single value in P131", and looking at Property talk:P131 you do appear to be correct; nevertheless it seems sensible to me in general to aim to have no more than one value that is unqualified.
The basic principle right across Wikidata is to try to give the most specific form of statement applicable, and let more general truths be handled by statements on the value items. This makes clear which is the immediate value, it exposes the hierarchical structure, and it minimises the amount that needs to be changed if parts of that hierarchy are altered (or if we decide to change our approach, as Nikki pointed out).
You write that it does help to be able to access the super-values directly. But the whole set is merely a click away on Reasonator, if the "link to Reasonator" gadget is enabled (an overwhelmingly good idea, for most users); or, even failing that, one can easily click up the chain of item pages.
So, as far as possible, I would prefer to have only the most local level of detail visible in P131 statements for places in England. This also seems to me what to aim for to get best results out of Magnus's "add Administrative Entity" game, which offeres the player sampled P131 values from the nearest other local places. Jheald (talk) 19:37, 27 October 2015 (UTC)

  • @Filceolaire: Ceremonial counties are still live; their more official name is "Lord Lieutenancy areas". But you may be right that a possible way to handle them might be as "deprecated" values for located in the administrative territorial entity (P131), with the qualifier reason for deprecation (P2241) = "administrative reorganisation", since their principal former administrative roles are now with unitary authorities or the reduced council areas. This would mean that they would appear on the subdivision items, but would not be found by "truthy" searches. Jheald (talk) 19:51, 27 October 2015 (UTC)

  • @Nikki: You make some very good points, in particular emphasising that only a small number of items are affected by this, especially if we work to try to get other items to have only the most local P131 value, so it's not so bad to leave the issue to re-visit later -- there's no shortage of other work to be done.
The issue with linking Lincolnshire ceremonial county direct to England is (i) how do you stop somebody else coming along and linking it to "East Midlands", which would then put all of it into the P131 tree; and (ii) how do you deal with people that currently have place of birth (P19) = Lincolnshire? Does one force this to be "Lincolnshire county council area" (which may not be correct); or does one lose them from searches for people born in the East Midlands? (And does one want to see them included in searches for people born in the North East, or not ?) Jheald (talk) 20:29, 27 October 2015 (UTC)

The most forceful response there was made by User:Nilfanion, who recommended the following as representing the administrative hierarchy of England (taking the label of located in the administrative territorial entity (P131) absolutely literally, even if that's not always necessarily the right way to interpret property labels here):
  • England
    • Regions
      • Metropolitan county/Non-metropolitan county/Greater London/Unitary Authority
        • London Borough/District (where relevant)
          • Civil Parish (where relevant)
This has the additional advantage of giving a reasonably regular structure, and nesting perfectly; although he did add that "Ceremonial counties would be highly desirable, but would not fit into P131 (especially Stockton-on-Tees) and are not adminstrative." Jheald (talk) 20:29, 27 October 2015 (UTC)

  • Another approach would be to see how far one can get by using qualifiers -- in particular applies to part (P518).
To try this out, I added P131 statements to the counties in North East England (Q47983)     , and in particular one to North Yorkshire (Q23086)      complete with qualifiers for the relevant sub-divisions, to see whether it might still nevertheless be possible to write reasonable queries.
In the simple case, where there is straightforward nesting, without any "applies to part" qualifications to have to consider, the recursive path specification
?item wdt:P131* ?top_level_adm
is essentially all that is needed to pull out all of the places and administrative entities located in the subdivision tree of ?top_level_adm. The page "/sub adms simple" gives a complete query.
In contrast, the query "/sub adms less simple" gives my current best attempt to write a corresponding query if there are "applies to part" qualifiers that need to be considered.
I think the query is structurally sound; though it only goes down a maximum of four levels, hard-coded, since one can no longer use the recursive path syntax; and probably one should check to make sure that none of the statements are deprecated, since to see qualifiers it needs to deal with the whole set of statements, rather than being able to limit itself to just the best most "truthy" ones. Nevertheless, it is significantly more involved and complicated that the simple query that is possible with the simple structure.
Also, for reasons I don't currently understand, the full query above doesn't actually complete in practice, but times out -- presumably due to a bug I haven't been able to find, either in the query or in the underlying Blazegraph engine. All I can say is that a query of this form restricted to two levels does appear to work:, but if the third level is activated, then for a reason I don't yet understand the query no longer completes. Any diagnoses very welcome! Jheald (talk) 21:18, 27 October 2015 (UTC)
Maybe the reality is not as simple as you tend to describe it? From 2016 in Sweden, the Districts will be the smallest administrative division. Most of them are, but some of them are not at all located in only one Municipality. The Parishes, that until 1999 was the smallest administrative division, were normally only in one Municipality. But the municipalities tended to change their borders more often than the parishes so they often crossed borders. The Civil parishes has part by part lost their administrative function. They were the smallest part of the local self-government until the 1860s. They kept many functions, as real estate-registration long into the the 20the century. They could then have different borders depending on which part of their administrative function you are referering to. They are still the smallest entity you refer to when you describe your etnicity. Ancient monuments are still registred in these entities. That function maybe will be replaced by the districts, but we do not know that yet. The Counties are the local representation of the central government. Their body is elected by the national government. Since the Municipalities were formed in the 1860s, no municipality have ever been located in more than one County, but they have sometimes changed county. That is the only typical hierarchy we have ever had in Q34. The County counsils are, despite their English name, not at all related to the counties. They are the typical Regional self-government with an locally elected administration. Neither these have ever crossed the borders of a county. But there can be more than one county counsil in a county and there can be county counsil-free areas within a county. Today, Gotland is the only county counsil-free area in Q34. Regions is the new name of County counsils with extended priviliges. But the name Region is also used for many local cooperations between municipalities and county counsils. Provinces have not had any administrative functions since the 17th century. Their borders have been fixed since then. When you ask somebody were (s)he comes from, the person will most likly tell you from which Province. It was the provinces that once formed the nation of Q34. The borders of a province may cross the borders of municipalities, counties or whatever. I still do not know if there is a fixed relation between districts and provinces, that is still to be proved. Q34 is also divided in three countries (landsdelar). They have no administrative functions, and are mostly used in weather reports. The borders between two of them are sometimes disputed, and depends on who you ask.
I have not started to talk about statistical entities, like Nuts, Urban areas, LA-regions, Metropolitan areas etc etc. These sometimes even cross the national borders.
You also have Ethnical areas like Sapmi that cover parts of four nations. Neither does Bothanical areas, Drainage basins etc have any relation to any man made administrative borders.
It is fiction to describe the administrative hierarchy as simple. It is simple only if you remove 95 % of them and forget all of their history. -- Innocent bystander (talk) 07:56, 28 October 2015 (UTC)

Update: I have now done a bit of work on the upper parts of the UK administative tree (ie regions / counties / districts) -- I haven't touched civil parishes yet.

In the end I decided to allow the ceremonial counties as values for P131 statements, and to accept them as full-status members of the tree.

It's still work in progress, but the data is now good enough to extract a reasonably good structure with this query (Run on query editor), that can be compared with the page at en:Subdivisions of England.

The issues with Lincolnshire etc have been addressed by the hard-coding filters at the end of the query, to reject an authority being put with the wrong region. Yes, it's a trap for the unwary. But I thought it was worth it to include the counties. Since there are so few examples where the nesting fails, it should usually only take a few lines of query text to explicitly check for and reject incorrect associations. (Though you do have to be aware of the issue).

I have also started creating new "county council area" items for parts of counties where the county council doesn't now administer the whole of the ceremonial county of the same name. These areas would be appropriate targets for applies to jurisdiction (P1001) for the items for county councils, like Essex County Council (Q5399679). I have only done some so far -- progress summarised in this query (hit 'execute' to run), so I would welcome people's input as to whether they think this is on the right track. Jheald (talk) 17:51, 29 October 2015 (UTC)

Wikispecies batches

Just as a heads-up, I will be running a lot of Wikispecies matches and item creations today and tomorrow. The matches have been checked by name, uniqueness, and taxon rank in the respective Wikispecies pages; the de novo creations have been checked to not exist as a label or taxon name. But as these will be many tens of thousands, individual edits can be wrong. Please keep an eye out, and alert me if there is a systematic issue. I will use User:Reinheitsgebot for this. There are still ~175K Wikispecies pages without Wikidata item, so using bots where we can is the only viable option; there will be enough "stragglers" left as it is. --Magnus Manske (talk) 09:16, 27 October 2015 (UTC)

@Magnus Manske: I don't think this is a good idea. Wikispecies includes lots of taxon names that are invalid, dubious, deprecated, or that have been synonymized with other taxons long ago. Wikispecies has just never had enough eyeballs to be organized and up to date. I think mass importing items from Wikispecies may do more harm than good as far as Wikidata's taxonomic data quality. Kaldari (talk) 17:20, 29 October 2015 (UTC)
I see that I'm probably too late. In the future it would be good to have more than 1 day's notice of mass imports. Kaldari (talk) 17:22, 29 October 2015 (UTC)

STOP with bot import

We will have to choose: quality or quantity. But we can't continue like that. I spend several weeks to clean chemical items using the constraint report (reduction of constraints violation during the last months) and I just find that most of my corrections were erased due to an import form the Serbo-Croatian Wikipedia (increase of constraints violation in the last week).

We can't work on the items, clean them and at the end just see a bot import destroying all the previous work. Can we definitively forbid bot import from Wikipedia according to the facts that 1) WP is not a source and 2) we have now enough data on WD now to start to work ? Snipre (talk) 13:01, 27 October 2015 (UTC)

  • I haven't looked at your case in particular (which sounds frustrating enough to be sure), but I think that we (a) do not have do "decide" between quality and quantity, and (b) we do not have enough content to scoff at Wikipedia yet. We may have enough in some specific areas, but with >4M items without even instance of (P31)/subclass of (P279), there is still a whole lot of whitespace to be filled. The thing we should take from your experience is that we need to be more careful with indiscriminate mass import (ironically, while writing this, I am mass-importing from WikiSpecies...). For example, we could detect manual edits of a specific statement (as opposed to bot edits) from the item history, and make bots skip these on mass edits. However, this would require much better bot frameworks than currently available, as such functionality would have to be "abstracted away". Also, constraint checking before a bot edit would be such a function, as would "revert detection". --Magnus Manske (talk) 13:32, 27 October 2015 (UTC)
You always do a choice between quality and quantity. By increasing filters to get high quality data you always loose some data and at the end you have less data than if you didn't filter your data.
And I think it's time to decide if we want to invite people from Wikipedia to improve Wikidata or not. I had enough complaints from WP contributors saying that data from WD is bad or wrong (and that data is coming WP!!!) and I can't propose them to come in WD to correct because there is a high risk to see their corrections erased.
The critical thing is errors shared between several WPs. You import the error once from WP:xx, then you correct in WD, then you import from WP:yy and you see the same error again. Then you have the constraints system on WD which is not applied in some WPs.
We can't import 5 times the same data after one or two imports fromWP we should close that raw data to go to the direction of authorized databases in order to clean and complete data sets. Snipre (talk) 22:37, 27 October 2015 (UTC)
Well, your initial sentence "We will have to choose: quality or quantity" implies either-or. Of course this is a more "fluid" scale, where we should increasingly look for quality. Which is what I said in my reply. --Magnus Manske (talk) 12:25, 28 October 2015 (UTC)
I like the idea of detecting if things have been removed before. Letting people know if they're re-adding something that was deleted sounds like it would even be useful for people editing on the website by hand. - Nikki (talk) 15:15, 29 October 2015 (UTC)
  • As a note, I do semi-automated edits on my work account, and I plan on doing some as a volunteer as well. I don't use Wikipedia as a source (as a Wikipedian of 11 years, I cringe at the thought ;), but if any batch edits I do manage to screw something up despite my meticulous planning, please let me know immediately. I will take responsibility for my own messes. Harej (talk) 17:38, 27 October 2015 (UTC)
@Magnus Manske: et al. "with >4M items without even P31/P279, there is still a whole lot of whitespace to be filled". When P31/279 is the only statement in an item, I am not sure it gives much value other than filling the charts at Wikidata:Statistics. And I am not interested in filling those charts. I have started to add "Wikidata-related maintainance categories" on svwiki. "Articles missing basic properties P31/P279", "Articles with labels in less than two languages" (it is not of much value to you here, if the only label is in Swedish) and others. This far, I have only added this to featured articles, but it is a start. With a few exceptions, sitelinks is the only thing the sv.wikipedians are interested in here. I hope this can change that. -- Innocent bystander (talk) 18:25, 28 October 2015 (UTC)
It was just an example of how much there is left to do, and that it's not "we can do the rest by hand" time yet. And even just an instance of (P31) can be helpful beyond mere statistics; it can, for example, attract people who like to work on a specific subject area, by hand or bot. Like you did on svwiki, around Featured Articles. Labels are another important aspect of this. Again, I think bots can help here (which is the original topic of the thread); arguably, people's names in German will most likely be the same in English. Taxon names will mostly be correct labels for many languages, unless there are "trivial names". But many require manual work indeed; I look forward to get some more Swedish speakers from Wikipedia here! --Magnus Manske (talk) 21:33, 28 October 2015 (UTC)
@Magnus Manske: I do not think "we can do the rest by hand time" ever will come. The number of new articles in the X-hundred clients will outnumber us here for time indef. -- Innocent bystander (talk) 09:15, 29 October 2015 (UTC)
About labels and translations: There was also interest from Wikiproject Medicine to get more people involved in translations in order to provide basic medical information in as many languages as possible. I think it would be an excellent thing to gamify. For example a game on the desktop could involve typing. On mobile we could connect to some translation-API with which we could suggest one translation and have a yes/skip/no button. That would also help fill the various Wiktionary sites as soon as we can connect to them. I started to see how this could be done and discovered that Google is already gamifying translations: - They did a really good job on that page and the game is responsive, fun, and gives out lots of points and badges. It really shows that we should invest more time in creating such games to harness swarm intelligence. --Tobias1984 (talk) 21:56, 28 October 2015 (UTC)
There's still huge amounts of data we're missing and I don't think it makes sense to prevent all bots from importing data just because there are some bots which are misbehaving. That would be like preventing all humans from editing just because some of them are spammers or vandals. Have you let the person operating the bot know that there's a problem? - Nikki (talk) 15:15, 29 October 2015 (UTC)

I think that there has to be a possibility to protect single snaks, statements and whole items against bot edits. This possibility (user right?) should be given to any regular Wikidata editor. Bot edits to get quantity to Wikidata are good in many cases, but after a human being has cleaned up certain statements or even a whole item, this work should not be overwritten by a bot. Furthermore, there should be a way to protect existing statements of an item against bot edits while allowing bots to add further statements for that item. Is there anything that opposes this suggestion? Otherwise we should ask the developers to develop different protection possibilities. Yellowcard (talk) 08:57, 29 October 2015 (UTC)

I don't think that would be a good idea. Not all bots make the same sort of edits and most bots are making useful edits, so it would prevent good edits as well as bad ones (nothing is ever really complete). If a bot is making repeated mistakes, it's better to let the person operating the bot know so they can fix it (and if it continues to be a problem, we already have the option of blocking the bot). - Nikki (talk) 15:15, 29 October 2015 (UTC)
@ Nikki: What is a good edit to a snak, once a human being has checked and sourced this snak? I'm not thinking about misuse of the bot, but simply the fact that a bot cannot decide if certain data's quality is good or bad. Yellowcard (talk) 15:52, 29 October 2015 (UTC)
It's not only a bot, it's a general problem. We need a quality and a security system. If we want for example use Wikidata for replacing en:Template:BibISBN we have to be able to mark an item with a kind of flag system. Flag 1: this item has been checked and can be used for citation. Flag 2: this (protected) item is used for citation and can't be changed by an IP or regular bot. --Kolja21 (talk) 16:07, 29 October 2015 (UTC)
+1, sounds good. Has there ever been a discussion about such a system? Yellowcard (talk) 18:40, 29 October 2015 (UTC)
My impression is that bot operators simply too seldomly propose and discuss the new tasks they are inventing for their bots after they have gained the bot rights and performed the initial task(s) the permissions had been granted for. Other users may know about (general or specific) pitfalls with regard to the dataset for the intended action and they even might have some overview about the level of manual cleanup already invested to the properties in question. So perhaps an announcement of the intended bot action on the discussion pages for the affected properties could prevent most damage? -- Gymel (talk) 18:23, 29 October 2015 (UTC)

Statistics Finland - Can I use these on Wikidata?

See General Terms of Use for free-of-charge data. There is that you can use the data but you have to mention the source (© Statistics Finland) etc., so I want to make sure it's ok. For example is very interesting and would be nice to add those data (all postal codes, Finnish and Swedish name of villages of Finland, NUTS codes and much more) on Wikidata. --Stryn (talk) 19:00, 27 October 2015 (UTC)

meta:Wikilegal/Database Rights may be helpful to you. I think it says you, mostly, can't import from CC-BY databases. Joe Filceolaire (talk) 03:57, 28 October 2015 (UTC)
@Stryn: According to the General Terms of Use, the corresponding common licence would by Creative Commons BY and not CC0. I am not a specialist but the General Terms of Use require the mention of the Statistics Finland service as author. Snipre (talk) 13:44, 28 October 2015 (UTC)
Yes, I realized, thanks. --Stryn (talk) 15:46, 28 October 2015 (UTC)
@Stryn: We had similair problems with Statistics Sweden. They demanded credits like those Statistikcentralen claims. But after some talks between the Swedish Wikimedia chapter and SCB, most (but not all) of the data is now free to use here. -- Innocent bystander (talk) 18:08, 28 October 2015 (UTC)
@Stryn, Innocent bystander: We need a OTRS ticket to be clear about that. We can't just import the data because someone has an email from a guy working in an administration saying we can do it. We already got in touch with OTRS system but we were waiting on Wikilegal comment in order to edit a template for data set. Snipre (talk) 09:39, 29 October 2015 (UTC)
@Snipre: In the case of Statistics Sweden, they have changed their general Copyright to most of the publicly published data. It is not information we have got in an e-mail explicitly for us. They changed their Copyright-policy after some of us met them and explained that their data wasn't "free enough" for us to use. They also do reports for private companies, and such data was not free before and is still not free. But such data is normally not publicly published on their website. -- Innocent bystander (talk) 10:16, 29 October 2015 (UTC)

Happy 3rd Birthday, Wikidata!

Hey everyone :)

Today we are celebrating Wikidata's 3rd birthday. I've been with the project since we started development 3.5 years ago and I can't believe what a ride it has been and how far we've come.

As for every birthday celebrations are in order. We've created a page at Wikidata:Third Birthday. There you can find editorials (by Harmonia Amanda, Ash Crow and me) about the past year and what is coming. Please take a moment to read it. There you will also find a section for congratulations and wishes, presents and more.

Here's to many more years of Wikidata. Stay as awesome as you are!

Cheers --Lydia Pintscher (WMDE) (talk) 23:29, 28 October 2015 (UTC)

PS: The development team has presents as well. I'll send an email about them in a few hours. Ohhh the suspense :D

Distinguishing between names and taxa

I would draw your attention to the proposal at Wikidata talk:WikiProject Taxonomy#Distinguishing between names and taxa. It touches on issues which apply to more than just biological taxonomy. Your comments will be welcome there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:27, 29 October 2015 (UTC)

birthday present: improved UI

Hey folks :)

Here comes birthday present number one from the dev team. Jonas took some time and reworked the UI of to make it prettier and easier to understand. It also has more examples now. You can expand them on mw:Wikibase/Indexing/SPARQL Query Examples. The auto-completion has also been improved. Check it out! I hope you like it.

Cheers --Lydia Pintscher (WMDE) (talk) 10:57, 29 October 2015 (UTC)

+1. Snipre (talk) 19:04, 29 October 2015 (UTC)

Fusion problems

Can someone fusion en:Ben Hodges with German article de:Frederick B. Hodges (Q19798487) ? 00:47, 30 October 2015 (UTC)

  Done --LadyInGrey (talk) 01:11, 30 October 2015 (UTC)

Can someone fusion en:Seventh United States Army (Q7457752) with German article de:Seventh United States Army (Q181197) ? 01:21, 30 October 2015 (UTC)

Not possible, there are two different English articles.--Ymblanter (talk) 02:28, 30 October 2015 (UTC)

Can someone fusion en:Category:United States Army generals (Q7329430) with German category de:Kategorie:Kommandierender General (United States Army) (Q16660620) ? 01:30, 30 October 2015 (UTC)

  Done, if you'd like to do this yourself, take a look at Help:Merge de (sorry for the late reply after performing the merge, I got distracted) --BurritoBazooka (talk) 03:37, 30 October 2015 (UTC)

National teams vs. Women's national teams

Hi, has anyone a good idea what statement to use to distinguish national teams from women's national teams? I used competition class (P2094) with the value women's association football (Q606060) for women's national teams to express this fact; however, I got reverted by user:La femme de menage ([11], [12], [13]). Has anyone a good idea how to express the difference by using a statement? Yellowcard (talk) 09:12, 29 October 2015 (UTC)

FYI - I explained to Yellowcard on my talk page the reasons of this reverts, and the opposite need to identify that a lot so-called football national teams were in fact football men's national team, which cannot be understood just by reading the name. --La femme de menage (talk) 13:34, 29 October 2015 (UTC)
Again, football national teams are NOT men's national teams as women are also permitted to play – but not the other way round. However, this is not what my question was about, but how to express this by a valid statement. Up to now I consider my version a good way to go, but I'm open to any other suggestions. Yellowcard (talk) 15:54, 29 October 2015 (UTC)
If women are allowed in "men's team" depends on the Sport. I know, in bandy (Q183018) a Finnish woman was denied to play in the national league. It probably depends on both the sport and which tournament they play in. -- Innocent bystander (talk) 16:03, 29 October 2015 (UTC)
Oh sure, that might be. I was referring to soccer. However, the question is how to determine the class of people who are allowed to participate in a certain team. In other sports, it might be necessary to mark men's team also, of course. Yellowcard (talk) 18:42, 29 October 2015 (UTC)
Hi. FYI, the Regulations of the 2016 UEFA European Football Championship specifies that „Every UEFA member association may enter its senior men’s national team for the competition“ (article 3.02). In the UEFA point of view, senior national team is men's or women's one, not both. --H4stings (talk) 13:54, 30 October 2015 (UTC)
Don't forget the Under 21's team, the veterans, the mixed doubles, and all the multiple paralympics classes. We definitely need a property to identify these but I have no idea what to call it. Joe Filceolaire (talk) 19:29, 29 October 2015 (UTC)
@Filceolaire: We do have a property, it's competition class (P2094). However, it might be that the labels in certain languages are not appropriate, yet. It works for Under XX teams, see for example Switzerland national under-23 football team (Q769632). It should also work for all the other groups you mentioned, we just have to find the linkable iteam for each group. Yellowcard (talk) 19:33, 29 October 2015 (UTC)

Related question (Sport participants)

Is there a way to specify who is eligible to participate in a competition or event?

E.g. a women's tennis open? Or a football league for teams in England ? Or for youth teams from London ? Jheald (talk) 23:46, 29 October 2015 (UTC)

Wikidata Birthday: ArticlePlaceholder

It’s Wikidata’s third birthday! Wohoo \o/

So it’s finally time to show you, what I worked on over the past weeks and months. I started working on an extension, called ArticlePlaceholder as part of my Bachelor’s thesis at the HTW Berlin in cooperation with Wikimedia Deutschland and espacially with the Wikidata team.

The idea is, to have automatically generated content pages on the different Wikipedias displaying data of items that don’t have an corresponding article in that language. This is will be especially helpful for smaller Wikipedias with a small contributor base and aims to make more knowledge accessible to more people. On the long run we even might be able to reduce the number of bot generated stubs. This will mainly support Wikipedia editors in maintaining but also in editing.

I’m very excited to let you know, that there is actually something that I can show you! So instead of telling you in a lot of words, how cool this all is going to be, I want you to just see yourself:

This is for example an auto generated placeholder for Ada Lovelace.

It’s all very much work in progress, I just wanted to give you a first sneak peak. :) So, just in case you wonder, even though Special:FancyUnicorn is an awesome name for a special page I will obviously change it to something more fitting. Also, the design of the page isn’t done at all yet. So far I just bothered about the loading of the data rather than the layout. I just chose something, that would be easy to read for people familiar with Wikidata for this presentation. But you know what is actually super cool about the layout? It’s completely written in Lua- so you can overwrite every part of the (upcoming super beautiful default) design and make it fitting to exactly what your local community wants and needs. Only the title and the language links are always set, the rest is adjustable with on wiki scripts. Awesome, right?

There are some other bugs I already know of ( and probably a lot more I have no idea about yet, so if you find them, feel free to file a bug.

The documentation is not up to date yet either, I will take care of that in the next days. But here is the page anyways:

So have fun discovering the article placeholder! I’m looking forward to your feedback.

And happy birthday Wikidata! I just want to thank you all for the chance to be part of all this awesomeness, all the help and support over the time, and the great work you all do! <3

You rock! Lucie (Frimelle)  – The preceding unsigned comment was added by Frimelle (talk • contribs).

Interesting start. But this looks like too much to Wikidata items at that level and perhaps somewokr on the display will help to feel more in the Wikipedia environment (a lot of contributors don't want or like to be in a different than the one they are used to). Then if you can find a way to group similar properties (properties about death and birth, properties about family, properties about activity, and the famous authority identifiers), this would do the difference. Good luck for the end of your thesis. Snipre (talk) 20:18, 29 October 2015 (UTC)
As far as I understand, the grouping will be open to the Lua-programmers. I don´t understand how Lua-Modules and special pages could work together? Maybe there is a Module called "FancyUnicorn", which Wikipedia editors are able to edit, that is formating the special page? --Molarus 00:45, 30 October 2015 (UTC)
Yeah there will be several Lua modules which will provide the default and that each can locally be overwritten. Pretty much all of the display can be controlled this way. --Lydia Pintscher (WMDE) (talk) 10:34, 30 October 2015 (UTC)

No world record(s)?

Hi there,

I've found the property for Record held (P1000), where a date can be attributed, and even an item "100 metres dash" (Q164761), but in neither case can I find a manner to give a value for these (e.g. 9"58). Am I missing something? Leo Fischer (talk) 12:19, 30 October 2015 (UTC)

See Wikidata:Property proposal/Event#Time, where this has been proposed. Thryduulf (talk: local | en.wp | en.wikt) 14:14, 30 October 2015 (UTC)

How many items about people (P31=Q5) have sex or gender (P21) defined?

There is a comparison by Wikipedia at sitelinks and P21.

Most items have sex or gender (P21) defined (94%). This has increased slightly since June (93.8%).
There is some gap in items linked to the Chinese (46%), Korean (75%) and Japanese Wikipedia (80%). Given their size, these have fairly large impact, also due to the low overlap with English Wikipedia (Q328).
Items with no link to Wikipedia have a fairly low percentage as well (80%). Languages written in Latin script with low percentages are: Turkish, Azerbaijani, Vietnamese and Indonesian.
Lithuanian Wikipedia (Q202472), one of the Wikipedias with the least overlap with English Wikipedia (Q328) (only 45%), reached 100%.

--- Jura 12:53, 30 October 2015 (UTC)

My article

I cannot locate my article on A Gifted and Talented Musician Tom Powers - it was once in edit mode - now I have no idea where it is. And can I contact Gorilla Warfare from your site? She is an editor.

I find your site extremely confusing and convoluted - OMG its amazing there is any information published - it is so sad that I cannot write something about Tom Powers the talented and note worthy accomplished musician - And the article that I had started was all composed by me. Some of the information I did copy was from my husbands website, but I composed that as well

Thank you, Joann Powers Joannpowers (talk) 18:06, 30 October 2015 (UTC)

Try The English Wikipedia (this is Wikidata, a different project, we don't handle articles). Complication is necessary when you are only a couple thousand editors but have to handle millions of articles and probably over a hundred languages(?). This project's (which is Wikidata and not Wikipedia) initial goal was to try and coordinate between all the different languages Wikipedia articles are written in. --BurritoBazooka (talk) 18:33, 30 October 2015 (UTC) edit: I found your draft article on the English Wikipedia. Draft:Tom Powers (musician) --BurritoBazooka (talk) 18:36, 30 October 2015 (UTC)

Python script to add interlanguages link automatically

Hello all, after a mass commit in the French wikipedia, we would like to link the new French articles to their counterparts in other languages. The list can be seen here [14] (see 2nd section, missing items). We know that the pages are already available in eo,pl,it (at least), and for the big part of them (99%), titles will be the same or could be easily deducted. Can someone help in pointing a good python script that could help us create the wikidata interlanguages links . I tried with the -hint options, but it does not seem to save the page though the script is able indeed to retrieve the interlanguages links ! thanks to who can help --Gallicbot (talk) 10:22, 31 October 2015 (UTC)

If your primary objective is to quickly have interwiki in the pages, why not add the "old style"-interwiki-link to them? An interwiki-robot will most likely very soon replace these links and add these pages to Wikidata. -- Innocent bystander (talk) 11:03, 31 October 2015 (UTC)
Are there actually any bots still doing this? I still come across old-style interwiki links regularly and it doesn't seem like any bots ever come along and link them. - Nikki (talk) 19:08, 31 October 2015 (UTC)

With QuickStatements (Q20084080), you can add them directly to Wikidata.

You need the article title in another wiki. A sample from Q1191484 for pl would be:

(5741) Akanemaruta	Sfrwiki	"(5741) Akanemaruta"

and select "first column are articles from:" "plwiki".

If the article titles aren't safe, you would need to match them up with Wikidata items first and then use the format:

Q1191484	Sfrwiki	"(5741) Akanemaruta"

Hope this helps. --- Jura 11:42, 31 October 2015 (UTC)

Hello Jura! this helps indeed ! merci beaucoup, with the CREATE tag (which is required here), I'll be able to complete the whole task!! --Gallicbot (talk) 14:37, 31 October 2015 (UTC)
CREATE makes new items (if we don't have one yet) .. BTW, the tab in my sample isn't rendered correctly, one needs the version in the wikisource text of my comment.
If the first version doesn't find an article, it just moves to the next line. --- Jura 14:41, 31 October 2015 (UTC)

displayed value and uncertainty do not match entered value and uncertainty

The number of participants of Colectiv nightclub fire (Q21282871) is given as 200-500. As we can't enter a range, I entered 350±150 but this is displayed as 400±200 - which is incorrect. Is this Phab:T105623 or something else? Thryduulf (talk: local | en.wp | en.wikt) 18:49, 31 October 2015 (UTC)

I think that's phab:T95425. In the JSON (Special:EntityData/Q21282871.json) it has 350 with 200 as the lower bound and 500 as the upper bound, so it seems it's stored correctly, it's just being rounded for display, which matches that ticket. - Nikki (talk) 18:56, 31 October 2015 (UTC)
Thank you, I've commented there. Thryduulf (talk: local | en.wp | en.wikt) 20:14, 31 October 2015 (UTC)

Template broken

The template {{Pre2}} is broken. I tried to fix it, but I couldn't fix it. Can someone with some more knowledge of Wikitemplates take a look?

This is how it currently looks

. Mbch331 (talk) 20:04, 31 October 2015 (UTC)

I changed something [15]. Paweł Ziemian (talk) 20:29, 31 October 2015 (UTC)

Wiki extension

I propose to develop a web-server. The server should be very easy for installation and using. This web-server and templates of Wiki web-page should be distributed free of payment. The main idea: everybody can download and install a simple web-server and publish own web-page (articles, books, etc.). Indexing of the new user`s web-page in the Wiki will be after reviewing the web-page. As the result: i) extension of Wikisource, ii) The data will be stored on the user`s hardware, iii) Increasing of IT education of Wiki users/autors :) -- – The preceding unsigned comment was added by (talk • contribs). at 21:58, 26 October 2015 (UTC)