Wikidata

Wikidata:Project chat/Archive/2014/01

< Wikidata:Project chat‎ | Archive‎ | 2014
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

Clarification on Template:[…]/doc items

Hi there! I have a doubt: according to Wikidata:Notability /doc subpages of templates shouldn't be moved to Wikidata, am I right? Just because I found plenty of them. Thanks for answering me. — TintoMeches, 21:35, 19 December 2013 (UTC)

Yes you're right. /doc kind of pages should be deleted. But then someone will add them propably back here, so I don't know... should we made an abuse filter to block doing so.. --Stryn (talk) 21:56, 19 December 2013 (UTC)
A filter would help identify new links being added. We also need a bot to create a list of items already on Wikidata that have mainly /doc links. Del♉sion23 (talk) 22:18, 19 December 2013 (UTC)
Legoktm as far as I know believes /doc pages should remain on Wikidata. Just pinging him so he can justify himself here. John F. Lewis (talk) 23:14, 19 December 2013 (UTC)
If there are /doc pages in multiple languages, I don't see why they shouldn't have Wikidata items. Legoktm (talk) 04:57, 20 December 2013 (UTC)
We normally have interwiki between the templates, not between their documentation. -- Lavallen (talk) 08:31, 20 December 2013 (UTC)
It was quite common practice to put interwikis for templates into documentation pages. --EugeneZelenko (talk) 15:34, 20 December 2013 (UTC)
Yes, but they were transcluded to the template, the documentation itself had no interwiki. -- Lavallen (talk) 16:08, 20 December 2013 (UTC)
I agree with Lavallen, the Item is the template itself, not the documentation subpage. — TintoMeches, 16:20, 20 December 2013 (UTC)
Two quick points. First, it's trivial to modify bots so that they stop creating items for subtemplates and as far as I can tell, nobody is manually mass creating these items for interwiki purposes. Secondly there is a clear downside to including these items. Indeed they show up in search results. Mergehappy (talk) 16:52, 20 December 2013 (UTC)
I'm not directly against adding items about documentation, if we find any good purpose for it. (This far, I fail to see any such purpose.) But if it give us interwiki between documentation-pages, I'm afraid it will confuse the users in WP et al, more than it gain. -- Lavallen (talk) 18:37, 20 December 2013 (UTC)
Users who worry about /doc pages in the template name space are typically quite tech-savvy, I wouldn't worry too much about confusing them. On the other hand, I would worry about creating items that explicitly go against the current policy. Mergehappy (talk) 20:05, 20 December 2013 (UTC)
And I do not worry about explicitly acting against current policy, when it gain the project. Our policy are some random bits in a server, not rock-carvings. -- Lavallen (talk) 15:05, 21 December 2013 (UTC)
We shouldn't have to create item page for documentation, just because documentation pages are sub-pages like the page discussion of article, which don't have any wikidata item... Moreover, lot of template item are partially mix with documentation link. --Nouill (talk) 11:36, 23 December 2013 (UTC)
bugzilla:28604. Helder.wiki (talk) 19:54, 24 December 2013 (UTC)

Since it appears to be consensus, can we start doing a bulk deletion request? — TintoMeches, 11:43, 23 December 2013 (UTC)

Please don't forget that sometimes (by error) some /doc pages has been linked to ordinary templates. It should be checked and changed by removing only prefix "/doc" in link and label. --Infovarius (talk) 08:21, 26 December 2013 (UTC)
Thank you, I'll pay attention. — TintoMeches, 14:04, 28 December 2013 (UTC)
There's an ongoing discussion related to this topic at RfC: Interwiki links for subpages. — TintoMeches, 14:16, 1 January 2014 (UTC)

Language support

From Wikidata:Contact_the_development_team/Archive/2013/03#list_of_supported_languages, it seems that the list of languages supported by Wikidata is defined in the MediaWiki source code file https://git.wikimedia.org/blob/mediawiki%2Fcore.git/master/languages%2fNames.php (and a messages file in here except for these). One way for a language to be added to that list is to localise the mediawiki software. I think that is the only way still available, but I would love to be proven wrong. That is the case even for extinct languages, like 'grc' (Ancient Greek) and 'ang' (Old English). Of course it is silly to use MediaWiki in Ancient Greek or Old English. Consider w:Aramaic language, which has 23 ISO 639-3 codes; should we create 23 largely ficticious translations of MediaWiki interface in order to use those language? It is also (nearly) impossible for the MediaWiki interface to be translated into dying languages, like w:Tiwi language, due to the practicalities of limited number of speakers - usually old people. Wikidata could be a repository of translations of terms in languages which don't have a MediaWiki translation. We already have word lists, like wikt:Appendix:Swadesh lists for Australian languages. And old people would be happy to help add translations of important concepts like man/woman/rice/cow/etc, but they are less interested in translating messages like MediaWiki:Bad image list, which is one of the 500 most used messages and must be translated before MediaWiki will add the language. With such limited support of languages, we often cant include the native name of an item. I've put up a list of missing languages at mw:Unrecognised languages. But that is only >1million native speakers. Look through the list at w:List of indigenous language names to see many more languages in this category.

When we look at incorporating Incubator, as supported at Wikidata:Requests_for_comment/Items_for_Wikimedia_projects_besides_Wikipedia#Incubator_pages, we'll need to allow languages which do not yet have an interface translation, as it is quite common for the Incubator project to start before the interface translation even begins. As mentioned above, the same problem occurs with Wiktionary, where translations for words exist in languages without a Wikimedia project. There are also Wikivoyage phrase books for languages which there is no Wikimedia project.

An improvement would be using the translatewiki:Special:SupportedLanguages list, which has a clear process at translatewiki:Translatewiki.net languages. Another option is for Wikidata to use mw:Extension:CLDR, which is installed on Wikidata. That is the approach taken by the Incubator project with its mw:Extension:WikimediaIncubator.

p.s. it will be a lot easier to translate mediawiki if the language is available here first, as many of the messages map to discrete items here. See User:John Vandenberg/MediaWiki interface. John Vandenberg (talk) 17:36, 29 December 2013 (UTC)

Wikidata, like every MediaWiki instance, can add more languages (even though they won't have their own interface). The first additional language was approved by the LangCom and is being added (bugzilla:57342). --Nemo 13:32, 31 December 2013 (UTC)
Excellent. @GerardM:, what is the process to request new languages? John Vandenberg (talk) 16:18, 31 December 2013 (UTC)
The process is that you ask the language committee for a language to be added. The language is approved for use in Wikidata when it has an ISO-639-3 code. The right is reserved for Wikidata only languages to review the content. When it is found to be problematic, a request can be made for it to be removed. Thanks, GerardM (talk) 07:57, 1 January 2014 (UTC)
Thanks GerardM. This sounds very promising. I don't completely understand what you mean by "The right is reserved for Wikidata only languages to review the content." Could you rephrase it, or expand on it, to help me understand what you mean? John Vandenberg (talk) 08:07, 1 January 2014 (UTC)
Languages can easily be added to Wikidata. The only requirement is that it is recognised by ISO. When we find that the trust to add content in that specific language is abused (it is not that language), the content will be removed. GerardM (talk) 08:11, 1 January 2014 (UTC)

Categories and pages

This may have been discussed. I just wonder why there are separate Category pages and pages, like Category:Psidium (Category:Psidium (Q8804336)) and Psidium (Psidium (Q320179))? In biology at least this refers to the same concept, and the concepts may be identical in most other areas also. Can't we get them on the same page? JMK (talk) 11:09, 30 December 2013 (UTC)

We use the category main topic and the converse property to link them. It was not in the initial Wikidata design to add the categories to an item, so this is a good solution. TomT0m (talk) 11:19, 30 December 2013 (UTC)
If I may say so, I think this leads to extra work and confusion. If someone could ever get them on the same page, I would be very pleased. JMK (talk) 13:30, 30 December 2013 (UTC)
As it is now, there seems to be thousands of biota that do not have category pages, but most do have normal pages, and a template on commons category pages reminds me time and again that the a wikidata page for it has not been created yet. I'm not going to create them all, as it would be impossibly cumbersome and time wasting, and I don't think they should be created. In short, I feel there should not be any stand alone category pages, as they will cause many mistakes, confusion and extra work. Only one solution, I feel, that the software must include the category pages at the bottom of normal pages. JMK (talk) 07:55, 31 December 2013 (UTC)
Related discussion: Wikidata:Requests for comment/Commons links#III. Restructure Wikidata items. — Ivan A. Krestinin (talk) 14:31, 1 January 2014 (UTC)
Thanks so much Ivan! That is exactly what I talking about. That extra bit of programming has to be done. And if it is a lot of extra programming, it still has to be done. JMK (talk) 16:05, 1 January 2014 (UTC)

Deleted Wikipedia articles

BlueSpice (Q885586) on English Wikipedia has been deleted, but the link remains in Wikidata after a few days. Isn't there some automation to remove those links? John Vandenberg (talk) 13:09, 30 December 2013 (UTC)

@John Vandenberg: Apparently no. And it's getting even worse. See:
Say hello to wiki(botgarbage)data. --Atlasowa (talk) 12:37, 1 January 2014 (UTC)

Template:Cite doi subpages

What should we do with pages like Q15085936. My rough count is 50 items like this. They are essentially an article (Q191067). However, a journal article can also be notable, which would mean we need two links to Wikipedia : the Template:Cite_doi/ subpage and the mainspace page. Our Fragile Intellect (Q7110639) is an excellent example of that problem, because it has two DOIs. (see Wikidata talk:Periodicals task force#Part publication of a journal article). John Vandenberg (talk) 17:15, 30 December 2013 (UTC)

AFAIK subtemplates are being deleted due to WD:N. At least the pages of de:Kategorie:Vorlage:BibISBN have been deleted, see Q1221443. Imho we should keep these templates and replace them later by WD. (The same way Wikipedia works with Commons.) --Kolja21 (talk) 19:38, 30 December 2013 (UTC)
I don't think template subpages are the best place to host metadata about publications even within one Wikipedia, but once we think beyond a single wiki, we surely need that information somewhere here. The two previous RFCs on the matter do not seem to have reached a conclusion about where to store that information, so I think keeping the items related to those template subpages could be a workable interim solution while something more long-term is being worked out. --Daniel Mietchen (talk) 03:28, 1 January 2014 (UTC)
This may not have been formalised, but we reached a concensus, see Wikidata:Books task force for example. As source are notable per WD:D, this is enough, we do not need a new namespace or anything. The point that might have not been settled is mass import of bibliographic datas, as the synchronisation and data quality is important here. TomT0m (talk) 13:15, 1 January 2014 (UTC) (PS: actually the real missing thing here is the ability to use only one item datas in Wikipedias article, probably. This does not allow to use items for Wikipedia articles bibliography yet) TomT0m (talk) 13:17, 1 January 2014 (UTC)

Reminder: Wikisource sitelinks coming January 14th

Hey folks :)

Just a quick reminder that Wikisource will get sitelinks via Wikidata soon. This is currently still planned for January 14th. Please help out at Wikidata:Wikisource to make sure this deployment goes smoothly.

Cheers --Lydia Pintscher (WMDE) (talk) 19:37, 29 December 2013 (UTC)

What is the state of the proposal? --Rschen7754 19:57, 1 January 2014 (UTC)

Happy New Year 2014

Happy New Year 2014 from France --Jitrixis (talk) 23:09, 31 December 2013 (UTC)

Happy new year! 2014 is the international year of crystallography (Q160398). We should try to work on crystallographic data. --Tobias1984 (talk) 01:37, 1 January 2014 (UTC)
Now from England. Not yes U.S and Canada.--DangSunM (talk) 01:43, 1 January 2014 (UTC)
Peter Murray-Rust (Q908710) has been working on extracting crystallographic data from the scholarly literature and may have some ideas. I pinged him. --Daniel Mietchen (talk) 03:18, 1 January 2014 (UTC)
It would be awesome to get an expert on board. Of course for a lot of data we need numbers first. --Tobias1984 (talk) 16:10, 2 January 2014 (UTC)

Enhancements for Template:Property_documentation

Please see Template_talk:Property_documentation#Parameter_enhancements and chime in with your thoughts. Thanks, Emw (talk) 03:36, 2 January 2014 (UTC)

Wikisource is coming on January 14!

Please read over Wikidata:Wikisource and make sure that we are good to go for the launch! --Rschen7754 07:34, 2 January 2014 (UTC)

Mmmk. Vogone talk 16:14, 2 January 2014 (UTC)

Wonder Women of Natural History: Wikipedia editathon at London Zoo, January 18th

Hi All

I'm organising a Wikipedia editathon at London Zoo on January the 18th, please have a look and come or join in online if you'd like. More info here. I'd really like to have some Wikidata involvement, any ideas would be very welcome.

Thanks

Mrjohncummings (talk) 11:11, 2 January 2014 (UTC)

Vandalism?

Has there been vandalism here? Thanks for checking. 130.88.141.34 14:04, 2 January 2014 (UTC)

Maybe not a vandalism, but just someone who don't know how to do things on Wikidata. Thank you for telling about this, I've fixed it. Though on that item is still some wrong links, some are about the whole game show, some about the British or U.S. version etc. --Stryn (talk) 14:37, 2 January 2014 (UTC)
Thanks for restoring the original version. It may not be perfect, but it is better than what it had been changed to. 130.88.141.34 14:38, 2 January 2014 (UTC)

<wikibase-itemlink>‎

I see everywhere in recent changes and on my watchlist no titles but only hundreds of <wikibase-itemlink>‎. That means no Q... number and no item name is displayed. Does it happen to you as well, or is it wrong only for me? Holger1959 (talk) 18:54, 2 January 2014 (UTC)

I'm getting this too. It also affects Special:NewPages. --Jakob (talk) 18:55, 2 January 2014 (UTC)
Developers are aware of that (per IRC). --Stryn (talk) 18:56, 2 January 2014 (UTC)
Maybe this is connected to the fact that commons: is down? I heard this is an i18n issue. -- Bene* talk 18:58, 2 January 2014 (UTC)
Commons and Wikivoyage are both down; this should be related.--Ymblanter (talk) 19:01, 2 January 2014 (UTC)

Thank you for the replies. Good that the developers already know. Holger1959 (talk) 19:01, 2 January 2014 (UTC)

Wikisource WD:N

Wikisource is coming in January 14th. We need to change WD:N for that. So I propose to change from "It contains at least one valid sitelink to a Wikipedia, Wikivoyage, or Wikimedia Commons page." to "It contains at least one valid sitelink to a Wikipedia, Wikisource, Wikivoyage, or Wikimedia Commons page."--DangSunM (talk) 01:36, 3 January 2014 (UTC)

  • @Rschen7754: So, because the Page namespace is somewhat like file, we can change "To be valid, a link must not be a talk page; MediaWiki page; special page; user page; file; translations page; or subpage of a template, portal, or module." to "To be valid, a link must not be a talk page; MediaWiki page; special page; user page; file; translations page; page namespace in Wikisource; or subpage of a template, portal, or module." (italic text is something in discussion and may be removed)--GZWDer (talk) 05:16, 3 January 2014 (UTC)

  Comment WHat about to write: Must be in namespace 0, 12 or 14; not subpage in namespaces 4, 10 or 100; Must not be in namespaces -1, 1, 2, 3, 5, 6, 7, 8, 9, 11, 13, 15, 101 or something like this. Specific namespaces in different projects are usually in NS 100+, so all other namespaces can be easily added to one of these 3 groups. JAn Dudík (talk) 07:18, 3 January 2014 (UTC)

Yes, but the 100+ namespaces have different purposes in different projects. The namespaces that are unique for Wikisource (Page, Author, Index) have different numbers in different languages. -- Lavallen (talk) 10:29, 3 January 2014 (UTC)
What about if we first create items for Wikisource pages that have at least one sitelink pointing to a Wikipedia page, and then handle the rest, evaluating namespace per namespace? — ΛΧΣ21 17:07, 3 January 2014 (UTC)
I think will be good idea to start from author and disambiguation pages. Content pages will need much more effort (in terms of import and using Wikidata on Wikisource) because interwikis model is completely different. --EugeneZelenko (talk) 15:24, 4 January 2014 (UTC)
  Comment English-language forum is not good place for rule change discussion. — Ivan A. Krestinin (talk) 16:52, 4 January 2014 (UTC)
Start from Wikidata_talk:Wikisource then. I do not think we need to change anything, only try to implement the present ideas on a new project. -- Lavallen (talk) 17:29, 4 January 2014 (UTC)

  CommentThere is a need for a greater consideration. The criteria needs to be focused on the principal of what the WSes are trying to present, and a means for clear concise statements in the policy. Page: and Index: namespace pages are not presentation pages, those ns are solely work spaces. Another issue is WS works, and subpages (chapters) to works, which can be interlanguage linked for side by side comparison and those individual subpage should NOT link to/from WD. I believe that this should be worked out elsewhere, and brought back to the community.  — billinghurst sDrewth 00:28, 5 January 2014 (UTC)

I think this is something in need of an RfC rather than just a project chat page.--Jasper Deng (talk) 02:13, 5 January 2014 (UTC)

Well, what do we do now, with deployment a week away? --Rschen7754 07:39, 7 January 2014 (UTC)

Try to focus more on Common sense than an outdated notability-policy. Wikidata:Wikisource and the Book task force give us enough guidance at the moment. The only thing I think we have to consider is the subpages in ns:0. I think they should be added to be able to add metadata and to support templates in those pages. But there is no hurry, we will have enough to do until we have a consensus in that matter. -- Lavallen (talk) 08:24, 7 January 2014 (UTC)
I have sent out a MassMessage to draw more attention to the issue, and restart the discussion in a more language-neutral forum. --Rschen7754 08:47, 7 January 2014 (UTC)

  Comment To me the WD:N needs to more incorporate what the respective wikis express as their presentation namespaces, and that should be presented as a schedule to the policy. It is a lot easier to update a schedule as you add sister wikis, rather than generate a really complex set of notability. The only thing outside of that is whether you also wish to collect project and help pages from these wikis too.  — billinghurst sDrewth 09:50, 7 January 2014 (UTC)

So in terms of the policy, that might make at least one valid sitelink from a presentation namespace ... (wiki list) (or think of another word than presentation, but please not content which has other meaning.  — billinghurst sDrewth 09:54, 7 January 2014 (UTC)

Seasons of European soccer

Hoi, The golden shoe is awarded to the best soccer player in the European national competitions. As far as I am aware there are no articles for the soccer seasons. They are the obvious qualifiers for awards like this. Are these series on items we want to create?? Thanks, GerardM (talk) 14:21, 5 January 2014 (UTC)

One other application is to include it as a qualifier for when a soccer player played for a team (like here. GerardM (talk) 14:31, 5 January 2014 (UTC)
Many sports have seasons from autumn to spring (or from summer to summer), so perhaps you should make items (or a property?) for sporting seasons in general. Bever (talk) 20:23, 6 January 2014 (UTC)

Logo entries mixed up

Q3294857 is mixing pages which correspond to respectively en:Wikipedia:Logos (usage of logos in general) and en:Wikipedia:Wikipedia logos (Wikipedia's own logos). I have limited language skills but looking at the images on a page usually makes it easy to guess which it is. Can somebody sort it out? Is there an easy way to split the entries without editing them one at a time? PrimeHunter (talk) 14:41, 5 January 2014 (UTC)

We have a "Move" gadget which can substantially ease the pain of moving IWs around. Have a look to see if that will help. --Izno (talk) 15:33, 5 January 2014 (UTC)
Thanks. It did help some. I have created Q15555389 for Logo usage in Wikipedia. Resolved. PrimeHunter (talk) 17:28, 5 January 2014 (UTC)

SubProperty

I just created a bug asking for a subproperty mechanism. [1].

Can anyone interested in this go over there and check it out and support if they agree. Filceolaire (talk) 14:53, 5 January 2014 (UTC)

What is going wrong?

Hi, I can't login, whatever I try. Wikidata won't accept my passwords, and even trying to request a new password goes wrong when I try to replace WD's temporary password with a new permanent one. Am I the only one with this problem? It's pretty frustrating. 94.209.26.76 16:32, 5 January 2014 (UTC) (w:User:Steinbach)

Okay, maybe it was just me who touched the wrong keys a few successive times. But this never happened to me before, so I started to believe it was the system. Steinbach (talk) 16:42, 5 January 2014 (UTC)
Edited and marking as resolved.
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --Tobias1984 (talk) 18:30, 5 January 2014 (UTC)
@Tobias1984: Section resolved template does not work: because {{Autoarchive resolved section}} is not on the top of this page, so SpBot (talk • contribs • logs) cannot understand where to archive. --by Revi레비 at 08:54, 7 January 2014 (UTC)

Property proposals

Can a few people head over to the property proposal pages. Voting has gotten really slow. --Tobias1984 (talk) 19:51, 5 January 2014 (UTC)

Looks like procedure change is needed. There is no oppose comments during week - create property. Somebody want to leave oppose comment after week - welcome to property deletion page. — Ivan A. Krestinin (talk) 20:30, 5 January 2014 (UTC)
I think @Emw: and some other people have recently stated that a thorough review of proposals is a good idea. I also think we should stick to the current system, but try to motivate more people to participate in the discussions. In my opinion the pages are not visible enough. In the left side menu we have a link for "Create a new item". Why not another link "propose a new property". That would certainly draw more people to the proposal pages. --Tobias1984 (talk) 22:11, 5 January 2014 (UTC)

Unresponsive script

Hi. I tried to add a link to wikidatapage Q421219 for the finnish article fi:tioredoksiini but I get an error message that says

Warning: Unresponsive script
A script on this page may be busy, or it may have stopped responding. You can stop the script now, or you can continue to see if the script will complete.
script:https://bits.wikimedia.org/www.wikidata.org/load.php?debug=false&lang=en&modules=jquery%2Cmediawiki%2CSpinner%7Cjquery.triggerQueueCallback%2CloadingSpinner%2CmwEmbedUtil%7Cmw.MwEmbedSupport&only=scripts&skin=vector&version=20140104T034256Z:9

and I can't add the link. No other wikidata page gives that error for me and I can edit them normally. My browser is Firefox 26.0. --MiPe (talk) 11:32, 6 January 2014 (UTC)

I added the link now, but I also had to click away the "Unresponive script" warning twice when loading the item. --YMS (talk) 11:45, 6 January 2014 (UTC)
Happens to me also. I guess it's because there is so many properties on the item. The same error I get also e.g. at Q40030. This is very frustrating, because it also freezes the browser. --Stryn (talk) 12:29, 6 January 2014 (UTC)
It's high priority for me to get this situation improved. The bug for tracking it is bugzilla:54098. --Lydia Pintscher (WMDE) (talk) 15:29, 6 January 2014 (UTC)

Is MediaWiki talk supported?

I'm trying to convert en:MediaWiki talk:Spam-whitelist#Other projects with active whitelists to Wikidata. What I've tried here doesn't seem to be working. Is this supported, and if so, how do I get the Languages links to show on the left? Thanks, Wbm1058 (talk) 00:29, 7 January 2014 (UTC)

Uh, interface messages are not eligible to be included on Wikidata, per WD:N. Talk namespace pages of any kind are also not eligible.--Jasper Deng (talk) 01:39, 7 January 2014 (UTC)
Thanks. Is it possible to get those links into the left margin using any older, pre-Wikidata technique? Wbm1058 (talk) 01:53, 7 January 2014 (UTC)
Also, I'm a little puzzled then that a page like en:Wikipedia:Requested moves is supported by Wikidata, as an internal project page hardly seems notable outside of the project. Wbm1058 (talk) 02:00, 7 January 2014 (UTC)
If you can find good arguments, it's possible to change WD:N. But I doubt that adding interwiki to a page who looks the same in every MediaWiki-project is an argument good enough today. -- Lavallen (talk) 07:27, 7 January 2014 (UTC)

Wikisource is coming in 1 week!

Apologies for using English; please help by translating this message!

Wikisource is coming in 1 week, on Tuesday, January 14! Help us prepare for the deployment at Wikidata:Wikisource. Specifically, we need your input in determining how Wikidata items will be created and used for each Wikisource page. Please join the discussion at Wikidata talk:Wikisource, and notify anyone who is interested in the discussion. --Rschen7754 08:44, 7 January 2014 (UTC) (using MassMessage)

Search page takes forever to load

If I click Special:Search, the page takes more than a minute to load, and I suppose that happens to you too? If I enter a search term and click the Search button, the results show up in a timely manner. I first noticed this behavior yesterday. I just played with the search parameter in the querystring for a few seconds: If it is empty, the page is slow. A known problem? - Soulkeeper (talk) 11:42, 7 January 2014 (UTC)

Yesterday the new search engine became the default. It may have to catch up a bit.. fill its caches .. We will know in a few days if this persists. Thanks, GerardM (talk) 12:19, 7 January 2014 (UTC)
Until now I experience a lot of searches yielding no search results at all, while there are items containing the search term. I hope this improves soon. Lymantria (talk) 14:03, 7 January 2014 (UTC)
I still don't understand why the page is slower with no search string and no results, than it is with a search string and results. But we will wait and see. - Soulkeeper (talk) 17:08, 7 January 2014 (UTC)
I cannot imagine that this is an intended behaviour. Perhaps notify at Wikidata:Contact the development team. -- Bene* talk 20:25, 7 January 2014 (UTC)
I've reported it at bugzilla:59821 and hope someone from the search team will have a look soon. --Lydia Pintscher (WMDE) (talk) 14:53, 8 January 2014 (UTC)

Workshop on mark-up of biodiversity literature

Hi, I am organizing a workshop on February 10-11 at Natural History Museum, Berlin (Q233098), where both semantic search and crowd-based approaches to markup will be amongst the topics to be discussed. If anyone here with an interest in biological taxonomy would like to participate, please let me know. --Daniel Mietchen (talk) 14:33, 7 January 2014 (UTC)

Obsolescence

Can I reflect my frustration about how some properties all of a sudden appear to be OBSOLETE, seemingly out of the blue, and without clear conversation. As a quiet worker in this place, who is busy in other places around the WMF, who is spending time working through violation reports, to find that something to which you have been working, is now obsolete is a kick in the (name your body part). Your RFD and obsolence processes seem to have an element of "wham! bam! thank you maam." If you want to know about obsolescence, the effort that I have been putting in to these reports definitely feels like it falls into that space, and makes me reflect that should I be doing it.

While I know that this community is youthful and still developing, there needs to be consideration to the people, outside of the technical. I understand that is probably a comment from someone unyoung talking to the young, though I hope that this can be seen in the perspective of the wise talking to the intelligent. You need to learn to communicate and to consider people, and their time frames.  — billinghurst sDrewth 01:57, 8 January 2014 (UTC)

I share your frustration with Properties for deletion. Currently, there is no requirement for relevant people to be noted, and there are too many editors doing knee-jerk nominations of properties. Something needs to be fixed. --Rschen7754 02:08, 8 January 2014 (UTC)
On the topic of fixing PFD, there should be a bot to remove links to properties that don't exist. Currently I don't like deleting properties because I manually have to remove hundreds of links. This would eliminate marking properties as obsolete also. --Jakob (talk) 02:50, 8 January 2014 (UTC)
Yes, but that doesn't resolve the issue at hand. --Rschen7754 03:26, 8 January 2014 (UTC)
Please don't forget that sometimes it is not enough just to remove links, you should also to change deleted property to more relevant one. --Infovarius (talk) 14:33, 9 January 2014 (UTC)
There was consensus about half a year ago at this thread for notifying relevant users and task forces when a property was nominated for deletion. I noted this consensus about two months ago here, and I only got one reply, which I didn't (and still don't) really understand. I'm not sure if the user was telling me to inform all users and task forces of all the discussions at PFD or what, but he just didn't want a bot notifying people, which I didn't even imply. TCN7JM 03:37, 8 January 2014 (UTC)
Some users could have understood your proposal like you have to add a new thread on a number of user_talks, instead of in some relevant projectpages. Even WMF and the Developers love to spam user_talks sometimes, and I do not like to see that habit spread to this project more than it already have. -- Lavallen (talk) 09:38, 8 January 2014 (UTC)
Yeah, but I didn't imply running a spambot. I just thought notifying relevant users and projects of deletion discussions was common courtesy. TCN7JM 11:12, 8 January 2014 (UTC)
I was never worried about how you should do, I was worried about how Mr and Mrs Average_User would respond to your proposal. What I saw, was that it could be interpreted as "spam is allowed when you have a good cause", even if it wasn't your intension. -- Lavallen (talk)
I get that sv.wikipedia hates these sorts of "spambots", but pushing those views on a neutral and multilingual project used by all Wikipedias (and Wikimedia sites, really) is not appropriate. --Rschen7754 08:35, 9 January 2014 (UTC)
We have something called "Netiquette". And with "We", I do not refer to the Swedishspeaking community within the wmf-projects, I refer to Internet. If we are supposed to give up "Netiquette", then write it out, but I'm afraid it will hurt the project. -- Lavallen (talk) 09:25, 9 January 2014 (UTC)
I am puzzled about why you insist that it would be in keeping with "netiquette" not to "spam" previous or interested contributors as well as potentially interested Wikidata talk space pages with a indication that a discussion is ongoing that is relevant to them. The reason we're even having this discussion is that people are not being communicated to. If you personally don't want to receive a notification, that's something we can fix. But the rest of us probably want one. And don't presume to think every one keeps the same notion of "netiquette" as you do. --Izno (talk) 14:48, 9 January 2014 (UTC)

It would help if

  • for properties under discussion that the front page had the identification of the discussion, not rely on it being found on a talk page. It is wiki standard to utilise the front, not the back of a page
  • links that are on talk pages that point to discussions, actually pointed to discussion, so a good linking process to a discussion would be more resilient
  • if there is a nomination of a property, then it would be useful to nullify/suspend any violation criteria and report that is in place. This is so that people do not continue to work on the "supposed" violation fixing things that are potentially going to waste.

all these things are way without any notification to people, they are evident to those going past.  — billinghurst sDrewth 04:25, 8 January 2014 (UTC)

I suppose that deleting of property worths at least pinging users who used it (can be gathered by "what's link here"). Infovarius (talk) 14:33, 9 January 2014 (UTC)

RFD archives 2014

Can someone please spend 5-10 minutes fixing up the archives for 2014 so that they all link up nicely. Seems to have escaped notice in the new year. Thanks.  — billinghurst sDrewth 02:02, 8 January 2014 (UTC)

Fixed. It always requires a redirect on every month like Wikidata:Requests for deletions/Archive/2014/01. --Stryn (talk) 08:32, 8 January 2014 (UTC)

Lime

When someone searches for the word 'lime', are they looking for the fruit or the color? Do we need the same aliases for them? TeleComNasSprVen (talk) 10:13, 8 January 2014 (UTC)

Help:Label explains how Wikidata handles labels for items. The description is used to disambiguate items with the same label. 130.88.99.229 11:06, 8 January 2014 (UTC)

Wikisource

Hello all. We really need an RFC about Wikisource, beside the notability thing, there is another concern that I'm worried about. s:Author:Hafez this page should be linked to Hafez (Q6240) or we need to make a new item, and what about his poems for example this is one of his poems but we really need to create an item for one-lingual content? I think almost everything on Wikisource is one-lingual Amir (talk) 10:18, 8 January 2014 (UTC)

There is discussion going on at Wikidata talk:Wikisource. My understanding though is that if they are the same person, they should be linked on the same item, though. --Rschen7754 10:21, 8 January 2014 (UTC)
another thing: s:Category:1390 deaths needs to be added to this Category:1390 deaths (Q6724975) or another item? Amir (talk) 10:26, 8 January 2014 (UTC)
Again, I don't see why those can't use the same item. --Rschen7754 10:27, 8 January 2014 (UTC)

redirects

The project fails in handling redirects. As an example, the german 'Spitzendiode' is the same as the dutch 'puntcontactdiode'. But the english item is 'point contact diode', which is not an article of its own, but redirected to the general article 'diode'. It should be made possible to link 'Spitzendiode' to the redirect 'point contact diode', in order to provide the right information. Nijdam (talk) 12:27, 8 January 2014 (UTC)

I just added 'point contact diode' as an English-language label to point-contact diode (Q15613408). --Daniel Mietchen (talk) 12:38, 8 January 2014 (UTC)
WD item links to WP REDIRECTs were discussed for a long time : Wikidata:Project_chat/Archive/2013/02#Linking_to_redirect_pages, Wikidata:Project_chat/Archive/2013/02#Interwiki_links_with_anchors, Wikidata:Project_chat/Archive/2013/08#How do we merge 2 items? and many more. I don't think there ever was a commitment of dev team to support that. The current workaround (stated in Wikidata:Project_chat/Archive/2013/07#Interwiki links) is: "edit [the WP redirect page] to make it an ordinary page, add the page to [WD item], and then immediately make it a redirect again". - LaddΩ chat ;) 23:03, 8 January 2014 (UTC)

Wrong precision for coordinates

Please see Property talk:P625#Wrong precision. --JulesWinnfield-hu (talk) 16:12, 8 January 2014 (UTC)

MediaWiki:Gadget-HotCat

Can we exempt this from notability considering how widespread it is? Very useful for cross-project wikignomish categorization. TeleComNasSprVen (talk) 11:31, 27 December 2013 (UTC)

Exempt or include? Right now, pages in the MediaWiki namespace are already excluded per Wikidata:Notability. The Anonymouse [talk] 16:55, 27 December 2013 (UTC)
Yes I know, but I believe this one has cross-wiki significance to Wikimedia projects. TeleComNasSprVen (talk) 21:42, 27 December 2013 (UTC)
To clarify if you don't understand the "but..." part, it is to include in this project, and exempt from the stringent criteria. TeleComNasSprVen (talk) 21:43, 27 December 2013 (UTC)
OK, now I understand. I have no opinion either way whether that page should have links. The Anonymouse [talk] 16:35, 2 January 2014 (UTC)
Why we should include it on Wikidata, even if it's widespread? It's like every other pages on MediaWiki namespace. --Stryn (talk) 17:45, 9 January 2014 (UTC)
For fast interproject travel? I've used this many times. TeleComNasSprVen (talk) 17:51, 9 January 2014 (UTC)

I created the links for it, and now it's deleted again (ref Q15544703, Q15544664, Q15544637). User:Delusion23, could you please participate in this discussion and explain your thoughts on this matter? TeleComNasSprVen (talk) 17:27, 9 January 2014 (UTC)

There is nothing to be said. The deletion is within the notability policy therefore the only dispute it with the policy and not the admin whether the admin agrees or disagrees with the notability policy. John F. Lewis (talk) 17:50, 9 January 2014 (UTC)
The notability says: "When in doubt, discuss your idea on the project chat." That is what I have been trying to do, but nobody seems interested. Speaking of which, why aren't there sitelinks for Special:Watchlist and Special:RecentChanges? Those are the most heavily trafficked pages out of all the special pages. TeleComNasSprVen (talk) 17:55, 9 January 2014 (UTC)
There is for Q6293548. John F. Lewis (talk) 17:58, 9 January 2014 (UTC)

Suggester

Hey :) A group of students is working on suggestions to make it easier to add missing information to items. Here's their first status report with a demo: http://lists.wikimedia.org/pipermail/wikidata-l/2014-January/003301.html --Lydia Pintscher (WMDE) (talk) 18:29, 9 January 2014 (UTC)

Tried it out. Really nice tool. Can the students turn it into a gadget that suggests these properties below the last statement when viewing a Q-item? --Tobias1984 (talk) 20:18, 9 January 2014 (UTC)
Yeah it will be integrated in the UI properly. This first demo is basically to demonstrate that their suggestions work :) Integration with the UI comes next. --Lydia Pintscher (WMDE) (talk) 20:21, 9 January 2014 (UTC)

I don't understand how this tool works. Suggestions are displayed like this on Firefox and Safari on Mac OS 10.9.1 :

{"name":"Geschlecht","id":"21","correlation":"0.8133400082588196"}
{"name":"Staatsangehörigkeit","id":"27","correlation":"0.508866012096405"}
...

— Ayack (talk) 21:19, 9 January 2014 (UTC)

Yes. It's just the beginning of the functionality. What you're seeing there is probably this: You chose a property. (I don't know which one.) It then thinks that Geschlecht and Staatsangehörigkeit are very likely to be properties this item should also have. The number at the end says how strong this correlation is. High means it is very likely that the same property should be used.
In its current form it isn't too useful yet for users. But I wanted them to show you this first demo so you can see we are making progress on suggestions. Once the project is done you will for example be able to go to an item and it will show you some stuff that is likely missing on that item so you can add it. --Lydia Pintscher (WMDE) (talk) 21:24, 9 January 2014 (UTC)

Wikisource now enabled on test.wikidata.org

Hey :) Just wanted to let you know that Wikisource is now enabled on test.wikidata.org. Nothing spectacular but some people asked. You can add sitelinks to Wikisource. --Lydia Pintscher (WMDE) (talk) 20:24, 9 January 2014 (UTC)

Interwikis (again?)

I'm sorry if this is the n-th time you guys have to answer this, but I could not find anything related to this in the FAQ, so here I am. This page at ptwiki showed an interwiki to enwiki that has been deleted in 01/07/2013, more than a year ago ([2]), up until 01/04/2014). Is this the normal behaviour or something wrong was going on at the time immediately before the bot actually did what it was supposed to? Jbribeiro1 (talk) 21:39, 9 January 2014 (UTC)

The bots which have been tracking deletions and moves can be lazy sometimes. We kick them whenever we get the chance, and sometimes they do work. :) --Izno (talk) 23:19, 9 January 2014 (UTC)

Aliases for years

I've just added the aliases ""2004 CE" "2004 AD" and "MMIV" for Q2014, the year 2004. Does someone have bot that could do the same for every other year? What about alternative ways to write them, say "2004AD", "2004 A.D.", "AD 2004", "2004 Anno Domini", etc? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:57, 4 January 2014 (UTC)

Do you expect someone is searching for "MMIV" (=2004)? --Succu (talk) 21:07, 4 January 2014 (UTC)
Why not? It's used in movie credits, on buildings, etc --YMS (talk) 21:35, 4 January 2014 (UTC)
But if we add a numeric property to each year, that says something like "year = 2004", then we can compute the Roman Numerals and don't need to add them as aliases? --Tobias1984 (talk) 07:52, 5 January 2014 (UTC)
We need to add them as aliases in case someone searches for them. Filceolaire (talk) 13:32, 5 January 2014 (UTC)
I think our search engine should be able to handle this at some point. In any case this sounds like a good bot task. --Tobias1984 (talk) 12:25, 7 January 2014 (UTC)
I would vote for adding Roman numerals as aliases. In Russian. --Infovarius (talk) 19:48, 7 January 2014 (UTC)
This does not directly affects me, since I'm not an en-N, but remember that there are two ways to write 2004 as a Roman number, MMIV and MMIIII. -- Lavallen (talk) 20:17, 7 January 2014 (UTC)
As I just learned, that second variant is the original one, but it became uncommon in medieval ages. Is it used anywhere today in some prominent way? Because it quite definitely doesn't make too much sense to convert the years in every numeric system possible. Nobody writes years in Babylonian or in hexadecimal numbers, nobody searches for those, and thus we don't need them as labels. As I wrote, I think this is different with the Roman numerals, but - as far as I'm aware of - only for the subtractive notation. --YMS (talk) 17:11, 10 January 2014 (UTC)
I have seen the "old system" in the Page numbers in some books and also in some watches, but as I said, I do not have an Englishspeaking pov. - Lavallen (talk) 17:25, 10 January 2014 (UTC)

Cirrus search is now default

Hey :)

Cirrus search is now default. This should considerably improve the search situation on Special:Search for example for special characters. It might still need some tweaking though. It'd be awesome if you could give the search a try and collect things here that still don't work as they should. --Lydia Pintscher (WMDE) (talk) 09:38, 10 January 2014 (UTC)

When will the search on the right top corner use Cirrus search? It (still) doesn't give any suggestions when starting to type something what is not in the item namespace. --Stryn (talk) 14:42, 10 January 2014 (UTC)
That's an independent issue. We have bugzilla:46251 for that. It is unfortunately not trivial to fix. However with the students working on the entity suggester some changes are needed that could make it easier. It is annoying me too and I want it fixed. (As usual go and vote for bugs that are important to you please.) --Lydia Pintscher (WMDE) (talk) 18:33, 10 January 2014 (UTC)

Italian flag and CoA templates

I've been working on the templates found in the categories linked in Category:Italy regional flag templates (Q9173335). German and Luxemburgisch, Occitan templates and and most of those in Lumbart, Ligurian, Napolitan and Ventetian were not linked. I found that even though all categories seem to claim (by name) that the templates use flags, a lot of them use coats of arms in stead. Some, including all Corsican, Italian, Tarandíne, Romenian and Ukrainian use CoA's exclusively. The Breton, French, German, Japanese, Liguian and Luxemburgisch use flags (almost) exclusively. The rest use one or the other, depending on the region. I found no cases where both existed on the same local wiki. As the local situations may change, when local users change some templates from flags to CoA or vice versa, this would need to be monitored constantly. Alternatively, as all foremensioned templates are used for the same purposes and as no conflicts seem to exist, we could merge all items, keeping one per region, and let the local communities decide if they want to use flags or coats of arms. -      - (Cycn/talk) 13:00, 10 January 2014 (UTC)

100000000

In a few hours Wikidata will log the 100 millionth edit (Special:Statistics). After the first birthday of the project another reason to celebrate. Nothing left to do but to lean back and watch the flow of information [3]. --Tobias1984 (talk) 13:55, 30 December 2013 (UTC)

I wonder who will make the 100000000th edit. Probably a bot. --Jakob (talk) 14:33, 30 December 2013 (UTC)
Less than 1000 edits to go now. Any minute... --Jakob (talk) 15:34, 30 December 2013 (UTC)
It's happened! We have 2,136,500,447 edits. --Jakob (talk) 15:38, 30 December 2013 (UTC)
I don't know why the numbers on that special page are so high, but they're probably incorrect. We're actually probably still at about 97 Million. - Hoo man (talk) 15:41, 30 December 2013 (UTC)
MariaDB [wikidatawiki_p]> SELECT c1 +c2 AS edit_count FROM ( SELECT COUNT(rev_id ) AS c1 FROM revision) AS rev, (SELECT COUNT(ar_id ) AS c2 FROM archive) AS arch;
+------------+
| edit_count |
+------------+
|   96992583 |
+------------+
1 row in set (29.85 sec)
Yep, diff?100000000 does not exist yet. --Stryn (talk) 15:43, 30 December 2013 (UTC)
Does your SQL query consider edits to deleted pages? --Denny (talk) 18:19, 30 December 2013 (UTC)
Yes it does, those are stored in the archive table which is also being counted in the query. Cheers, Hoo man (talk) 18:45, 30 December 2013 (UTC)
Anyway, when building the Croatian Wikipedia I was always making a conscious effort to let round numbers be done by humans, not by bots. Anyone thinks it is a good idea to ask for a bot moratorium once we get near diff 100M? --Denny (talk) 18:23, 30 December 2013 (UTC)
Sounds like a nice idea, having a bot do edit #100M sounds a bit boring - Hoo man (talk) 18:45, 30 December 2013 (UTC)
Pause 75 bots, some of them really highly active, for an unknown amount of time, just to have one diff be dine by a human, though likely one workig with script assist anyway? --YMS (talk) 09:32, 31 December 2013 (UTC)
Yeah, that would be the suggestion :) I guess there is not much excitement about the idea, though. --Denny (talk) 01:40, 1 January 2014 (UTC)
currently 97506181 edits. not yet--DangSunM (talk) 01:42, 1 January 2014 (UTC)
@Tobias1984, King jakob c 2: Very accurate is to watch Special:Log/patrol, only edits by non-autopatrolled users are not shown. Matěj Suchánek (talk) 10:56, 3 January 2014 (UTC)
At 99.1M now. Can we have a decision if we *want* 100M to be a human or if we don't care and just let it happen? There is still enough time either way. --Denny (talk) 23:54, 6 January 2014 (UTC)
I think it's reasonable but if no one wants it, I won't push in favor of it. BTW, it'll be about 3 days, 9 hours, and 3 minutes until 100000000 edits at the current rate. --Jakob (talk) 00:34, 7 January 2014 (UTC)
  • I don't have an opinion either way if the bots should be stopped or not. However, to stop the bots, an edit filter could be created temporarily to disallow all edits from users in the 'bot' user group. This is just an idea, I don't know if this would work or be practical. The Anonymouse [talk] 16:56, 9 January 2014 (UTC)
  • Lets just celebrate the diff and leave the bots running. Everybody on this project likes watching it grow and it doesn't matter who makes the edit. And who knows - maybe some human will beat the odds. --Tobias1984 (talk) 19:50, 9 January 2014 (UTC)
Looks like consensus to let the bots run. It's at 99.92M. Well then, let's see if a human can make it :) --Denny (talk) 17:40, 10 January 2014 (UTC)

You did. It was close :D It was a human who made this edit. Congratulations, User:Sjoerddebruin! --Denny (talk) 00:25, 11 January 2014 (UTC)

Crap, I missed it. --Jakob (talk) 01:15, 11 January 2014 (UTC)

Pause the bots

  1. I vote for having a human make the 100 millionth edit. Emw (talk) 00:38, 7 January 2014 (UTC)
  2. Even though I agree with the arguments for letting the bots continue, I still think it would be nice to have a human make the 100M edit. --Denny (talk) 18:22, 7 January 2014 (UTC)

Don't stop the bots

  1. Why is this a choice between man and machine? My only concern is that the bots don't stop, because I want all of the delicious data. Maybe that's a vote for machine, but that's not how I see it. I'm not getting any younger! --Izno (talk) 00:41, 7 January 2014 (UTC)
  2. Would much prefer that bots continue to feed info to the hungry Wikidata :D Ajraddatz (Talk) 01:10, 7 January 2014 (UTC)
  3. Why is this even being considered?--Jasper Deng (talk) 01:14, 7 January 2014 (UTC)
  4. (edit conflict) Our goal is to improve the database, and that preferably as efficient as possible. Having a human to make that edit (and stop the accounts which edit the most) doesn't help that much for this. Vogone talk 01:16, 7 January 2014 (UTC)
  5. I change my mind. Stopping the bots for a few minutes might be OK, but not every bot owner can stop their bot a few minutes before the 100,000,000th edit. Also, this wouldn't even work unless every bot owner agreed to it. --Jakob (talk) 01:35, 7 January 2014 (UTC)
  6. The truth is: bots rule this wiki. --Succu (talk) 19:57, 9 January 2014 (UTC)

About RfD

I have seen that Requests for Deletion is usually been filled up with a big amount of uncontroversial deletions that sometimes even reach the houndreds of items. Although this is not bad, I think that we should separate procedural deletions (such as when an uncontroversial merge happens, or when we have items that are obviously outside of our scope like spam or empty items, or items about userpages) from deletions that could actually help from discussion. Maybe we could refactor RfD into two different pages, one for procedural deletions, and other for requests that need actual discussion.

I was thinking about moving the latter to a new page like Wikidata:Items for Discussion or something, where we could put items not only for outright deletion, but also discuss possible merges (if applicable), and so on, and then leave Wikidata:Requests for Deletion for the procedural ones. Another thought was to move procedural deletions to Wikidata:Requests for Speedy Deletion (since most of these requests are handled within 30 minutes) and then leave the original RfD page to work just like Properties for Deletion does. Any thoughts? — ΛΧΣ21 23:35, 2 January 2014 (UTC)

I say; grab a steward. Delete WD:RfD, make deletion requests on the talk page using {{Delete}}. For discussions, also use {{Delete}} on the talk page. Perhaps make them more separate or so. But, WD:RFD is a poor system in my opinion even though it is used on a lot of projects. John F. Lewis (talk) 23:45, 2 January 2014 (UTC)
We'd still need some kind of overview page then, preferably not only a plain category as that seems rather uncomfortable to me. Vogone talk 00:03, 3 January 2014 (UTC
The reason we don't, and shouldn't do anything related to the talk page is that it just makes an extra page to cleanup. Good luck getting that to make sense to the admins who actually deal with RFD.... --Izno (talk)
Actually deals with RfD? Who is that then? Because I am one of them when I have time. John F. Lewis (talk) 02:32, 3 January 2014 (UTC)
You can't seriously tell me that you working on RFD would rather delete 2 pages than one, given the massive number of pages we need to deal with? /boggle --Izno (talk) 02:35, 3 January 2014 (UTC)
(edit conflict) One could also discuss about these items on the corresponding talk page, instead of creating a further, probably most of the time empty page as there is really rarely need to discuss. Vogone talk 23:48, 2 January 2014 (UTC)
Corresponding talk page...? --Izno (talk) 02:35, 3 January 2014 (UTC)
Every item has a talk page and in cases where users think an item and its sensefulness needs to be discussed (rather than speedy deleted) they could also use the talk page of the item to explain their point of view. This doesn't need any extra page in the Wikidata namespace as proposed by Hahc21 above, in my opinion. Vogone talk 04:24, 3 January 2014 (UTC)

By the way, I still wonder why we delete empty items at all. Isn't that just a big waste of time? I mean, empty items have no influence to any site. Vogone talk 00:08, 3 January 2014 (UTC)

It makes us feel cleaner. Given that the community at some point decided not to reuse item IDs, which is really a silly thing (though I can understand the permanent identifier attraction)... --Izno (talk) 02:29, 3 January 2014 (UTC)
The argument of not re-using items makes sense, and is a good enough reason. It would be nice to further streamline the process though. Ajraddatz (Talk) 02:35, 3 January 2014 (UTC)
I wasn't talking about reusing these items, rather about not touching them after they got emptied. They do not cause any harm and I do not see the big difference between an item with no content and a box saying the item in question was deleted. Vogone talk 04:32, 3 January 2014 (UTC)
  • (edit conflict) Hmmm, I don't like the idea of using {{Delete}} on talk pages. It is not a visible solution. I would like to know, what are your thoughts about creating a second Wikidata-namespace page for potentially controversial item deletion requests. I'm not trying to get rid of RfD as it is; I'm trying to make sure that we have a place to properly discuss controversial item deletion requests, which are usually handled at RfD but pretty poorly because of the big amount of uncontroversial deletions. Another way might be to move all these big loads of RfDs to a category, and then leave RfD only for the controversial ones. — ΛΧΣ21 04:34, 3 January 2014 (UTC)
    I agree. Creating talk pages for marking some page for deletion is not a good idea IMO. It would be good to have a second page for those deletion requests which needs some discussion. --Stryn (talk) 15:47, 3 January 2014 (UTC)

How about a "To be deleted"-property? -- Lavallen (talk) 06:20, 3 January 2014 (UTC)

The devs have promised us a redirect mechanism real soon. Once we have this merges will no longer result in a delete - the unused item will get a redirect to the retained item so that Qids are persistent. This will get rid of most of the deletes. Lets wait till we have this before we change anything. Filceolaire (talk) 11:06, 3 January 2014 (UTC)

+1 Ajraddatz (Talk) 01:55, 11 January 2014 (UTC)

Merge?

neighborhood (Q123705) and Ortsteil (Q253019) --Rippitippi (talk) 18:05, 9 January 2014 (UTC)

No. The first one can be inside of the latter one. --Stryn (talk) 18:28, 9 January 2014 (UTC)
My last analysis about these two items on a call by Infovarius:
"There is an object for the disambiguation pages Q1437134, one for the subdivisions within cities Q123705, and one for the subdivisiosn of other places (in German: 'Orte, Gemeinden') Q253019. The question what is a 'city' and what not, is not in every country handled equally. However, the English page en:Quarter (urban subdivision) looks for me like the it speacks about a subdivision of cities AND other places, i.e. this would be the union of both indivdual objects. Are there other languages which looks similar to that? Then, one could create a new object linking to the English page. But the Russian page should stay linked to the disambiguation object Q1437134, and ru:Городской_квартал does not yet exists." see Wikidata:Interwiki_conflicts/Unresolved/2013#quartier --Zuphilip (talk) 21:42, 9 January 2014 (UTC)
I think there is an big big confiuson in that two item, wikilink,category ecc. --Rippitippi (talk) 23:14, 9 January 2014 (UTC)
I tend to say no. For me, a "quarter" is close to and, often, part of a major human settlement; while a neighbourhood could be anywhere (often at crossroads, bridges &c, but not necessarily tied to a much bigger settlement). --80.114.178.7 23:10, 10 January 2014 (UTC)

Need a broader title property, is there something?

For positions which change name over time, I looking for a property akin to P357 (P357). Something that is text only, that doesn't link, but to which I can qualify with a start and end date. Sort of item that I would look to apply this is the religious bishops that can be bishops, then later get a version upgrade to archbishop. Thanks.  — billinghurst sDrewth 01:33, 10 January 2014 (UTC)

@billinghurst: Perhaps position held (P39)? --Jakob (talk) 02:14, 10 January 2014 (UTC)
@King jakob c 2:, that is what the person has, and what is at the end of the link. On the page to where position held links is where I need to put the alternative names for the position, eg, Bishop of ..., Archbishop of ...; all of which we use as aliases, but these are the formal titles just change over the years. It is not (preceded|succeeded) because it would be a self-reference as the entity does not change, it is the running name. Businesses do it with name changes, etc. so it is broader than a person, it is objects, administrative bodies, things that have formal names/references.  — billinghurst sDrewth 03:13, 10 January 2014 (UTC)
Hell, I would use P357 (P357) except it is very specifically identified as being the "title of a work" when you dig into it. Now if that was broadened in scope, it would be ideal, the issue is that I don't understand the consequences of such a change.  — billinghurst sDrewth 03:15, 10 January 2014 (UTC)
I wouldn't recommend broadening the scope of the property. "Title" as in name of works and "title" as in position of a person may share the same word in English, but they are very distinct entities. FYI, the property is being used mainly by written works and edition items, and will be used eventually for sources. With the phase 1 of Wikisource rolling out next week, the use of P357 (P357) will increase significantly. --Wylve (talk) 03:40, 10 January 2014 (UTC)
Yes and no. I would argue that if someone's official title is "Bishop of New Zealand" and it is an English title, that is what I am looking to have against what is now "Bishop of Auckland" which is where the successor role became. Trying to configure that for every language is nonsensical to me when it already appears in the actual position name.  — billinghurst sDrewth
It sounds like there should be a more integral change in the software to address this problem, since using a property to show the transformation of names will not change the label of the item. For example, ABC company changes its name to DEF company in 2010. An employee who worked at ABC company between 2006 to 2007 would have employer (P108):DEF company shown on their item. This doesn't make sense, because the employee never worked for a DEF company. I'm not sure if it's technically possible, or if anyone would be interested in this, but it would be great if we can set time constraints on labels of items. Again using the example mentioned, "ABC company" would be shown instead of "DEF company" on the employee's item when qualifier end time (P582) under employer (P108) indicates that he/she left the company before the name change. --Wylve (talk) 10:52, 10 January 2014 (UTC)
That is not what I am trying to do, you are taking this further than I intend. I am solely focusing on a continuing entity that has a variety of official names, and wanting to record those for the item, and put dates against that item for the particular names.  — billinghurst sDrewth 16:36, 10 January 2014 (UTC)
I think the way it's supposed to be done is to create more items. Not sure how that falls with WD:N. I'm guessing that you can probably pass that hurdle pretty easily else the use case wouldn't be causing you grief. --Izno (talk) 18:56, 10 January 2014 (UTC)

New project to reflect the information that is available because of Wikidata

Hoi, when you look at the data in Wikidata itself, it is difficult to get much information out of it. For any and all WMF projects it is possible to get up to 50 search items in the extended search and, it gives access to Wikipedia and Wikivoyage articles, Commons categories and the "Reasonator". As it is now possible to have languages that do not have any Wikipedia, it becomes important to provide a search to them as well.

When you are adding information on a subject, I have found it really stimulating when more information becomes available in the Reasonator. I have discussed with several people if a separate project that lets the information shine that is in Wikidata. So far I have only been given positive reactions. The initial functionality will be based on what Magnus Manske has been building for us.

Thanks to support from the Dutch chapter, it is possible to create a proof of concept project. It aims to make information available in all the languages Wikidata supports. I hope it will be possible to make use of the "Universal Language Selector" and it would be really cool when it ties in with SUL, the "Single User Login".

I hope that it will drive the addition of labels in many languages. I hope that it will grow the interest in Wikidata.

Thanks, GerardM (talk) 15:53, 11 January 2014 (UTC)

RfC on property documentation

I'm not sure if it's appropriate to link to my RfC here, but I don't see anything in the guidelines against it, so I am being bold.

I couple of weeks ago I created an RfC, Property documentation redux, and I just updated it to try to make it more explicit and clear. Please join that discussion and/or vote, if you are so inclined.

Klortho (talk) 19:26, 11 January 2014 (UTC)

Is it an instance of if there is more than one of it?

  1. The Lincoln bible is an instance of:book but is The Hobbit an instance of:book or a subclass of:book? There are millions of copies of it but there is only one copyright.
  2. Is Bambi and instance of:film or a subclass of:film? After all there are multiple different editions.
  3. Is Good Friday and instance of:religious holiday or a subclass of:religious holiday? after all there have been over 2000 of them?
  4. Is Diet Coke an instance of:soft drink or a subclass of:soft drink?
  5. What about cars, computers, and other standard designs/mass produced items?
  6. What about ships? We seem to treat them differently.

'Instance of' and 'subclass of' are standardised properties. How do other ontologies handle this? There is a case for treating a mass-produced item as a class but if every other ontology treats it as an instance (of a standard design?) then I think we should follow them. Filceolaire (talk) 01:54, 5 January 2014 (UTC)

Filceolaire, bringing the topic up here was a good idea, for the reasons you cite. However, I think we should avoid trying to fulfill the domain and range axioms of P31 and P279 through statements like instance of (P31) class (Q217594) and subclass of (P279) class (Q217594). I just finished an in-depth explanation of why I think that here. Emw (talk) 17:37, 5 January 2014 (UTC)

What about number 3? Is 'Good Friday' a class? if so then what type of class is it? What is it a subclass of? Filceolaire (talk) 14:09, 5 January 2014 (UTC)

  • It's a Holiday class (or more specific type of Holydays). This class beeing itself a specific kind of class, hence it is a subclass of <class>. The property of this type is the definition of Holydays : all of its instances are days or period of times in which people are allowde not to work and in wich people goes to school. TomT0m (talk) 15:54, 5 January 2014 (UTC)
The class of 'Good Fridays' is a subclass of 'Religious holidays' (every member of one class is a member of the other class) but I'm not sure what type of class is the class of 'Good Fridays' is an instance of. Filceolaire (talk) 16:53, 5 January 2014 (UTC)
If it's an instance of something, as it is a class, it's an instance of some Holydays type. So the subclass tree of the classes it is an instance of is rooted into <Holydays type>.
  • Let's try to find a practical use of this hypotetical item. An interesting type of Holydays is the periodical Holyday that come back every year. Every instance of <periodical Holydays type>, like the Easter's Holyday, has a period in the year associated, a reason why it is at that time ... We can associate this to the typeclass : every periodical Holyday, whatever it is, has a period of the year associated and might have a story. So if a reasoner knows that, throw the class instance of (P31) typeclass he can suggest the user to add this statement ... but maybe it's also possible with staying with the fact that <Periodical Holyday> is a subclass of <Holydays>.
  • One other practical use: let's say that transitive properties are not implemented, and you want to know how to class a certain Holyday. If transitive properties are implemented, a relevant request would be give me all Holydays subclasses. But maybe there is a lot of them and it's hard to find what you are looking for. Now let's say transitive properties are not implemented. We can have a similar effects by searching every item that is an instance of the metaclass <Holydays type>, and maybe more relevant answers if the subclass tree is overloaded ... Anyway both solutions can be used at the same time. TomT0m (talk) 17:48, 5 January 2014 (UTC)
  • The Hobbit should be treated as a single work of writing, not as a class of pages glued together on which that work is printed. That's how we treat other works of writing such as poems, plays or short stories and more importantly that's how we think about books. (i.e. you'll never read the sentence "Tolkien has written millions of books".) Note by the way the ever sensible WikidataUsefuls script uses P31 for books. Mergehappy (talk) 23:01, 5 January 2014 (UTC)
    It's more accurate to state that The Hobbit is a (instance of (P31) sense) novel, or a fantasy novel, there is an item for that. What is important is not how you say that in english, it's how the item we use is defined. The definitions are moreother language indepedant, not like a common language expression. It is also a set of book objects, so a subclass of books. TomT0m (talk) 00:19, 6 January 2014 (UTC)
    Consistency and intuitiveness is important. I think we all agree that P31 should be used to classify the Jack London short story To Build a Fire (Q515819), the Anne Brontë poem The Passionate Pilgrim (Q1199790), the Samuel Beckett play Waiting for Godot (Q19871), and the memoirs of Winston Churchill. So it's counter-intuitive to argue against P31 in the case of London's novel White Fang (Q152267), Brontë's Agnes Grey (Q517172), Beckett's Molloy (Q1971806) and Churchill's Savrola (Q2228245). It's much more natural to treat all works of art in a similar fashion rather than sticking with the metonymy that confuses the work and the material it's printed on. Mergehappy (talk) 19:16, 6 January 2014 (UTC)
    To agree with that first I would have to understand what it means, and I fail. TomT0m (talk) 19:28, 6 January 2014 (UTC)
    Let's try this again then. My basic idea is that similar items should be treated in similar fashion. Short stories, poems, plays, memoirs have a P31 statement so it would be consistent to use it for other books. P31 is also used by other works of art including albums, songs, paintings and so on. As for my second point, I am pointing out that the issue is caused by our dual use of the word "book" as both a physical object and the work of art. That ambiguity is not problematic on Wikipedia but it's hard to handle on Wikidata. One solution would be to create a new item for written work and keep book (Q571) to deal with books. Mergehappy (talk) 20:06, 6 January 2014 (UTC)
    Hmm I agree mostly with that, but as novel (Q8261) is a subclass of work (Q386724) (it probably should be also a subclass of written work (see Talk:Q8261), all of this is already entailed. I'm not sure we really have something to discuss :) TomT0m (talk) 20:23, 6 January 2014 (UTC)
  • Mergehappy, the subclass relation denoted by subclass of (P279) does not indicate metonymy. In ontology, metonymic relationships are essentially mereological. They are represented with part of (P361) or a subproperty of it. A claim 'X subclass of book' is not asserting that X is a collection of pages; it is not confusing the work and the material it is printed on. It asserts that X represents a sets of things that are instances of the book X.
The trouble with considering most items Y about books to be instances is that it seems to preclude us from saying an item about another instance of a book Z that is an instance of Y is an instance of a book. This is the case with Bible (Q1845) and Lincoln Bible (Q1816474). To which item does instance of (P31) book (Q571) apply? Both, giving 'Bible instance of book, Bible subclass of book', 'Lincoln instance of Bible'? Is that valid metamodeling? I don't know for certain, but I would be surprised if it were. It implies that we need two definitions for 'instance of book': one an abstract thing that is a valid subject for properties like 'number of copies sold', one a different, concrete thing that is a valid subject of properties like 'location'.
Thus I think it would be better to say 'Bible subclass book', 'Lincoln Bible instance of Bible'. This is in line with the suggestion in An object lesson in choosing between a class and an object that "When in doubt, make it a class." This practice is seen in large ontologies like those of the Open Biomedical Ontologies -- which declare things like chemicals and genes classes (via rdfs:subClassOf / P279) and do not declare them as instances (rdf:type / P31 is generally not used). Since we have items about actual instances of things that are mass-produced (e.g. cars, books, organisms), it is it is best to consider those classes of thing as classes, not instances. Also, if we can foresee an item having subclasses, I think that's another basis to declare something as a class via P279. Emw (talk) 13:39, 8 January 2014 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘You still have to explain why individual films, paintings, poems, short stories, plays, ballets and all other individual creative works are considered instances with the sole exception of books. Consider the novel Fear and Loathing in Las Vegas (Q23395) and the film adaptation Fear and Loathing in Las Vegas (Q772435). The film is currently considered as an instance of film (Q11424), not as a class of DVD (Q5294) or something like that. My question to you is a) do you think that the film item should be repositioned as a class? and b) if not, don't you think it's wiser to have a coherent way to handle creative works? Mergehappy (talk) 20:17, 9 January 2014 (UTC)

Mergehappy, to answer your question in A: yes, I think subjects that have instances should be classified with subclass of (P279). This makes things coherent when we occasionally want to talk about instances of things that we would otherwise generally treat as individuals. A work can be treated as an individual even when it is classified with 'subclass of' and not 'instance of' -- this is done through punning. Stating that the film Fear and Loathing in Las Vegas (Q772435) to be a class of film does not mean it would be a considered a class of DVD; it would be a class of film (Q11424) (motion picture). As an aside, it is worth noting that paintings are a type of creative work with items that almost always refer to an instance and not a class, e.g. Mona Lisa (Q12418)
If we classify all creative works with 'instance of', then how do we handle cases like Bible (Q1845) and Lincoln Bible (Q1816474)? To which item does instance of (P31) book (Q571) apply? Having it on both would be incoherent, as I explain in my previous post. This situation also applies to automobiles: if Ford Model T (Q182323) is a instance of a car, then what is ThrustSSC (Q826174) an instance of? Emw (talk) 15:50, 11 January 2014 (UTC)
  • Sorry that I am coming late to this discussion. I am still playing catch-up. I think this is a really important set of questions, and User:Filceolaire started out by asking, "How do other ontologies handle this?" I am aware of a spec, FRBR (Functional Requirements for Bibliographic Records), that provides a useful framework for thinking about these sorts of questions. It describes four main concepts: Works, Expressions, Manifestations and Items:
In FRBR, a Work is a distinct intellectual or artistic creation, an abstract concept recognized through its various expressions (for example, your latest research paper); an Expression is the specific form that a Work takes each time it is ‘realized’ in physical or electronic form (for example, as a journal article); a Manifestation of an expression of a work defines its particular physical or electronic embodiment (for example online, print or PDF); and an Item is a particular copy of that you might own (for example the print copy of a journal issue on your desk).
The FRBR core ontology is documented here, and there is a related ontology based on it, called FABIO, that could also be used for guidance. I think that if wikidata items are aligned with the concepts from these ontologies, perhaps it could help to resolve some of the ambiguities.
From what I can tell, the four FRBR concepts are not related by subclass relationships. The classes "work", "manifestation", "expression", and "item" are all disjoint. So, for example, if you have many editions (each of which is an "expression") of Bambi (Q43051) (which is a "work"), this is encoded with frbr:realization statements pointing from the work to the expression. In other words, Bambi (Q43051) does not act as a class of which each of the editions is an instance.
So, trying to apply those concepts as best I can, here are my answers to the original questions:
  • The Lincoln bible is an instance of :book but is The Hobbit an instance of :book or a subclass of :book? There are millions of copies of it but there is only one copyright.
Lincoln Bible (Q1816474) → instance of (P31) → Bible (Q1845). 'Bible' is a class, and is a subclass of (P279) → book (Q571). This is just my opinion -- I don't think 'bible' could be identified as a single work, even though it's tempting. I'd suggest it is more akin to a class of works, like 'novel'.
The Hobbit (Q74287) → instance of (P31) → book (Q571), if 'book' is aligned with the FRBR concept of "work", which is not at all clear. I see that 'the hobbit' is also an work (Q386724), and that seems like an obviously good fit for FRBR "work". 'Book', on the other hand, is a subclass of both physical object (Q223557) (which would align with a FRBR "item") and a subclass of work (Q386724), so that is problematic.
  • Is Bambi and instance of :film or a subclass of :film? After all there are multiple different editions.
Bambi (Q43051) → instance of (P31) → film (Q11424). 'bambi' should correspond to a FRBR "work", and each separate edition should correspond to a FRBR "expression". From what I can tell, film (Q11424) seems to align with a class of works, analogous to "novel".
  • Is Good Friday and instance of :religious holiday or a subclass of :religious holiday? after all there have been over 2000 of them?
Good Friday (Q40317) → instance of (P31) → religious and cultural festive day (Q375011). This seems pretty obvious to me. If there is an item for one specific day, like March 29, 2013 (which I have yet to see) then it could be instance of (P31) → Good Friday (Q40317), in which case, Good Friday (Q40317) would be both an instance and a class (punning).
  • Is Diet Coke an instance of:soft drink or a subclass of:soft drink?
In my opinion, Diet Coke (Q1815345) → instance of (P31) → diet soda (Q3246353). I would say 'diet coke' is analogous to a "work", and diet soda is analogous to, for example, "novel". Each individual brand of diet coke could be also be instance of (P31) → diet soda (Q3246353), also be related to Diet Coke (Q1815345) by some other relationship analogous to frbr:realization.
Klortho (talk) 05:13, 12 January 2014 (UTC)
Interesting but ... this system disqualifies itself because it does not consider qualifiers. They are part of the Wikidata architecture and prevents much of the making up items and labels that are not used. Thanks, GerardM (talk) 08:43, 12 January 2014 (UTC)

Question about classes, and 'instance of' vs 'subclass'

Note: this conversation was copied from User_talk:Emw#Question_about_classes.2C_and_.27instance_of.27_vs_.27subclass.27 on 2014-01-08 by Emw per Klortho.

---

This is maybe a dumb question, but I want to understand classes vs. instances better.

What makes something a class? Taking plant (Q756), for example, I see that it is both:

Would either of these be sufficient for me to infer that plant (Q756) is a class (Q217594)? The latter, because I'd assume that the property subclass of (P279) has a domain constraint to class (Q217594).

What is the actual (inferred) statement that says that plant (Q756) is a class (Q217594)? (As I write this, I realize I'm pretty confused.) Is it:

I'd also guess that if there are any statements that specify this item as the object of instance of (P31), then it could be inferred to be a class. Is that right?

For a given (perhaps new) item, is there a canonical way to specify that it is a class?

If you have a minute or two to explain, I'd really appreciate it. Thanks! Klortho (talk) 20:07, 4 January 2014 (UTC)

Klortho, those are great questions. The answers are important, but not at all apparent without diving into the W3C specifications for RDF, RDFS, OWL, etc. I'll try to accurately summarize what they say about your questions.
In a nutshell:.
  • Classes are sets, and instances are elements of those sets.
  • Being part of an instance of (P31) or subclass of (P279) claim alone is sufficient to infer that an item is a class. As the definitions rdf:type and rdfs:subClassOf say, such items are implicitly instances of rdfs:Class.
  • Relatedly, it is invalid to assert in a Wikidata P31 or P279 claim that an item is a class as it concerns those properties (i.e. rdfs:Class). rdfs:Class is reserved vocabulary, and thus fundamentally different than any item on Wikidata, even those labeled 'class'.
  • There is no direct way in Wikidata to specify that a given item is a class. P31 and P279 are currently the only ways indicate an item is a class, but they are somewhat indirect.
  • An item can be both an instance and a class.
In full:
First, some definitions and background. A class represents a set of individuals, and individuals represent objects from a domain. As an OWL 2 formalism, instances are individuals that fulfill a class expression, but I have never noticed this fine distinction between 'individual' and 'instance' used in practice. The terms 'individual' and 'instance' seem effectively identical. A class is a set of instances.
When the specifications describe a class as a set, they mean it in a rather formal sense. For example, the OWL model theory specification defines subclass of (P279) as SubClassOf(d1, d2) = d1 ⊆ d2 -- 'subset of'. instance of (P31) is effectively ∈ -- 'element of'. The basis of P279 is rdfs:subClassOf and the basis of P31 is rdf:type.
The fact that P31 and P279 are based on rdf:type and rdfs:subClassOf -- special, built-in properties in RDF and RDFS -- entails certain things. First, to answer your questions, if an item is the subject or object of a subclass of (P279) claim, or if it is the object of an instance of (P31) claim, then it is a class. In other words, those conditions are sufficient to infer the item is an instance of a built-in 'class' concept, i.e. rdfs:Class.
I think we should avoid linking rdfs:Class to class (Q217594) or any other item in Wikidata for a few reasons:
  1. In Section 5.1: Classes, the OWL 2 syntax specification says "IRIs from the reserved vocabulary other than owl:Thing and owl:Nothing MUST NOT be used to identify classes in an OWL 2 DL ontology." In other words, classes in the ontology -- here, Wikidata -- should not make claims involving resources in the rdf:, rdfs: or owl: namespaces other than owl:Thing and owl:Nothing. (Small digression: 'owl:Thing' exists on Wikidata as entity (Q35120). Q35120 is the top concept, conventionally referred to as   in description logics, the basis of OWL.)
  2. class (Q217594) is actually ambiguous. There are different theories of sets, each of which assigns notably different properties to the term 'class' -- which one does rdfs:Class have the semantics of? (ZF, NBG, MK?) rdfs:Class is defined in RDF model theory. Per this technical note in the RDF MT document, rdfs:Class fulfills a notable condition of Zermelo-Fraenkel set theory, but I don't know if rdfs:Class and ZF classes are equivalent. (The second-to-last paragraph here and page 5 here seem to suggest they are not.)
  3. When we start talking about upper ontology on Wikidata, the project's assertions enter the domain of metaphysics and philosophical ontology to some degree. (A side note: I think Wikidata claims upper ontologies are useful as a way to practically organize our claims about more specific domains. I do not consider them to be political assertions about the nature of the universe.) While claims about metaphysics and philosophical ontology are in scope for Wikidata, the W3C Semantic Web standards aim to separate the languages themselves from those concerns:
[RDF model theory] tries to be metaphysically and ontologically neutral. It is typically couched in the language of set theory simply because that is the normal language of mathematics - for example, this semantics assumes that names denote things in a set IR called the 'universe' - but the use of set-theoretic language here is not supposed to imply that the things in the universe are set-theoretic in nature.
—Specifying a formal semantics: scope and limitations. RDF Semantics specification
Relating rdfs:Class to class (Q217594) or any other item on Wikidata seems like it would unnecessarily erode the separation between ontology as a philosophy and ontology as field of computer science that the W3C desires. Wikidata is in the business of making content claims (big and small) about the universe based on reliable sources from domain experts, but RDF, RDFS and OWL are designed to provide only the structure of those claims, not their content.
You also ask "is there a canonical way to specify that it is a class?" There is currently no standard way in Wikidata to directly specify that an item is a class. In OWL, this is done with <owl:Class ...> ... </owl:Class> (see e.g. trans.owl), but there is no analog of owl:Class on Wikidata. The most direct way we currently have is to put a P279 claim on the item; the next most direct is to make the item the object of a P31 or P279 claim on another item. These mechanisms seem OK to me for now. They encourage users to connect the items to our growing class hierarchy.
Since you note how plant (Q756) has claims for both instance of (P31) and subclass of (P279), you might be interested in the notion of metamodeling in OWL 2. I wrote about it a bit in Project chat a little while ago: Is Mary a scientist or a profession? Punning in Wikidata with OWL. In brief, a class can be an instance. Think of it this way: a set can be considered a subset or an element of another set. In other words, it is valid to state 'plant (Q756) instance of (P31) taxon (Q16521) and plant (Q756) subclass of (P279) organism (Q7239)'. Note that this is fundamentally different from how plant (Q756) is an instance of rdfs:Class and a class of subclass of (P279); this case mixes the syntax of the ontology (rdfs:Class) with the content of the ontology (organism (Q7239)) and is not what "metamodeling" refers to.
In that Project chat discussion, Filceolaire made a proposal similar to your question. I didn't address the proposal very directly, but my points above about rdf:Class and class (Q217594) do. I think we should avoid adding explicit claims like instance of (P31) class (Q217594) or subclass of (P279) class (Q217594) on items to fulfill a constraint that "Anything using the 'Subclass of' property should be an 'instance of:Class' or some subclass of 'Class'." That constraint is built into the definition of rdf:type and rdfs:subClassOf, and thus 'instance of' and 'subclass of'. Filceolaire's concern is prescient (as usual), it's just that it seems already addressed by default in the linked terms from the RDFS specification, and indicated against by the numbered arguments above.
Please let me know if clarification or more information would help. Emw (talk) 17:21, 5 January 2014 (UTC)

My name was mentioned above so I got a site notification so I'm going to put in my $0.02 worth.
First: Emw this discussion was fascinating and helpful. I hope my comments below add something.
'Plant - instance of - Taxon' tells us what type of class 'Plant' is. It is a taxon, which is a specific type of class, and it is an instance/member of the 'Taxon' class along with every other taxon.
'Plant - subclass of - organism' tells us that every instance/member of the 'Plant' class is also an instance/member of the 'organism' class. Note that none of these instances/members are Taxons; they are plants. None of them are instances/members of the 'Taxon' class. I think that this is the basis on which we should decide whether something is an 'instance of' or a 'subclass of'.
We can get even more meta. 'Plant' is an instance of 'Kingdom' and 'Kingdom' is a subclass of 'Taxon' but 'Kingdom' is also is an instance of 'Taxon rank'. It is a member of the Taxon rank class along with 'Species' and 'Genus'.
At least that is my opinion. Hope that helps. Filceolaire (talk) 22:05, 5 January 2014 (UTC)

This is great, thanks very much to both of you for the discussion!
"The terms 'individual' and 'instance' seem effectively identical." → I guess, not quite: an individual that is not declared to belong to any class is still an "individual", but not an "instance".
"'is there a canonical way to specify that it is a class?' ... The most direct way we currently have is to put a P279 claim on the item; the next most direct is to make the item the object of a P31 or P279 claim on another item. These mechanisms seem OK to me for now." → The problem is that in the absence of any "subclass of" statements, there is no easy way for an editor to see that an item is a class. It may be a class because other items are "instances of" it, but those statements are not visible on the original item's page. I understand your reasons (I think) for not wanting to *require* it, but would making such an item be "instance of" → "class" cause any problems?
I think I now understand the issues that I brought up originally, but one (new?) area of confusion is how classes are used in Wikidata as the objects of properties other than instance of (P31) or subclass of (P279). Three recent examples:
  • In your, "Is Mary a scientist or a profession?" post, you wrote about "statements of the form 'X occupation Y' entail that X is an instance of Y". The example was Mary → occupation → scientist, where scientist is a class.
  • In the pathogen transmission process property proposal, the objects of that property are all classes.
  • Likewise, for the sex or gender (P21), the objects are typically subclass of (P279) "male" or "female"
So, I think these are examples of punning, right? In the statement Chelsea Manning (Q298423) → sex or gender (P21) → trans woman (Q15145782), I think trans woman (Q15145782) is being treated as an individual, not as a class. But on the other hand, since trans woman (Q15145782) → subclass of (P279) → female (Q6581072), I think one should be able to infer the sex or gender (P21) → female (Q6581072) relationship. But maybe there is no general rule like that.
Klortho (talk) 04:32, 6 January 2014 (UTC)
I do not understand why you say that 'trans woman (Q15145782) is being treated as an individual'. I think the P21 statement means that the person (in this case, Chelsea Manning (Q298423)) is an instance of a certain sex or gender (in this case, trans woman (Q15145782)). So the gender or sex is a class. Bever (talk) 05:17, 7 January 2014 (UTC)
@Bever:, that is what I thought at first, but I don't think it is right, and the distinction is crucial, see my last comment under "still to do", here. If Chelsea Manning is an instance of transgender female, then we would be justified in inferring that she is a biological female, which she is not. In the Semantic Web (Emw can correct me if I'm wrong) you're not allowed to infer an "instance of" relationship from some other property, unless that property is declared to be a subproperty of "instance of". And so it still seems to me that the object of sex or gender (P21) should be interpreted as an individual/instance, rather than a class, but that begs the question, what exactly are the subclass relationships there for? Klortho (talk) 02:13, 8 January 2014 (UTC)
Klortho, I do not think it is valid OWL DL to declare a subproperty of P31, i.e. rdf:type. Filceolaire and I have been talking about this intermittently for few a months. My argument has been that I think we should avoid declaring subproperties of rdf:type (P31) because I have never seen such usage in any of the ontologies I've looked at, and it's unclear whether it's even supported by OWL.
I found a note by Ian Horrocks, an editor of the W3C OWL recommendations, that seems to support the notion that it is not valid in OWL DL to have a statement 'X rdfs:subPropertyOf rdf:type':
...the restriction being discussed here, i.e., not being able to create a subPropertyOf rdf:type, is nothing to do with DLs per se, but is required in order to keep the language inside what I think we agreed to call "conventional" FOL. In fact separating the syntax of the language from the domain of discourse is fundamental to most logics.
—Ian Horrocks, Re: Inferring Class Membership w/o OWL Full?
I also found a relevant note in a report about expressing Dublin Core in RDF:
dcterms:type vs rdf:type
There is an issue with declaring a sub-property
of rdf:type. OWL-DL will not accept the assertion that
dcterms:type is rdfs:subPropertyOf rdf:type.
Proposal: This is really an issue for the DCMI Usage Board.
-- Remove statement that dcterms:type is rdfs:subPropertyOf rdf:type.
-- State that in mapping from DCAM to RDF, dcterms:type maps to rdf:type
—Mikael Nilsson, Report from public comment period on Expressing Dublin Core metadata using the Resource Description Framework (RDF)
Regarding sex or gender (P21), I think the objects of such claims are conventionally considered classes, not instances. All classes can be considered instances, but 'man' and 'woman' are canonically treated as the former. See for example http://www.w3.org/TR/owl2-primer/.
You also say "In the Semantic Web (Emw can correct me if I'm wrong) you're not allowed to infer an 'instance of' relationship from some other property, unless that property is declared to be a subproperty of 'instance of'." We can certainly infer 'instance of' relationships from some other property (and, as noted above, we shouldn't do it with a subproperty of P31). This is the point of P279 (rdfs:subClassOf): if Methuselah (Q590039) is a tree, then Methuselah is an organism, because a tree is a subclass of organism. In fact, almost all properties can be used to infer rdf:type relationships, since domain and range entail that the subject or object of a claim with a given property is an instance of the class denoted in the domain or range. (See the definitions of domain and range in those RDFS specification links.) So we see that what classes an item is an instance of can be inferred in many ways. Emw (talk) 05:37, 8 January 2014 (UTC)
The restrictions on declaring subproperties of rdf:type are interesting. And, I hadn't considered subClassOf, domain, and range, when I wrote that 'instance of' relationships can't be inferred. I'm also familiar with a lot of OWL, and I reviewed the primer you referenced. However, the outstanding question here is very simple, and has not been addressed: given the statement "Chelsea Manning (Q298423) → sex or gender (P21) → trans woman (Q15145782)", where trans woman (Q15145782) is a class, can one infer "Chelsea Manning (Q298423) → instance of (P31) → trans woman (Q15145782)"? If so, by what mechanism? By convention? Does the Wikidata RDF export reflect that? If so, how is that different from declaring sex or gender (P21) to be a subproperty of rdf:type?
You pointed me to the OWL primer as a reference to suggest that objects of properties such as sex or gender (P21) are often classes, but I don't see anything in that document that supports that. Yes, they have classes :Man and :Woman in the sample ontology, but I don't think there are any implied properties similar to Wikidata's sex or gender (P21). In other words, :Man and :Woman are only used as classes, in class-type relationships like type, subClassOf, and all of the OWL restriction statements, for example. The more I dig into this question, the more I am convinced that statements of the form "sex or gender (P21) → trans woman (Q15145782)" are just wrong, and should be replaced with "instance of (P31) → trans woman (Q15145782)".
Since I doubt that I'm the only one confused by this, do you think it would be good to move this discussion to Project_chat? Klortho (talk) 14:56, 8 January 2014 (UTC)
In OWL, there is several ways to a class (see http://www.w3.org/TR/owl-ref/#ClassDescription about OWL1 but it should not be so different with the version 2). The one that answer to your question is, in this link property description. With a property restriction, it is possible to define the <female> class (or an anonymous class as all the individual who have a sex statement with the <female> (assuming the fact it is both a class and a value in the same definition is not a problem) value. This way you do not have to explicitely say that an individual belongs to the female class without stating it explicitely, a resoner deduces that. Of course this is more interesting with more complex class definition. TomT0m (talk) 17:12, 8 January 2014 (UTC)
So, User:TomT0m is suggesting that female be considered both an individual ("value") and a class, and that a property restriction could be defined to, in effect, say that if ":foo → sex or gender (P21) → female (Q6581072) (qua individual)", then ":foo → instance of (P31) → female (Q6581072) (qua class)". I think the property restriction would look like this:
EquivalentClasses(
    :female (qua class)
    ObjectHasValue( :sex_or_gender :female (qua individual) )
)
Two points:
  • I don't think this is any different from declaring sex or gender (P21) to be a subPropertyOf rdf:type. User:Emw, I reviewed some of the references you gave above about not being able to do this, and they all talk about it being a limitation of OWL DL (hence, OWL 1) in which punning is not allowed. So I think that the restriction is because, in OWL DL, properties are limited to either Datatype or Objecttype properties, and are not allowed to take classes as their objects. So if punning is allowed, as above, in the convoluted way of defining the sex or gender (P21) to be a subClassOf rdf:type, then I don't see any reason it couldn't be just specified explicitly. In other words, "sex or gender (P21) → rdfs:subPropertyOf → rdf:type" is cleaner, and does the exact same thing.
  • My thinking has evolved, and whereas before I said I thought that this was wrong, I now think that it's fine, but that we should explicitly acknowledge that we are defining sex or gender (P21) to be a subPropertyOf instance of (P31), and that that's a good thing. It seems to be what everybody intuitively understands about how this property works. It's just saying "anything that has a (biological) sex of female (Q6581072) is in the class female (Q6581072)".
I'd still like to know how this manifests in the RDF dump (and by what mechanism) but I guess that's a research project for me. Klortho (talk) 03:26, 9 January 2014 (UTC)

Klortho, per the quote from Ian Horrocks, the restriction on creating an rdfs:subPropertyOf rdf:type is because it is required to keep the language (OWL DL) in conventional first-order logic (FOL), which separates the syntax of the language from the domain of discourse. That doesn't seem to have anything to do with whether something is an instance or a class or both via punning.
This restriction on making statements about the language's syntax in an OWL DL is reiterated in a paper by another editor of the OWL W3C specification. It notes:
OWL DL differs from OWL Full mainly in that (i) it does not allow one to state axioms about the built-in vocabulary (i.e., the symbols, such as owl:allValuesFrom, used in the definition of the semantics), (ii) it strictly separates the sets of symbols used as concepts, roles, and individuals, and (iii) it enforces the well-known restrictions required for decidability, such as allowing only simple roles in number restrictions"
—Boris Motik, On the Properties of Metamodeling in OWL
The paper goes on to provide a describe a decidable method to handle (ii), but maintains that violating (i) would make an ontology undecidable, and thus is not permitted in OWL DL. This is a theoretical basis for metamodeling in OWL 2 DL; see the reference labeled 'Metamodeling' http://www.w3.org/TR/owl2-new-features/#References. It also indicates that making a statement about the language's syntax -- which Horrocks says is what rdfs:subPropertyOf rdf:type would be -- is not enabled by the kind of metamodeling done via punning. Given that, it seems creating a subproperty of rdf:type would not be valid in OWL 2 DL. Class membership is a basic function in inference, and something we really want to be decidable. So, whereas creating subproperties of rdf:type (or rdfs:subClassOf, another built-in vocabulary term) seems undecidable, and whereas instance of (P31) and subclass of (P279) are designed to have the semantics of rdf:type and rdfs:subClassOf, I do not think we should be creating subproperties of P31 and P279.
Clearly, though, we need a way to infer that an item is a "woman" or "man" if it has certain values for certain properties. The way to do this is via property restriction, as TomT0m notes. It is supported in OWL 2 as described in http://www.w3.org/TR/owl2-primer/#Property_Restrictions. Property restriction brings core logical operators to OWL: particularly, the existential quantifier ∃ -- "for some" -- and universal quantifier ∀ -- "for all". (Note: quantifier != qualifier.) This would enable us to make statements like "All people who are woman and parents are mothers", which is important. Emw (talk) 15:49, 12 January 2014 (UTC)

Invalid coordinates

I noticed a few items with coordinates that can't be retrieved from Wikipedias but seem to work fine in the item page. When I corrected them, the old (invalid) coordinates doesn't even show in the diff. [4]

@Michiel1972: as far as I saw, the invalid coordinates were imported by RobotMichiel1972 from nlwiki.--Pere prlpz (talk) 12:56, 10 January 2014 (UTC)

Hmm, I thought I fixed all these errors. Can someone please produce a list of items that still contain errors? I have a bot to fix these errors. Multichill (talk) 20:07, 10 January 2014 (UTC)
I can't produce such a list, but there are 4 articles in ca:Categoria:Manteniment plantilla coord. I tested {{#property:P625}} and it yields "The value is invalid and cannot be displayed".--Pere prlpz (talk) 12:45, 14 January 2014 (UTC)

Samuel De Wilde

I am having difficulty editing "Samuel De Wilde" (Q7411251). The Commons category has been renamed from "Category:Samuel de Wilde" to "Category:Samuel De Wilde", but Wikidata won't let me change the de to De, and keeps reverting to de. Please help! Smuconlaw (talk) 17:50, 12 January 2014 (UTC)

The name is a Dutch name. In the Dutch language the "de" part is exclusively in lower case. Thanks, GerardM (talk) 18:01, 12 January 2014 (UTC)
Hmmm. English Wikipedia and the National Portrait Gallery, London – presumably a reliable third-party source – both have it down as "Samuel De Wilde", which is why I changed all related Wikimedia Commons categories to match. Note that De Wilde lived his entire life in the UK, so it is conceivable that he did not spell his name according to Dutch conventions. Anyway, the problem I mentioned resolved itself after "Category:Samuel de Wilde" was deleted by an administrator. Smuconlaw (talk) 19:49, 12 January 2014 (UTC)

Mistress of ?

Harriet Howard (Q639964) was the mistress of Napoleon III (Q7721). How would you indicate this ? cheers, --Teolemon (talk) 22:30, 13 January 2014 (UTC)

unmarried partner (P451) is probably the closest property we have. Note that there's some divergence between label and description, or between labels in different languages. While I don't think you'd call a mistress a "cohabitant", the English description just says "Someone with whom the person is in a relationship without being married", which matches here, I'd say. The property was originally proposed as "Partner". --YMS (talk) 07:06, 14 January 2014 (UTC)

Blood type property

It would be useful to have a property for blood type of humans, as it is recorded on some of the wikis as in Ishikawa Taisuke's page on ja:wiki.

Presentime (talk) 22:55, 13 January 2014 (UTC)

Maybe it's against privacy in some countries for some people, but if it's shown in some wikipedias with references on reliable sources that are available to public, there is no problem.--Pere prlpz (talk) 11:35, 14 January 2014 (UTC)

Merging

as you can see in this RfBA. I'm told not to merge into lower Q-id but to bigger item, I don't like this, do you think it's okay? Amir (talk) 14:50, 14 January 2014 (UTC)

This was just recently discussed on Wikidata talk:Requests for deletions#Why does the item number matter?.
I always was in favor of the "merging into the more complex item" thing anyway, but in your bot's case even more, as you always have one item that is empty and therefore (other than in many other cases) it's totally clear which is the more complex and more likely to be used item. --YMS (talk) 14:56, 14 January 2014 (UTC)

Q15418575

Is it really useful?--GZWDer (talk) 15:51, 14 January 2014 (UTC)

I've deleted the page. --Jakob (talk) 15:53, 14 January 2014 (UTC)

Q15476144

Is it a duplicate of Q4663903?--GZWDer (talk) 16:06, 14 January 2014 (UTC)

Yeah, but Q15476144 has hundreds of links. Anyone up to delinking it so it can be deleted? --Jakob (talk) 16:18, 14 January 2014 (UTC)
@King jakob c 2: That is User:BeneBot*/movelinks.js.--GZWDer (talk) 16:20, 14 January 2014 (UTC)
Ah, thanks. I've done that. --Jakob (talk) 16:44, 14 January 2014 (UTC)
I tried your request again with the <> removed since the bot didn't understand it. The Anonymouse [talk] 16:51, 14 January 2014 (UTC)
Thank you both and apologies for the extra brackets - I haven't used that tool before. --Jakob (talk) 17:01, 14 January 2014 (UTC)

migrating away from P107

My bot finished everything I programmed it for the migration away. Now we have less than 1.8M statements that needs to be removed.

My main question: What will be the next step? Best Amir (talk) 13:47, 8 January 2014 (UTC)

Amir, thank you for your help, your work has been essential to improving how we model data. Are there lists of the problematic cases anywhere? More importantly, do you have an impression of what is preventing the remaining P107 claims from being removed -- especially for P107 (P107): person (Q215627)? Do those items have claims we could use to infer an appropriate P31 claim? Emw (talk) 13:59, 8 January 2014 (UTC)
We could replace it with instance of (P31) --> human (Q5) --Jakob (talk) 14:16, 8 January 2014 (UTC)
There is no way to infer being human from statements (because fictional characters also can have birth date, death date, birth place, ...) I used articles in English Wikipedia and analyzing them in order to infer it (just simply checks categories like "Category:2014 births" is used in the content or not and checks "Category:Fictional" is used or not) what my bot needs is another way of analyzing (I can do some l10n and expand the code to check other language and I'll do it very soon but I think it won't make so much difference) Amir (talk) 14:19, 8 January 2014 (UTC)
@King jakob c 2: We can't because P107 (P107):person (Q215627) can imply on human beings, deities, fictional charterers, and sometimes more strange things like twins. for more information see the RFC (I can't find the link right now) Amir (talk) 14:19, 8 January 2014 (UTC)
Amir, how about the replacement: P107 (P107)person (Q215627) + country of citizenship (P27) ==> instance of (P31)human (Q5) ? See some of the 125K statements here http://208.80.153.172/wdq/?q=claim[107:215627]_AND_claim[27]. Are there also examples of fictional characters with a nationality? Can we tolerate some (small) errors in such automatic taks (I guess there will always be errors in a database)? --Zuphilip (talk) 16:38, 8 January 2014 (UTC)
[5] Currently, about 1000 items with instance of (P31) → fictional character (Q95074) have country of citizenship (P27). The Anonymouse [talk] 16:50, 8 January 2014 (UTC)
As The Anonymouse mentioned I don't think it's a good idea Amir (talk) 07:16, 9 January 2014 (UTC)
I guess for creative works and organisations, checking some more things with a bot should not be a big deal (I checked some of those in the last days, and most of them were clearly marked as companies, bands, sports teams etc. or albums, books etc. with according infoboxes, categories etc.). However, they're the small portion anyway.
@Amir 62k items have P107=work (Q386724) and P31 set [6]; just deleting P107 on those would be OK I believe; similarly for 42k organization (Q43229) having a P31 [7]. LaddΩ chat ;) 23:22, 8 January 2014 (UTC)
For persons, maybe a user script could help better than a bot can. So everytime a P107=Q215627 statement is displayed on an item, the user can choose to transform it iunto a P31=Q5, P31=Q215627 etc. statement by just one click. For two million items, this probably would take a while even if we make it a default gadget and have some eager users, but in the end I imagine it could work. Similar could be done for geographical features, where a user could transform it into towns, asteroids, lakes etc. In any case, such a script should provide as many reasonable options as possible, to avoid that the users just choose the one that's closest but not correct (e.g. if the geographical feature can be converted by one click into a lake, but not a river, people might choose lake for rivers as this is as good as it gets from what they see, just like people willingly tagged everything as a person before). --YMS (talk) 15:42, 8 January 2014 (UTC)
@Ladsgroup: For the last items about persons we need to have a more complex analysis: 1) check if the property creator (P170) exists or not (most fictional characters have a creator), 2) check if a birth date and a death date exist: few fictional characters have both. Then we can add another check for birth and death date: which calendar is used, is the life duration higher than 130 years,... 3) for deity take all subclasses of item deity and check if these value are present in the items. Snipre (talk) 19:17, 8 January 2014 (UTC)
I'll do it, thank you for the idea Amir (talk) 07:16, 9 January 2014 (UTC)

If no one objects on removing P107:work (Q386724) and organization (Q43229) if there is a P31, I'll do it via my bot. Any objections? Amir (talk) 07:16, 9 January 2014 (UTC)

You may want to run over the P107 set looking for P279 as well (WDQ is being slow for me else I'd produce a link). --Izno (talk) 14:55, 9 January 2014 (UTC)
Sum up to 100K = 85 % of the two small groups, which would left us with 24K statements there. --Zuphilip (talk) 15:45, 9 January 2014 (UTC)
@Ladsgroup:@Zuphilip: I guess that arbitrary classes is not appropriate (see reason by Bever below). Better to use all descendants of specific class, like for creative works: http://208.80.153.172/wdq/?q=claim[107:386724]_AND_tree[386724][][31,279]. Infovarius (talk) 11:50, 13 January 2014 (UTC)
Infovarius Sure, fine for me. If someone wants to work on the remaining 1000 items here is the list (loading takes its time): http://208.80.153.172/wdq/?q=claim[107:386724]_AND_claim[31,279]_AND_noclaim[31:(tree[386724][][279]),279] for creative work. --Zuphilip (talk) 12:07, 13 January 2014 (UTC)
I analyzed the remaining 1000 objects to which class instance of (P31) the belong to. Actually, there are only a few classes which occur very often. See
--Zuphilip (talk) 22:57, 14 January 2014 (UTC)
also remove P107 if is present instance of (P31)->human (Q5) --Rippitippi (talk) 04:11, 10 January 2014 (UTC)
I guess you can remove P107 also when Swedish urban area code (P775) is present. All those items should have instance of (P31) urban area in Sweden (Q12813115) with qualifiers and sources. But I guess the latter can wait. We are not in a hurry. -- Lavallen (talk) 12:50, 10 January 2014 (UTC)

I started the bot again with new rules, It'll be finished very soon Amir (talk) 10:16, 11 January 2014 (UTC)

I hope I am not too late, but just removing P107 may not be a good idea in all cases, in my opinion. Especially, when P107 and other statements contradict each other, this may reflect that the item has different meanings, a problem which has been brought up before. For example when P107 says 'creative work' but other properties indicate a person, or the other way around, the item is probably about a creative work and main character with the same name (like Maya the Bee (Q752027), which until recently had a sex property indicating that Maya is a female). So when there will be a good solution for these kind of items, it is nice that the items where the problem exists can still be find through this non-intended use of P107. Bever (talk) 19:41, 11 January 2014 (UTC)

Parsing the Wikipedia article, it should be possible in most cases to distinguish a human from a fictional character. In an article about a human, the date of birth is written right after the name. In an article about a fictional character, the date of birth isn't mentioned at the beginning of the article but words like fictional or character are showing up in the first sentence. --Pasleim (talk) 19:32, 11 January 2014 (UTC)

False positive in property format check

The constraint violations for P1067 lists that the code 38 violates the pattern (\d\d){1,4} for that property - but it should be a match with that RegEx and not a violation. Is there something wrong with the regex constraint tests? Ahoerstemeier (talk) 23:21, 14 January 2014 (UTC)

I fixed the format constraint as it is described at Template:Constraint:Format. --Zuphilip (talk) 11:11, 15 January 2014 (UTC)
Didn't know about the need for nowiki, thanks for the fix. LaddΩ chat ;) 04:14, 16 January 2014 (UTC)

Lua and Commons

Hello,
I am a commons contributor writing templates and Lua modules.
I currently have a big issue:
My lua module fails to retrieve data from wikidata as the lua instruction mw.wikibase.getEntity() does not work anymore.
More technicaly: mw.wikibase is Nil even if the wikicommons page is linked to a wikidata item.
Any idea what is happening?
If you don't know, where can I ask for help?
Best regards Liné1 (talk) 20:07, 15 January 2014 (UTC)

If nobody can answer here you may go to WD:DEV. -- Bene* talk 20:21, 15 January 2014 (UTC)
Actually, phase 2 support is not live for Commons at this time. --Rschen7754 04:57, 16 January 2014 (UTC)

Request for comment on Commons: Should Wikimedia support MP4 video?

I apologize for this message being only in English. Please translate it if needed to help your community.

The Wikimedia Foundation's multimedia team seeks community guidance on a proposal to support the MP4 video format. This digital video standard is used widely around the world to record, edit and watch videos on mobile phones, desktop computers and home video devices. It is also known as H.264/MPEG-4 or AVC.

Supporting the MP4 format would make it much easier for our users to view and contribute video on Wikipedia and Wikimedia projects -- and video files could be offered in dual formats on our sites, so we could continue to support current open formats (WebM and Ogg Theora).

However, MP4 is a patent-encumbered format, and using a proprietary format would be a departure from our current practice of only supporting open formats on our sites -- even though the licenses appear to have acceptable legal terms, with only a small fee required.

We would appreciate your guidance on whether or not to support MP4. Our Request for Comments presents views both in favor and against MP4 support, based on opinions we’ve heard in our discussions with community and team members.

Please join this RfC -- and share your advice.

All users are welcome to participate, whether you are active on Commons, Wikipedia, other Wikimedia project -- or any site that uses content from our free media repository.

You are also welcome to join tomorrow's Office hours chat on IRC, this Thursday, January 16, at 19:00 UTC, if you would like to discuss this project with our team and other community members.

We look forward to a constructive discussion with you, so we can make a more informed decision together on this important topic. Keegan (WMF) (talk) 06:47, 16 January 2014 (UTC)

Rank not yet used for inclusion in Wikipedia?

At Mueang Bueng Kan (Q476000) I have added both the current value for Thailand central administrative unit code (P1067) as well as its old value before the creation of the new province, and given the current value a higher rank. However, when I try to include this statement into the Wikipedia article, it shows both codes separated with a comma. Is it a known bug (or better - not yet implemented feature) that the rank not works the way it is supposed to be? Ahoerstemeier (talk) 08:20, 16 January 2014 (UTC)

I think this is working as intended. Ranking is more for the API side (currently) I think where you can get both values (and should be able to get both values IMO (Lydia Pintscher pinged because I know you were thinking about this; we should let the data users decide)) as well as their rankings. It seems to me that a deprecated ranking maybe should not be returned in infoboxes... --Izno (talk) 15:01, 16 January 2014 (UTC)
Hey :) This is indeed not done yet. It's on my short-term list to change the parser function to only return preferred values. I will however have to think some more about how to best do this. I'll make a larger announcement when I have details. I hope I can discuss the last details with devs over the coming week. The bug for tracking it is bugzilla:58403. --Lydia Pintscher (WMDE) (talk) 15:42, 16 January 2014 (UTC)
Help:Ranking says that high rank will be showed in infoboxes by default, so I would expect to see only the statement(s) with the highest rank, maybe omitting deprecated ones if these are the only ones. Thus if there is no high ranked statement, the normal ranked one will be shown. As none of those statements I am working on is used anywhere in Wikipedia yet it's no big deal this is not working yet - I can continue to set the data with ranks to differentiate between currently correct statements and outdated ones. Ahoerstemeier (talk) 20:28, 16 January 2014 (UTC)
Note that it is better to use Lua for infoboxes, and in Lua you can retrieve the rank easily (actually, by default fr:Module:Wikidata only returns preferred values when they exist and normal values otherwise). I certainly hope the parser will not be changed to return only preferred values without any fallback mechanism, at least not in the short term. The overwhelming majority of statements are currently ranked "normal". --Zolo (talk) 21:44, 16 January 2014 (UTC)

LIGA - interwiki link doesn't work

The en:WP entry at LIGA includes (other languages) a link to de:wp which doesn't work. The wikidata entry says the German version is (between asterisks) *LIGA (Fertigungsverfahren)*, but the link goes to the de:disambiguation page for LIGA. I can't see how to fix this - perhaps it is a bug. Imaginatorium (talk) 10:39, 16 January 2014 (UTC)

There was a local interwiki link at enwiki, that caused the problem. I removed it. Kind regards, Lymantria (talk) 10:45, 16 January 2014 (UTC)
Thanks! Now I should know how to fix it next time. Imaginatorium (talk) 13:52, 16 January 2014 (UTC)

Merge

Hi, is possible to merge Q4592255 and Q13407457? Lugusto (talk) 17:29, 16 January 2014 (UTC)

  Merged by ValterVB. Thanks :) Del♉sion23 (talk) 20:11, 16 January 2014 (UTC)

Automatic update when pages are renamed

Are items updated when pages are renamed? I read about this kind of automatic update once, but I don't know about the developments, if any.--Erasmo Barresi (talk) 17:54, 16 January 2014 (UTC)

Yes, they are updated when renaming a page. -- Edinwiki (talk) 18:01, 16 January 2014 (UTC)
It does work in Wikivoyage since July 23 2013 and in Wikipedia since July 25.--GZWDer (talk) 05:14, 17 January 2014 (UTC)
It's not flawless though; under certain conditions, it doesn't happen. --Rschen7754 05:36, 17 January 2014 (UTC)

Chinese Wikivoyage

Just a note to say that Chinese Wikivoyage went live today so the import onto Wikidata begins. There is a list here of articles that have not been connected yet. Hopefully it will go down quite quickly. Del♉sion23 (talk) 21:50, 15 January 2014 (UTC)

@Delusion23: Well, first of all, congratulations. :) +1 for the automatic upload of pages that already have valid links to other Wikivoyages, not so sure about the ones who currently have no connection. --Sannita - not just another it.wiki sysop 14:49, 16 January 2014 (UTC)
  • I launched my bot, but there are very few articles were added to wikidata. In first view, the main problem is preffix :voy in interwiki that was actual when the project was in the incubator. But now it is incorrect. I plan to modify my bot's code and will run it in a few days again. --Emaus (talk) 22:49, 16 January 2014 (UTC)

Unlinked item deleted

Some days ago, I manually created items for the local government units in Phuket province which had no corresponding Wikipedia yet. I certainly set both a label and a description, but did not yet many statements yet, as I planned that to be done by bot. Now User:TambonBot was trying to add more data, and at least Q15507095 was deleted as being "Empty / unlinked item". As I cannot check what I did add at that time - what is the policy of deleting items without sitelinks? How many statements do they need to be kept? Or they need to show in one "What links here"? Wouldn't it have been nice to ask me? Now I have to check all the item IDs I saved in my data again if they are still valid, or if I have to created them new - and scare that they might disappear again. They are notable administrative units, just apparently not yet that much of interest that someone started an article on them. Ahoerstemeier (talk) 10:40, 16 January 2014 (UTC)

That's a bit difficult for the Thai municipalities, since there are two administrative structures in parallel in Thailand. The central administration has a clear hierarchy, there it is easy to add that backlink. But the local governments in Thailand are no direct subunit of any of the central administrative unit, thus they IMHO don't belong into contains the administrative territorial entity (P150). Adding them there only to prevent them from being deleted by over-active admins is not a good solution. Would it be sufficient to add links on a dummy sub-page of my user page like User:Ahoerstemeier/Unlinked Thai municipalities, until there is a property where these items can be linked correctly? Ahoerstemeier (talk) 11:17, 16 January 2014 (UTC)
I do not think it will help to link from a "dummy sub-page". Use shares border with (P47), it gives you links between the items, and ask for help at WD:AN if something have to be restored. -- Lavallen (talk) 16:44, 16 January 2014 (UTC)
That is a possible work-around, though it is sometimes difficult to have the neighboring municipalities correctly - I haven't yet found a good authoritative map of Thailand going to that detail. But I still wonder why these items were deleted. They did fall into the notability criteria, especially the one item mentioned above should have had the description "town in Phuket, Thailand", and thus it would have been easy to see that it fits into criteria 2 of the notability criteria. At least I could have been contacted, it took me maybe one hour to set these 17 items with some basic info and store them into my XML files, to use for the bot now. Shouldn't there be an additional clause on the create new item page, explaining that without links an item may get deleted without warning? Ahoerstemeier (talk) 20:15, 16 January 2014 (UTC)
I agree that they probably have Notability. But it's not always obvious for all users, especially since we have many many items that are proposed for deletion every day, and many of them are not proposed at all, we have to find them through RecentChanges and lists made by robots. It's as simple as some admins have to much to do, and therefor make mistakes sometimes. If you can find a way to create links to them from other items, admins and robots have to take more caution. You can for example identify items about places in these "local government units" and add located in the administrative territorial entity (P131) to them. -- Lavallen (talk) 11:54, 17 January 2014 (UTC)

Can I chose nameform of a property?

The capital of the Republic of the Congo (a.k.a. Congo-Léopoldville) was Léopoldville, now known as Kinshasa. Léopoldville is now an alias for Kinshasa, which means that the capital of Congo-Léopoldville will be displayed as Kinshasa. Is there a way (a qualifyer?) to select another name form of the property for this item? Or would it make more sense to create a new unique item for the city under it's former name? Rotsee (talk) 07:48, 17 January 2014 (UTC)

I am not aware of any such property, but I would need it as well for the Thai subdivisions, which sometimes change their name. If the name were a property, one could easily add two statements each with qualifiers to mark the date when the name change became effective. But it would have to be a multilingual property, as the name changes in the local alphabet as well as in all the transcriptions. For persons, we have P513 (P513), but that of course not fits here. Ahoerstemeier (talk) 08:16, 17 January 2014 (UTC)
Such properties are proposed and we have agreed to create them as soon as we have the technical tools to create them. This since we do not think "string-datatype" is good enough for this issue. -- Lavallen (talk) 11:59, 17 January 2014 (UTC)
Ok, thanks! Rotsee (talk) 13:39, 18 January 2014 (UTC)

Q12078

Q12078 doesn't load for me completely/times out (XP, FF 26). Is it a problem with my browser? Or corrupt data? --Murma174 (talk) 16:52, 17 January 2014 (UTC)

This item is definitely a cancer. Seriously, this seems to be related to Wikidata:Project_chat/Archive/2014/01#Unresponsive_script. --Stryn (talk) 17:10, 17 January 2014 (UTC)
Thanks for the link! But even pressing the 'continue' button doesn't help. Appearently Kompakt-Bot added some nonsense-links, maybe an admin could delete these additions? Or someone with access would be able to add the language-link to Nordfriisk (frr): 'Kreeft' Thanks in advance! --Murma174 (talk) 17:42, 17 January 2014 (UTC)
  Done with Chrome and some patience. The page contains 98 statements with the property ICD-10 (P494) and 100 statements with the property ICD-9 ID (P493) everything with the source to en:Cancer. Does this mass of statements really make sense? Or could it also be given by a range as in English wikipedia? Actually, the ranges are even part of the classification itself, see http://www.icd9data.com/2013/Volume1/140-239/default.htm and http://apps.who.int/classifications/icd10/browse/2010/en#/C00-C97--Zuphilip (talk) 17:59, 17 January 2014 (UTC)
Im am to blame for the import of the ICD-codes and the other medical identifiers (Kompakt-Bot just fulfilled the request). I did think about ranges at the time, but they didn't seem structured enough and I am still ignorant about how the query system will handle ranges. The ICD codes are somewhat primitive, because instead of single code for a group of diseases, it lists every single code of the diseases that belong to that group. In general the medical identifiers don't overlap well with our items. The constraint violations (Wikidata:WikiProject_Medicine#Data_aquisition_progress) are in the thousands for some of the identifiers. Maybe somebody has an idea how to improve the situation? --Tobias1984 (talk) 19:52, 17 January 2014 (UTC)
Tobias1984, I would suggest to use superclasses whenever possible, which are in these case ranges. I would even think about a single value constraint for these properties. The single statement cancer (Q12078) ICD-10 (P494) "140" i.e. Malignant neoplasm of lip, seems wrong or at least incomplete. Cancer is not restricted to lips. When you have "enough" information about the class 140, I guess you can create a new wikidata object for this specific illness as a subclass of the general cancer. The query for an individual class might not work easily, but anyway it is critical when you want to query whether there is a "corresponding" wikidata object for a given ICD code. Does "corresponding" means idendentical concept, subconcept or superconcept? (Compare maybe also with geographical codes: we don't add the country code for every administrative unit, but we save the information in which country it is in.) --Zuphilip (talk) 13:59, 18 January 2014 (UTC)
I began a class hierarchy of cancers a while ago: http://tools.wmflabs.org/wikidata-todo/tree.html?q=Q12078&rp=279&lang=en. I would be interested in how ICD 10's hierarchy of cancer differs from NCI's. Zuphilip, I think "corresponding" would be ideally be as close to identical as possible, with the realization that two concepts from different vocabularies in a certain domain (like medicine) with often not be equivalent classes. Finding corresponding concepts between different ontologies is known as ontology alignment (a good, more in-depth review is here). Aligning Wikidata's assertions with those of external ontologies is an exercise in that.
I agree that we should not be applying all ICD codes for a given disease to the top concept of that disease. Emw (talk) 14:22, 18 January 2014 (UTC)
(Off-Topic remark to Emw : I guess that the classifications are thesauri. The ontology matching for thesauri seems to be considered far less and the results are - well - improvable, see http://web.informatik.uni-mannheim.de/oaei-library/results/2013/ --Zuphilip (talk) 14:48, 18 January 2014 (UTC) )
Zuphilip, the NCI Thesaurus is one of the biggest biomedical ontologies; see Thesaurus.owl (warning: big file, 230 MB) and http://owl.cs.manchester.ac.uk/research/ncit/. ICD-10 also exists as an ontology; see http://bioportal.bioontology.org/ontologies/ICD10/. Mappings between NCIT and ICD-10 can be found via NCBO BioPortal, e.g. their ontology mapping for breast cancer. Emw (talk) 15:11, 18 January 2014 (UTC)

Why is wikidata so retarded?

moved from Wikidata talk:Main Page.

  • Why can't one edit it like a normal wikipedia page and set links simply by writing [[:en:Article name]] (or maybe with enwiki or something)?
  • Why can't one add comments when editing? This is especially useful when deleting a link which might look like vandalism and so might get or get's retardedly reverted.
  • Why is there no solution when there are e.g. 1 english article and 2 foreign ones, or 2 english articles and 1 foreign one? ["Site link LINK is already used by item ITEM (QNUMBER)."]

-80.133.121.58 09:20, 18 January 2014 (UTC)

  • You can add links even easier than in Wikipedia: just write "enwiki" and add name of the article.
  • It's not possible yet, here is some automatic comments, but you can't add your own comments, however what I know, it's coming in the future.
  • What do you mean exactly?

--Stryn (talk) 09:36, 18 January 2014 (UTC)

I think number 3 refers to the problem that there can be two pages in one wiki and just one in another language. In my opinion Wikidata helps us sort out a problem that is clearly on the Wikipedia-side. Many articles are too long and cover to many concepts. Wikidata helps to bring structure into articles and shows us where merging or splitting makes sense. I sometimes think that people still view Wikipedia-pages as self-contained-essays and don't understand the concept of the hyperlink.
Also in the future the laguagelinks could just take the next item in the "subclass-of-chain" if the same page is not available. Example: If species article is not available just show the language link for the genus, or the family. --Tobias1984 (talk) 10:30, 18 January 2014 (UTC)
Please notice that:
  • Problem 1 is not Wikidata problem. It's just that you are still used to the old way and you can't get used to the new (easier) way. New users would think the opposite. Please notice that editing the old way you can also put wrong links without noticing - and it did happen rather often.
  • Problem 2 should be adressed.
  • Problem 3 is not a new problem. When interwiki links were on Wikipedias the problem was the same, and bots couldn't deal with it. Anyway, you can't work a solution locally - a very complicated one, as it had been before.--Pere prlpz (talk) 11:05, 18 January 2014 (UTC)
Problem 1 is partly an optimization problem and partly a consistency problem. You want to limit the traffic so you don't want to collect data from the individual sites and then send it out again. You will also get consistency problems that way. You want to centralize it to make it consistent.
Problem 2 is actually quite fun. Because the actions are easily identifiable it is easier to spot troublesome edits, but then the patrollers act on false negatives and reverts the edit. If the edit wasn't easy to spot it would lower the number of false negatives and the number of erroneous edits would go down, but at the same time the number of errors kept would increase. You can't troll the automatic summary, it is an objective report. Still as I said a long time ago the automatic summary should be made as tags instead and the summary left for the users to fill in. It would then be a collection of tags that described what was done, and only values or key-value pairs should be left as one or more comments in the summary.
Problem 3 comes from the fact that Wikidata describes entities and Wikipedia is an encyclopedia that more or less by accident describes the same entities, but not quite so not even when each subproject are compared. The individual subprojects describes the entities according to context and cultural conventions and those are different. When a unified approach are used those variations will be visible and something envisioned as one entity in one context or culture might be two or more in another. Those differences will be very difficult to reuse by others, so some normalization must be done. Sometimes entities must be split, sometimes merged. If that creates to much trouble on some projects they can define overridden linking if they want to do so. </rant> Jeblad (talk) 13:49, 18 January 2014 (UTC)

Number (dimensionless) datatype

Where can I find out about the state of the development of the Number (dimensionless) datatype? -- HvW (talk) 11:53, 18 January 2014 (UTC)

See Wikidata:Project_chat/Archive/2013/11#Quantities_datatype_available_for_testing and ask the development team. Snipre (talk) 00:00, 19 January 2014 (UTC)
Thanks. -- HvW (talk) 00:41, 19 January 2014 (UTC)
You should also check out Wikidata:Requests for comment/Dimensions and units for the quantity datatype. The datatype is not nearly ready to be adapted yet and we should centralize the discussion on the RfC page. --Tobias1984 (talk) 17:23, 19 January 2014 (UTC)

Wikidata:WikiProject Informatics

I started the page to get an overview over our information-technology related properties. Anybody interested in scanning a few infoboxes and creating some property proposals? Wikidata:WikiProject Informatics --Tobias1984 (talk) 17:53, 19 January 2014 (UTC)

Wikisource is here!

Hey :)

We have now enabled language links for Wikisource via Wikidata. Let me know if you encounter any issues. In case of questions Wikidata:Wikisource is the place to be.

Let's make this a great home for the Wikisource people. Thanks to everyone who helped prepare this deployment.

Cheers --Lydia Pintscher (WMDE) (talk) 21:04, 14 January 2014 (UTC)

Thanks a lot! Tpt (talk) 21:41, 14 January 2014 (UTC)
Excellent. --Jakob (talk) 21:51, 14 January 2014 (UTC)
Thank you! I started importing categories, authors and templates of Wikisource Amir (talk) 08:57, 15 January 2014 (UTC)
Well, if you do that with DexBot, please stop, at least for Authors. Tpt has developped a specific bot which does that in a specific way. Come discuss it on Wikidata talk:Wikisource first. --LBE (talk) 09:18, 15 January 2014 (UTC)
I have blocked this bot at enWS. There was a clear discussion about the import process, and Dexbot does not align with that discussion. It is flooding RC, removing links, and the WSes had clearly asked for a "hasten slowly" approach with opportunity to review. I don't know why this bot operator has declined to cooperate, and not been involved in the discussion when the opportunity arose, rather annoying.  — billinghurst sDrewth 11:27, 15 January 2014 (UTC)
The bot is halted here too now! -- Lavallen (talk) 11:44, 15 January 2014 (UTC)
I suggest people to read my talk page and decide I'm co-operating or not, you said don't make new items, I said okay. you said don't run it on main namespace, I said I'm already doing it, you said don't run the bot because we have a bot operator ready to do it, and I said it's unacceptable. Amir (talk) 13:57, 15 January 2014 (UTC)
Is it safe to assume that author pages on Wikisource should be connected with their corresponding articles on Wikipedia? So that w:en:Charles Darwin = s:en:Author:Charles Darwin? Jon Harald Søby (talk) 23:42, 19 January 2014 (UTC) Wikidata:Wikisource#Author says yes, so don't mind me. Jon Harald Søby (talk) 23:45, 19 January 2014 (UTC)

What is happening here ?? REALLY !!

Hoi, there is ONE item for one person on Wikidata. It is no option of changing this without a discussion here on Wikidata. GerardM (talk) 10:11, 20 January 2014 (UTC)

Where do you find two items about the same person? -- Lavallen (talk) 10:52, 20 January 2014 (UTC)
a bot was stopped because "it did not align with Wikisource discussion ... It does not argue about anything that is wrong with what it does. The notion that another bot can do it as well, is not relevant. GerardM (talk) 11:20, 20 January 2014 (UTC)
I also don't understand what you are saying. What does a bot task have to do with "one item per person"? Can you post a link? --Tobias1984 (talk) 11:24, 20 January 2014 (UTC)
How is that related to "Hoi, there is ONE item for one person on Wikidata."? -- Lavallen (talk) 11:25, 20 January 2014 (UTC)

Why is a bot stopped from adding links to Wikidata .. there is no obvious discussion and there is not much more to be done that I am aware of than adding one item for one person to Wikisource. What argument could there possibly be ? GerardM (talk) 12:32, 20 January 2014 (UTC)

The bot flooded at least one Wikisource-project, and was therefor blocked temporarly. And there has never been any other discussion than "one item for one person". Nobody has proposed anything else, as far as I know. -- Lavallen (talk) 13:33, 20 January 2014 (UTC)
Well, the Wikisource introduction had been prepared by a group of people over the past few months, agreeing on the proper strategy and enumerating all the particularities that each language project could have. Then, barely 15mn after the enablement announcement was posted, a contributor who had not partipated to the discussions (on the project page at least) or volunteered in any way, launched his bot on the import.
So I asked him to stop for a while, until we could be sure that he followed the proposed strategy. And the fact was, there is a bot which has been developped only for that purpose, so why not use it? Another fact: Dexbot ended up being blocked on enWS because it did not follow the local policies.
Where is the problem here?
--LBE (talk) 13:27, 20 January 2014 (UTC)
I completely agree with you. --Tobias1984 (talk) 16:40, 20 January 2014 (UTC)

Merge or split?

Should we merge Wikipedia:What Wikipedia is not (Q4345841) (includes wikipedia, wikivoyage, commons) and Wikisource:What is Wikisource? (Q15627752) (includes wikisource) and do all projects together and give a description like project principles or should we split every project? --Bigbossfarin (talk) 17:29, 15 January 2014 (UTC)

Looks like there are three versions: "What Whatever is (i.e. project scope)", "What Whatever is not" and "What Whatever is and Whatever is not". Is it an alternative to divide by these three versions instead of what Project it is related to? -- Lavallen (talk) 18:30, 15 January 2014 (UTC)
They are NOT about the same thing. They should be split. GerardM (talk) 10:07, 20 January 2014 (UTC)

Watchlist

Hi, I know that we can follow the edits of wikidata items in the wikipedia watchlist but I don't remember how I can activate this feature. Can someone help me ? Thanks Snipre (talk) 13:17, 20 January 2014 (UTC)

Go to w:Special:Preferences#mw-prefsection-watchlist and check "Show Wikidata edits in your watchlist". (If you have "Group changes by page in recent changes and watchlist" in w:Special:Preferences#mw-prefsection-rc checked, you'll need to uncheck that, too.) --Yair rand (talk) 13:28, 20 January 2014 (UTC)

List of lonely interwiki links

Hi! I made a tool for generating lists of Wikidata items that link to only one language version of a Wikipedia page. This helps to find Wikipedia pages where the corresponding pages in other languages actually exist, but are not linked to the page. It is available at:

Regards, Sanyi4 (talk) 04:23, 12 January 2014 (UTC)

Fine, but I find one problem. I started to work from the page about Sámegiella Wikipedia (sewiki). That language contains many non-ascii-letters, and the links in the page cannot handle them correctly. -- Lavallen (talk) 11:30, 12 January 2014 (UTC)
I checked alswiki, and besides the problem Lavallen mentioned, I found two other ones. First, some items I've checked have already been deleted mid-December, so you're probably using a dump being at least about one month old. Of course that's better than nothing (thank you very much for the tool, by the way), but more up-to-date data would be better. And then the results were doubled. The tool claims to have found 1408 pages, but number 706 is the first one again, number 707 the second one, and so one. So the tool only finds about 700 pages and just lists them twice each (there were no duplicates for pages 704 and 705, though). --YMS (talk) 12:33, 12 January 2014 (UTC)
It is written "Current data is based on the dump of 5 December 2013" and for the character encoding you may have to change it by hand in your browser, e.g. in Firefox: Ansicht (View?) -> Zeichenkodierung (character encoding?) -> UTF-8. --Zuphilip (talk) 14:04, 12 January 2014 (UTC)
Argh, I read the dump sentence, but apparently I stopped at some point ;). So never mind this one. Sorry. --YMS (talk) 14:41, 12 January 2014 (UTC)
Had a quick look. This looks really good, well done! :D Del♉sion23 (talk) 18:10, 12 January 2014 (UTC)

In labs there is a copy of Wikidata that is always up to date. Why not have the queries run on this one ? Thanks, GerardM (talk) 18:25, 12 January 2014 (UTC)

Thanks for your comments! I'm aware of the problems you mentioned, but I couldn't resolve them by now. Character encoding problems may be the webhosting service's fault, but I try to fix them. The database will be updated soon, and I'm planning to update it regularly. @GerardM: What are you thinking of exactly? -- Sanyi4 (talk) 20:42, 12 January 2014 (UTC)

Your character encoding is set to "iso-8859-2", that's why. When you generate your HTML page(s), instead of
  <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-2">
use
  <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
That should fix the problem. :) -- Edinwiki (talk) 08:45, 13 January 2014 (UTC)

I use UTF-8 in the code, but I'm presuming the default encoding is set by the web hosting provider. -- Sanyi4 (talk) 18:26, 13 January 2014 (UTC)

Sanyi4, have a look at line 4 of your lonelylinks.php. --Zuphilip (talk) 18:56, 13 January 2014 (UTC)
try to add: ini_set('default_charset', 'utf-8'); at the script beginning --Rippitippi (talk) 03:49, 14 January 2014 (UTC)

The character encoding in your html was a problem in any case. But it now seems that you also have to configure php to output the data in UTF-8. If the suggestion of Rippitippi doesn't work, then you can also try adding this at the top: 

  header('Content-Type: text/html; charset=UTF-8');

-- Edinwiki (talk) 14:41, 14 January 2014 (UTC)

@Sanyi4: I think GerardM was suggesting to host your tool on Wikimedia Tool Labs, allowing access to real-time data. --Ricordisamoa 20:09, 14 January 2014 (UTC)

But overall: great to have an easy list of merge-candidates! It would indeed be even better if it was hosted on Wikimedia Tool Labs to have always up to date lists. Lymantria (talk) 14:02, 15 January 2014 (UTC)

Another tool

Hi, I created a similiar tool on labs that uses the data provided in the database so it is always up to date. See toollabs:bene/lonelyitems. Best regards, -- Bene* talk 19:33, 15 January 2014 (UTC)

Did you notice the button says "Serach"? --Tobias1984 (talk) 19:38, 15 January 2014 (UTC)
Oh, thanks, didn't notice. -- Bene* talk 19:44, 15 January 2014 (UTC)
I'm on Chrome and can't seem to get it to work. Can you give me an example of what to put into "Wiki"? e.g. ang, ang.wikipedia, ang.wikipedia.org? Thanks. Del♉sion23 (talk) 22:00, 15 January 2014 (UTC)
I got it working :) Thanks very much for the tool Bene*! Del♉sion23 (talk) 22:00, 15 January 2014 (UTC)

Very good! I'm happy that my idea is so inspiring! Although the always-up-to-date feature is very nice, this list generator is extremly slow (at least for me). I think the two tools would well complement each other, so I keep on trying to resolve the technical problems of my version. -- Sanyi4 (talk) 21:27, 20 January 2014 (UTC)

Yes, you are right. Unfortunately the SQL queries are quite slow for larger wikis, perhaps because of a lack of indices. This is a big advantage of your tool. ;-) Best regards, -- Bene* talk 21:49, 20 January 2014 (UTC)

Bit confused

Sorry for potential dumb questions, but here it goes. I have never contributed at Wikidata, but how come I have contributions under my account here? Is it somehow linked to other projects? Because these looks like what I did at Commons (where I am an admin). I can't edit the template pages here.

Also, I work on a lot of power station and dam articles at en.wiki. I would like to help import the coordinates of those here, but I cant seem to figure out how. I score my wikidata knowledge to about 3/10. Can anyone help, please? Thanks! Rehman 03:30, 18 January 2014 (UTC)

  1. When you move a page, Wikidata will automatically update. This edit is under your account.
  2. A lot of template is imported from English Wikipedia, Wikimedia Commons or Meta-Wiki. Your contributions include these imported edit though you make edits in other wikis instead of Wikidata.

--GZWDer (talk) 04:25, 18 January 2014 (UTC)

"Your contributions include these imported edits" is not always true, because someone can have account with the same user name. --Stryn (talk) 08:03, 18 January 2014 (UTC)
Perhaps it is a bit strange that you have contributed under the CC-0 license, without being aware of that and even before wikidata actually started. Lymantria (talk) 12:29, 18 January 2014 (UTC)
Contributions in template namespace are not under CC-0, otherwise imports from other projects would often have been copyvios. -- Lavallen (talk) 16:15, 18 January 2014 (UTC)

Hmm ok, weird. As long as it's some feature and not a glitch, I'm fine with it. Can anyone guide me on importing coordinates? Rehman 00:31, 19 January 2014 (UTC)

Hi Rehman (hey, remember me ?) As far as I know the coordinates are automatically imported to Wikidata from Wikipedia by bots, but if for some reason a coordinate couldn't be imported (for instance if this information is not properly formated in the original article), you can manually edit the statement on a Wikidata page or add a new one with the property name "coordinate location", like on this page. Cheers ! Nonox (talk) 12:43, 21 January 2014 (UTC)

trouble using "Add links" from enwiki

On the English Wikipedia, on an article page under Languages, I tried to use the "Add links" link, and a message box came up, saying "You need to be logged in on this wiki and in the central data repository to use this feature." The message was unhelpful, because I was already logged in to both sites. The box had a link to log in to wikidata, but didn't provide an obvious means to proceed after that. Rybec (talk) 23:52, 21 January 2014 (UTC)

The same thing happens on zhwiki. Rybec (talk) 05:02, 22 January 2014 (UTC)

development plan for 2014+

Hey folks :)

I have just published our plan for Wikidata development for 2014+. You can find it at Wikidata:Development plan. Some of the things on there have already been started. If you have any comments or questions please leave them on the talk page there. Adam and I will also be holding an office hour soon that will also give you a chance to ask questions about the plan. I will post another message about that in a bit.

Let's make Wikidata rock even more than it already does. There is a lot left to do. If you want to help out please ping me and let me know which area you are interested in.

Cheers --Lydia Pintscher (WMDE) (talk) 12:18, 22 January 2014 (UTC)

next office hour

Hey folks :)

Adam and I will be holding another office hour on February 3rd at 5pm UTC. See here for your timezone. We will be meeting in #wikimedia-office on freenode IRC. The topics will be:

  • short update on the current state of Wikidata
  • questions about the development plan for 2014+ (Wikidata:Development plan)
  • any other questions/discussion

We will make logs available for those who can't attend. Hope to see many of you there!

Cheers --Lydia Pintscher (WMDE) (talk) 13:22, 22 January 2014 (UTC)

Mara Wilson (Q237270)

Can some-one check how and why a date of death was added to Mara Wilson (Q237270). She is alive --Racklever (talk) 16:11, 22 January 2014 (UTC)

Vandalism in Spanish Wikipedia that went unnoticed for more than two months and so was present when the bot visited the article to import infobox data to Wikidata. --YMS (talk) 16:50, 22 January 2014 (UTC)

Member states

Hello,

Why does Q30 contain a list of member states, and Q458 or Q1065 not? Does this have something to do that phase 3 hasn't been launched yet?

Thank you for answering. Muelleum (talk) 00:14, 22 January 2014 (UTC)

Muelleum, there is no reason I'm aware of, other than noone has added them yet. Emw (talk) 00:36, 22 January 2014 (UTC)
@Emw: No, you're wrong. This is because of Wikidata:Requests_for_deletions/Archive/2013/Properties/2#(no label) (P100).--GZWDer (talk) 07:24, 22 January 2014 (UTC)
  • @Emw: Wrong again? I read "list of member states" which I guess was exactly the property P100 mentioned above. - However, if you would delete "administrative division" in the labels and descriptions of P132 (P132) as well as all of the 6 current constraints at Property_talk:P150 (for example it has to be inside a country), I guess you could use as you suggest. But what would this be good for? Note also there exists an (almost) generic property: has part(s) (P527). --Zuphilip (talk) 20:10, 22 January 2014 (UTC)
Zuphilip, note how Muelleum uses the present tense and asks "Why does Q30 contain a list of member states, and Q458 or Q1065 not?" If you look at United States of America (Q30), you'll see that the property that lists the "member states" of the United States is contains the administrative territorial entity (P150); that property is also not used in any statements in European Union (Q458) or United Nations (Q1065). So I think Muelleum is asking about why P150 is not used, not why P100 is not used.
Thanks for pointing out the constraints on P150, though. Interestingly, United States of America (Q30) violates at least the constraint the the subject of a P150 claim must have a 'type of administrative division' (P132) claim -- Q30 has always violated that P150 constraint, yet has plenty of P150 claims. So it seems the constraints aren't enforced stringently. The 'country' constraint seems somewhat superficial to me, but if there is a reason to not apply P150 to European Union (Q458) and United Nations (Q1065), that's probably it.
If users did prefer a generic property for this type of relation, I think has part(s) (P527) would be reasonable. Emw (talk) 01:57, 23 January 2014 (UTC)
Emw: If you think that the constraints on P150 are wrong, then please start a discussion there how improve them. If you think that there are some "real" constraint violations on certain object, then please try to correct them. If you don't understand how constraint violations works, then please ask. --Zuphilip (talk) 19:38, 23 January 2014 (UTC)
If I would use has part(s) (P527) for European Union (Q458), I would prefer to list a set of treaties, not a set of nations. From my point of view EU is an organisation, not a geographic entity. -- Lavallen (talk) 19:55, 23 January 2014 (UTC)
@Lavallen: What speaks against foundational text (P457) for the treaties? Muelleum (talk) 22:41, 23 January 2014 (UTC)
From what I can see, France (Q142) is an instance of member state of the European Union (Q185441), which has European Union (Q458) as part of (P361). Even if it may be a disputed point whether the EU is a geographic entity or not, United Nations Security Council (Q37470) seems to be no geographic entity, but still has member states. Is there a property for plain simple membership for states, just as membership in a bowling crew can be modeled through member of sports team (P54)? If not, what speaks against creating such a property, what for? I get the difference between a state consisting of sub-states and an international organisation consisting of sovereign states. I also see similarities between people signing a membership paper for a bowling crew, and states signing a membership treaty for an international agreement. Muelleum (talk) 23:20, 23 January 2014 (UTC)
Muelleum, member of (P463) would probably fit this use case. It was created essentially as a non-transitive version of part of (P361). Emw (talk) 05:02, 24 January 2014 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ On second thought, member of (P463) might not be best, at least not until its meaning is updated. The property's creation discussion (Wikidata:Property_proposal/Archive/6#P463) had different users supporting different semantics for the property, and the meaning that we ended up with is not what a fair number of them supported. For example, they supported something not limited to people, but 'member of' currently lists 'people' in its domain. We should probably flesh out some better semantics of 'part of', and consider different flavors of that fundamental relation as discussed in http://www.w3.org/2001/sw/BestPractices/OEP/SimplePartWhole/#Background. That section summarizes several different varieties of 'part of':

  • Containment: we have no generic 'contained in' or 'located in' property, only domain-specific versions of it (see here)
  • Membership: this is probably what member of (P463) should be. Several supporters of that property argued for this meaning, but that never got reflect in P463. My guess is that it would be fine to update the meaning of the property to accommodate this.
  • Connections and branches: I don't think we have a generic property to fulfill this relation. See linked document above for explanation.
  • Constituents: see linked document for details
  • subClassOf: this exists as subclass of (P279). (It's interesting to see this property considered as a "flavor" of 'part of'. I usually see the two properties described orthogonally.)

A Taxonomy of Part-Whole Relations seems to be an influential publication on this subject, for anyone really interested in it. Emw (talk) 05:39, 24 January 2014 (UTC)

Strange ages

Hi, I got a query for Bith and Dead these people's age (dead year- birth year) data may be incorrect (low age).

  • Q708055 age=0 corrected
  • Q83171 age=0 corrected by EugeneZelenko
  • Q982880 age=0 corrected
  • Q113775 age=0 corrected
  • Q49752 age=0 death date corrected, birth date still problematic, see note about birth date in fr.wp
  • Q181995 age=0 corrected
  • Q8404 age=0 is correct
  • Q82006 age=0 corrected
  • Q403564 age=2 corrected, importer errorneously imported time as head of state
  • Q53662 age=3 is correct, it's a dog
  • Q332958 age=4 is correct
  • Q1060353 age=5 is correct, it's a dog
  • Q86932 age=6 is correct
  • Q171433 age=7 is correct, it's a sheep
  • Q37089 age=7 is correct
  • Q191707 age=7 is correct
  • Q467662 age=8 is correct
  • Q281336 age=8 is correct
  • Q46840 age=10 is correct

Yamaha5 (talk) 09:21, 23 January 2014 (UTC)

I checked them all (and stroke them in the list) and many of them are just children (mostly kings) who died early or animals. Most real errors are due to import errors from en.wp. --Slomox (talk) 10:18, 23 January 2014 (UTC)

Merge direction, deleting item that is used

Deleted was Q6571331 - by User:Zolo and kept was Q9708251 - but why not merging into the lower Q number? And why was the deleted one left here: https://www.wikidata.org/w/index.php?title=Q6465&diff=prev&oldid=104508628 ? Androoox (talk) 19:14, 23 January 2014 (UTC)

Why are not all roses red?
Have you tried to ask Zolo him/herself? -- Lavallen (talk) 20:15, 23 January 2014 (UTC)

Prevent nonsense by using more constraints

I looked at https://www.wikidata.org/w/index.php?title=Q178712&diff=78786488&oldid=78579422 and thought User:Rippitippi made the state and territories of Australia a subclass of 2014 street riots in Ukraine. But no, in October 2013 this Q was not street riots, but administrative division of Australia. It got changed by User:Misha motsniy [8]. But the Italian label and maybe several others stayed at something like "divisioni amministrative australiane". OK, now that I wrote this, I see User:Arctic.gnome just reverted.

Could this be prevented by more constraints? Androoox (talk) 19:48, 23 January 2014 (UTC)

It looks like in October 2013, administrative territorial entity of Australia (Q4494320) was part of a batch of 81 deletion requests (see here) based on the incorrect assumption that it was the same thing as state or territory of Australia (Q178712). After all of -320's links were merged into -712, rather than -320 being deleted it was re-purposed to be an item about the Ukraine street riots. --Arctic.gnome (talk) 20:02, 23 January 2014 (UTC)

Category and template

There are 69 items which are marked as both category and template: http://208.80.153.172/wdq/?q=claim[31:4167836]_AND_claim[31:11266439] Do I right understand that we don't want to merge category and template about the same topic? --Infovarius (talk) 11:01, 17 January 2014 (UTC)

I fixed a few of them; it looks to me just like bad interwikis that need to be cleaned up. It's about 50-50 whether there needs to be a new item created to fix them or not. --Izno (talk) 05:18, 18 January 2014 (UTC)
I suppose that nearly all needs to be divided. Infovarius (talk) 11:15, 20 January 2014 (UTC)
  Done I've gone through the last 45 or so and fixed them. Some were because of incorrect interwiki links on Wikipedia. Most were due to people mistakenly thinking they could read Russian. Del♉sion23 (talk) 22:47, 24 January 2014 (UTC)

Old Wikidata-like proposals

I just stumbled upon a few project proposals from Meta-Wiki that resembled Wikidata to me, especially:

  • EncycloPDia: a public domain encyclopedia and computational knowledge engine – proposed by Burstling Miraculous Power
    Wikidata is actually under CC-0, and can be used by AI software such as Wiri
  • Meta World's Data – proposed by MetaConceptualArt
    Wikidata is actually made by bots
  • Slotipedia: a cross between an Encyclopedia, and a knowledgebase for some kind of AI application – proposed by SJK
    as above, see Wiri
  • Wikibot: a project proposal for making wiki content and meta content available via queries from external sites – proposed by Disanpoter
    very detailed description of a Wikidata-like structured syntax
  • WikiDatabank: a free databank which collects data all over the world – proposed by Πrate
  • Wikistatistics – proposed by Gnuser

I think we (as a whole community) should say a thank to all users who predicted in some way the development of our project. I would suggest designing a special barnstar for them, maybe with a formal invite to contribute to Wikidata – even if they haven't been active recently. --Ricordisamoa 00:22, 24 January 2014 (UTC)

Ricordi, thanks for noting those proposals. I'd certainly support a formal invitation to those users. Emw (talk) 02:28, 24 January 2014 (UTC)
Great ideas are usually the product of their time. Just look at the list of people that discovered evolution. Just like Darwin the Wikidata community will manage to get most of the credit. --Tobias1984 (talk) 08:43, 24 January 2014 (UTC)

Gadget/bots to standardize statements order

Are there gadgets or bots which could standardize statements order? For example, place start date before end date in qualifiers; or combine place/date of birth/death/burial in one block? Similar standardizations could be made for other kinds of items. --EugeneZelenko (talk) 16:10, 24 January 2014 (UTC)

See StatementSort (source). --Ricordisamoa 18:32, 24 January 2014 (UTC)
@Ricordisamoa: That's only for personal sorting. Eugene may be talking about the embedded statement sorting, which can be changed for everyone. --Izno (talk) 00:12, 25 January 2014 (UTC)
@Izno: honestly, I don't know much of how WikibaseRepo sorts statements in the GUI: please ask in WD:DEV. --Ricordisamoa 02:19, 25 January 2014 (UTC)

P31 for animals

We use "instnace of human". Human appears to be synonymous with homo sapiens, which is a species (Q7432). Can we decide that all individual animals should have p31 with the name of their species ? --Zolo (talk) 19:17, 17 January 2014 (UTC)

Individuals like en:List of individual dogs should use "instance of" (in this example instance of = dog) all other items should use "subclass of". At least I think thats how those two work. --Tobias1984 (talk) 19:39, 17 January 2014 (UTC)
think dog breeds is the same and use instance of dog breed (Q39367) --Rippitippi (talk) 22:33, 17 January 2014 (UTC)
Yes, what I meant, is : can we decide that the p31 value to use is the species, and so not use p31: labrador or p31: monkey ? --Zolo (talk) 09:46, 18 January 2014 (UTC)
P31: labrador and P31: monkey are very two different things with regard to classification. "Monkey" is a class for members of the simian infraorder, excluding apes. It is much broader than a species. "Labrador" a class that is more specific than a species. "Dog" is very ambiguous -- it could apply to a Chihuahua or a Great Dane. Dog breeds have useful properties like "weight", "height", "lifespan"; see Template:Infobox_Dogbreed. This seems to be true for breeds of animals generally. Given those facts, I am inclined to support classifying individual animals with their breed when known. That said, classifying animals with their species (or subspecies, e.g. dog) when more specific information is not available seems reasonable. Emw (talk) 13:09, 18 January 2014 (UTC)
I believe that we should point as precise class as possible. E.g. dog breeds for individual dogs, species or at least genus for monkeys. Of course when it is known. --Infovarius (talk) 14:41, 18 January 2014 (UTC)
I agree with this. Emw (talk) 14:46, 18 January 2014 (UTC)

I have a slighly different point of view. Here we are mixing too different kind of classifications : the average human classification would class dog by breeds, but this may not be that pertinent in scientific classifications. For sure a reasoner would infer that the breed is a subclass of the species in almost all cases, but what in the cases where the two different classifications are non consistent ? I think it may be a good practice in Wikidata to take into account the different classification, and to adapt the most specific class rule with most specific class in any relevant classification. TomT0m (talk) 16:07, 18 January 2014 (UTC)
Breeds might not be in the province of conventional Linnaean taxonomy, but they are pertinent in scientific classification and they are all subsumed by the species classification. Dog is a model organism for studying mammalian phenotypic diversity, and breeds are the taxonomic units that are used to classify dogs in scientific research. See for example A Simple Genetic Architecture Underlies Morphological Variation in Dogs, which used dog breeds as the unit of classification to study the genetic causes of variations in body size and shape.
All instances of a dog breed -- a particular instance 'Fido' of the class 'Chihuahua', for example -- are also instances of the broader class 'dog'. This means Chihuahua is a subclass of dog; species can be inferred from breed. The classification system of breeds is still biological, widely used, rather standardized, and very useful. Thus I think 'Fido instance of Chihuahua' would be significantly better than 'Fido instance of dog, Fido dog breed Chihuahua'. Emw (talk) 21:28, 18 January 2014 (UTC)
I'm not talking of Linnanean Taxonomy, rather of genetic classification. en:Dog_breed for example says the traditional dog breeds are something else than scientific genetic classification (even if breeds can be studied genetically, of course). That's why I would think both information are useful as classes, even if we can deduce safely one from the other. We got less dependancy to the powerfulness of the reasoning and query engine, which can be good as people might promiraly be interested in both depending of what they want to do. I quote The recognition of distinct dog breeds is not maintained by a scientific organization; they are maintained by a number of independent kennel clubs that need not apply to scientific standards and are often inconsistent. For instance, the Belgian Shepherd Dog is separated into four distinct breeds by some clubs, but not in others TomT0m (talk) 21:46, 18 January 2014 (UTC)
TomT0m, none of the assertions I see in the Dog breed article about breed not being a scientific classification have citations. The fact that some organizations have inconsistent standards for breeds does not somehow make them unscientific. There are several different standards for plant taxonomy -- does that make plant taxonomy unscientific? No. The fact that genetic research is reorganizing clusters of dog breeds also does not entail that the notion of a dog breed is unscientific. Genetic research reorganizes taxonomy above the species rank, but that doesn't imply that the notion of species is unscientific. And as noted in my previous comment, dog breeds are widely used in scientific research on dogs; this is prima facie evidence that breeds are a scientific classification. Consider another example: a 2010 paper in the scientific journal Nature used 81 dog breeds defined by the American Kennel Club as a basis for their research (see the first paragraph of Supplementary Table 1). If using classifications of dogs by breed in Nature doesn't mean breed is a scientific classification, then I don't know what would. So the notion that dog breeds are "unscientific" seems unsubstantiated.
Beyond being scientific classifications, dog breeds are a practical way to group dogs for knowledge representation. Knowing that a dog is a Great Dane immediately entails that values for the important properties of height and weight will be in a certain range, a very different range than if the subject were a Chihuahua. And I don't see how using species instead of breed decreases reliance on query engines. If we want a naive mapping to infoboxes -- which in my opinion would be myopic -- then we should use 'instance of animal' for all individual animals, then have separate statements for 'species' and 'breed'. Using 'instance of dog' for individual dogs already relies on query engines. Using 'instance of Great Dane' requires no more functionality in a query engine than 'instance of dog'. The former, however, gives us a more informative and specific (and still scientific) classification. Emw (talk) 17:06, 19 January 2014 (UTC)
There is a misunderstanding, I don't think we should use one instead of the other, I think we should use both. If only for the fact that there is several standards for dog breed. Of course Dog Breeds are interesting for genetic studies are they are probably a very good example of human selections and breeding, this does not imply the criteria used for breeds are relevant for phologenetic classification for example. So both informations are interesting. This is related to class tagging : this individual is classed as a wolf subspecie according to this classification, and and as a member of dog breed A according to American dog breed association, and as dog breed B in the British dog breed association. A user might want to know all of its classes, or just one, we don't know that and we don't have to make that choice for him. All of these classes are interesting, and even if there may be some kind of redundancy they do not say exactly the same things and classifications do not use the same criteria. TomT0m (talk) 17:22, 19 January 2014 (UTC)
TomT0m, dog breeds are very relevant for phylogenetic classification, in addition to other kinds of genetic studies. They are a case study of interesting phylogenetics, and research that considers dog breeds from a phylogenetic perspective is a regular topic of scientific journal articles and articles in more accessible publications for the science-interested public. I provide links to examples in my previous posts. Dog breeds are scientific classifications, interesting from phylogenetic and other perspectives.
There is some inconsistency in dog breed classifications, but all of the standards bodies certainly hold that if Fido is an instance of a Chihuahua, then Fido is also an instance of a dog. In other words, noone asserts that there are dog breeds that are not kinds of dog. For any 'X instance of dog breed', the axiom 'X subclass of dog' holds. Thus if 'Fido instance of X', then 'Fido instance of dog' -- regardless of which standards body is 'X' is defined as a dog breed by. So 'Fido instance of dog' is redundant with 'Fido instance of $dog_breed'.
I am willing to hypothesize that the inconsistency between dog breed standards bodies is not drastic, and not significant enough for us to avoid using dog breeds as a conventional P31 value where that information is known. But we don't have data on how much inconsistency there is. Having properties to capture information on how standards bodies like FCI, AKC, etc. classify dog breeds would help answer this. The 'Classification and standards' section in the infobox in Great Dane gives an example. I'll propose these properties soon, unless someone else wants to. Emw (talk) 18:55, 19 January 2014 (UTC)
I know all that, we won't repeat ad vitam eternam the same trivial arguments :) Nonetheless what is important is not whether or not dog breeds are studied, it's the conclusions of the studies. Until now, nothing proves that the scientific conclusions are totally consistent with the dog breeds organisation criteria. TomT0m (talk) 19:32, 19 January 2014 (UTC)
We have a Wikidata:Dog breeds task force and a Wikidata:Cat breeds task force, and Emw breeds are usually out of scope of „scientific classification”. --Succu (talk) 21:44, 18 January 2014 (UTC)
Succu, dog breeds might not be deemed taxons by ITIS, but they are units of scientific classification that are essentially taxonomic. I provided an example of a highly-cited scientific paper that uses dog breeds as a unit of scientific classification. Surely, there are many more scientific papers that treat breeds as units of scientific classification. I don't see what the argument is against classifying individual dogs as instances of a particular dog breed rather than the subspecies 'dog'. It is quite standardized, widely used in science, and matches colloquial use -- all indicating classification by dog breed is preferable. Emw (talk) 22:02, 18 January 2014 (UTC)
Thanks for citing some „highly-cited scientific paper”... BTW: Mind to read my wording? --Succu (talk) 22:15, 18 January 2014 (UTC)
Please explain what you mean. Emw (talk) 22:22, 18 January 2014 (UTC)
Wikidata:Dog breeds task force and Wikidata:Cat breeds task force are not really active, but I see no notification of yours. --Succu (talk) 22:32, 18 January 2014 (UTC)
it is a pity that the projects are abandoned this can be a good way to use Wikidata on wikipedia, use this data do not require script very complicated, is useful for box and list, at the time Wikidata is practically not used in wikimedia projects links excluded. Think we must complete it with the required properties and present it to wikipedia it could be a startig point --Rippitippi (talk) 02:47, 19 January 2014 (UTC)
Looks like a good idea to close this thread. I read it twice now and have no clue what this is about. Apparently users have some concepts in their mind and just paste words onto their thoughts without bothering to read back what they are actually saying. Just trying to guess makes my head hurt. - Brya (talk) 17:21, 19 January 2014 (UTC)
The dismissive and rude language in the comment above is not helpful. Perhaps you could ask for clarification or explanation, or explain why you think one of the assertions made above is nonsensical or incorrect. Emw (talk) 17:45, 19 January 2014 (UTC)
My, you are oversensitive (guilty conscience?). The first few posts are understandable, although the fifth goes on at length to say remarkably little (in an invitation to go off-topic?). After that it goes off into an entirely different dimension altogether. Asking for clarifications would obviously be counterproductive. - Brya (talk) 06:19, 20 January 2014 (UTC)
No, the language you used really was dismissive and rude. Suggesting that we stop the discussion because you think "users have some concepts in their mind and just paste words onto their thoughts without bothering to read back what they are actually saying" is clearly not helpful.
It's good to know which posts you find confusing, though. Simply put, the fifth post is proposing to classify non-human animals with the most specific biological class -- breed -- where known. Posts after that go back and forth on whether species or breed is a better P31 value for such items, all of which is on topic for a discussion labeled "P31 values for animals". Emw (talk) 23:06, 21 January 2014 (UTC)
So, now you have shown that you are not only careless in your phrasing, but also in how you read. What does this get you? - Brya (talk) 06:31, 23 January 2014 (UTC)
Vague insults are also not helpful. Please at least explain what you mean. Emw (talk) 13:59, 23 January 2014 (UTC)
Interesting thread, and I regret to say that this is far away from my usual areas of activity. I just want to throw this into the discussion, as I feel you are to reach a consensus on Fido instance of dog breed = Fido instance of dog: this is true in all cases I know of, and it is quiet improbable to find an individual purebred dog that does not fit into this scheme. But, as far as I know, there is at least one recognized cat breed which is a hybrid of two felid species, a common house cat and a species of wild cat. And, second example, the barbary duck breed is a hybrid of two genera. Hence you could run into trouble with a system that relies on a vertical hierarchy. On the other hand, I don't know of any individually named cats/ducks of these breeds, and it's yours to decide whether you must prevent any possible inconsistency. --Cimbail (talk) 21:16, 19 January 2014 (UTC) (noreply, rarely on Wikidata, sorry)
In my opinion we can use P31 with the breed for thoroughbred dogs. Similarly for other lifeforms where there is a formal way of registering and recording sub-species/breeds/races/varieties, including most domesticated animals and plants (and micro-organisms - bread yeasts?). These are sub-species however so dogs of different breeds can breed with each other. Where an individual dog is the result of such a cross breed then we need a convention for describing this cross. For mongrels where the breed of the parents is unknown then instance of:dog is appropriate until that mongrel is declared to be a type specimen for a new breed.
For humans the whole concept of assigning mongrel humans to various races is completely discredited and I would avoid using it. OK? Filceolaire (talk) 15:54, 20 January 2014 (UTC)
  OK --Rippitippi (talk) 17:05, 21 January 2014 (UTC)
@Filceolaire: Unfortunately your last mentioned problem has found his way to wikidata: ethnic group (P172) with values like Caucasian race (Q7129609) (= en:Caucasian race). --Succu (talk) 19:05, 21 January 2014 (UTC)

So, to recap, I glean these questions out of the above thread:

1. Should instance of (P31) be used to specify the species of an individual non-human animal, say, Koko (Q1348219) the gorilla? I would say, "yes". In this example, there is indeed a statement Koko (Q1348219) → instance of (P31) → Gorilla (Q36611)

2. Likewise, should instance of (P31) be used to specify the breed of an individual dog, say, Lassie (Q941640)? Again, my vote would be "yes". Again, there is already a statement Lassie (Q941640) → instance of (P31) → collie (Q1196071)

3. Are there other properties currently being used for these relationships? If so, is that a problem, and should they be nominated for deletion? Or, at least, annotated in the (well hidden) documentation that they shouldn't be used for these purposes?

4. In the case of individual dogs, would it be sufficient to only specify a dog breed, or would it be better to specify both the dog breed and the species (dog (Q144))? In the case of Lassie (Q941640), both are given. I would vote that that is as it should be: specifying both is better. Though it might be redundant, if you wanted to only specify dog breed, then it would be necessary to state, for each breed, that it is a subclass of dog (Q144). Note that this statement doesn't exist for collie (Q1196071), and I think it would be undesirable. I am not an expert, but a quick check of NCBI's taxonomy database, for example, shows that they do not have records for dog breeds. Klortho (talk) 20:17, 25 January 2014 (UTC)

Klortho, there seems to be no disagreement that the answer to (1) is "yes", at least where information on breed isn't known. The answer to first question in (3) is "no", which entails answers for the subsequent two questions in (3).
Regarding (4), I think in cases where the information is known, it would be sufficient to specify only the dog breed. Dog breeds are, by definition, subclasses of dog. All instances of Collie are instances of dog, etc. The fact that dog breeds are not captured by biological taxonomy databases like NCBI's or ITIS does not negate that fact; it simply indicates that breeds for the taxon have not been an area of focus for those entities. In terms of knowledge representation, specifying both species and dog breed with P31 is like specifying both kingdom and species with P31: it is redundant and not good practice.
As I point out to TomT0m in a discussion above, dogs are commonly classified by breed in Nature-caliber scientific literature; thus, dog breeds are scientific classifications. Levels of biological grouping below species or subspecies is often not covered by taxonomic databases, but that does not mean the grouping is not essentially taxonomic. Dog breed is a classification widely used in both scientific literature and the lay public, so I think it would be reasonable to dog breed as a P31 value for dogs. Emw (talk) 21:25, 25 January 2014 (UTC)
That's quite a big step to do. The fact it's used in scientific publications does not imply it's a taxonomy and a relevant taxonomy. If I say they are used because their parenthood are traceable it's probaly as relevant. What would prove your point is a paper related to taxonomy or a scientific dog taxonomy database. TomT0m (talk) 22:09, 25 January 2014 (UTC)
TomT0m, last weekend you began by suggesting that dog breeds are not a scientific classification, then when I rebutted that by showing that dogs breeds are used as a type of scientific classification in highly-cited scientific publications, you moved the goalpost to claim that dog breeds are not relevant phylogenetic classifications. I then explained how dog breeds are indeed relevant phylogenetic classifications, again with a link to a scientific journal article supporting the claim. Those detailed explanations were followed up with a dismissive response that "I know all that" -- which seemed odd, since I had just spent several paragraphs contradicting your suggestions that dog breeds aren't scientific or phylogenetic. The fact remains: dog breeds are clearly relevant scientific and phylogenetic classifications of members of the class Canis lupus familiaris.
I do not understand the motivation of these attempts to semantically wiggle around the fact that dog breeds are clearly relevant classifications of dogs as it concerns Wikidata. Let's assume for a moment that dog breeds are scientific and phylogenetic classifications but not "taxonomic" classification. Then what does taxonomic even mean? Is a class not taxonomic unless it is in a database with other taxonomic classes? That seems superficial and inadequate as a defining feature. Taxonomy is predicated upon classifying life based on salient morphology and phylogeny. Dog breeds satisfy both of those criteria.
As explained previously, the papers A Simple Genetic Architecture Underlies Morphological Variation in Dogs and Genome-wide SNP and haplotype analyses reveal a rich history underlying dog domestication show that dog breeds are scientific and phylogenetic subclasses of dog (Q144). The latter, published in Nature, used 81 dog breeds defined by the American Kennel Club as a basis for their research: see the first paragraph of Supplementary Table 1.
Furthermore, breed is used as a taxonomic classification by the US Department of Agriculture, as shown in their 2010 Standardized Taxonomy Usage Recommendations. This is maintained by the National Animal Health Laboratory Network, which bases its classifications on the SNOMED and LOINC data standards. That standard does not include dog breeds, but it includes breeds of several common animals and demonstrates that breed is supported as a level of taxonomic classification by relevant reliable sources.
So we see that breeds are not only relevant in their scientific and phylogenetic nature, but also supported as a level of taxonomic classification by relevant institutions. Given that, dog breed seems like a very relevant value for P31 in dogs. Emw (talk) 00:08, 26 January 2014 (UTC)
Emw, that all sounds good to me. So, the recommended best practice would be to make individual dogs instance of (P31) their breed only, and have the breed item be subclass of (P279) → dog (Q144). Klortho (talk) 03:07, 26 January 2014 (UTC)
Klortho, yes. When breed is not known, I think the recommended practice should be to omit breed and use dog (Q144) as the P31 value. Emw (talk) 14:39, 26 January 2014 (UTC)
When breed is not known A lot of dogs, I think of crossing outside of the control of dog breeds association simply cannot have an official breed and are just bastard, aren't they ? Anyway, you are probably right I still want to emphasise we should be careful about which classification we use and how we can and will use them. Thre is no harm with a little redundancy, especially if it can help someone with his usecase. TomT0m (talk) 15:14, 26 January 2014 (UTC)

have a Score variable for physical qualifying the articles

We can have a Score variable for each local wiki's article to show is that article has physical qualified! I mean if the article should have these items:

  • More than one References
  • Internal links
  • Images
  • Interwikis
  • Subsection
  • Minimum Size in byte (for example more than 2000 byte)
  • Category
  • Navebox or Sidebar
  • Wikified
  • Less than 2 Message boxs

we can define a formula and add score to each article. This qualification could be checked automatically by an extension or bot or JS tools in local wikis after that added to wikidata. with this option we can

  • find un-standard articles redirect disambiguate pages in wikidata items very fast
  • have good tool to compare local wiki's articles according to their articles
  • have Auto thank tool to thanking the new users which are made article with satisfying Score (i.e. more than 7/10) Yamaha5 (talk) 07:07, 24 January 2014 (UTC)
This is a matter for the wikipedias.
The Wikidata:Development plan is due to start work on 'badges' soon. This will let each wikipedia assign a quality mark to each article. A tool based on your proposal might be used to automatically add badges - distinguishing between stubs and starters say, provided the wikipedia agrees. I doubt the English Wikipedia would agree - there are too many ways such a tool could be fooled to attract a good score to an article full of lies and spam. 85.133.32.70 13:19, 24 January 2014 (UTC)
I do not think we should try to score the quality of Wikipedia articles here. On the other hand, I think we could really benefit from scores about the importance of the article, based on the number of the number of articles, of inbound links etc. That could be used to filter long queries, search results, whatever. But I do not think it can be done direcly for Wikidata, as we are supposed to simply add data from elsewhere, so someone would have to develop such a scoring system elsewhere, and once it seems relatively reliable and mature, we could conceivably use it. --Zolo (talk) 18:13, 25 January 2014 (UTC)

Şume

A village in Kahta  – The preceding unsigned comment was added by Ronipinar (talk • contribs).

What about it and which country? --Tobias1984 (talk) 13:59, 26 January 2014 (UTC)
@Ronipinar: --Tobias1984 (talk) 14:02, 26 January 2014 (UTC)
There is an entry, though the title is using the Turkish name: Q1890236. Added Şume as an alias. Rotsee (talk) 21:55, 26 January 2014 (UTC)

Data access for Wikisource coming in a month

Hey :)

Thanks to everyone who helped make the language link access for Wikisource so smooth. Let's get them data access as well now. We are currently planning this for February 25th. Coordination/questions/... on Wikidata:Wikisource as usual.

Cheers --Lydia Pintscher (WMDE) (talk) 13:52, 27 January 2014 (UTC)

Accessing Wikidata from Meta

Meta should not have sitelinks at all, but we should be able to use Wikidata data just like on a Wikipedia. Is it possible to get the translated sitelink of some item into some language currently, or to find out which ones exist from Meta? πr2 (t • c) 02:26, 28 January 2014 (UTC)

Well, we could link Meta's RFA to other projects' RFA... --Rschen7754 03:01, 28 January 2014 (UTC)
Or Meta's Checkuser Policy and Oversight Policy to the local wiki pages. Techman224Talk 06:02, 28 January 2014 (UTC)
I think it would be useful to have the links to meta and commons also visible on, for example, nl:Wikipedia:Overleden Wikipedianen. Trijnstel (talk) 11:20, 28 January 2014 (UTC)
Maybe, but Meta is not a content site. At best we could only link it to pages in the Project namespace, and maybe Project talk for certain request processes. Anyway, I'm more interested in being able to translate links to the user's language automatically, from Wikidata information. It would make some things a lot nicer on Meta. πr2 (t • c) 19:38, 28 January 2014 (UTC)
Then you are requesting that meta would have preferrably no phase 1 (interwiki links) support, but would have phase 2 (accessing data) support and access to arbitary data. Judging from Wikidata:Development plan - that would be possible in the fourth quarter of 2014 or in the first quarter of 2015, althrough it is unlikely that the links to pages would be as few as you suggest. Meanwhile you probably can use some of the many tools which use data from wikidata - altrough they would never be fully automatic, given your usecase.--Snaevar (talk) 13:33, 29 January 2014 (UTC)
Meta will get links via Wikidata as well just like all the other projects. --Lydia Pintscher (WMDE) (talk) 13:35, 29 January 2014 (UTC)

Authors and languages

Is there an agree-upon way to provide information about the languages an author published their works in? E.g. Mark Twain published in English. I think, that would be useful. --Slomox (talk) 12:37, 28 January 2014 (UTC)

For persons, there's only native language (P103) as of now, which is not quite what you're looking for, as of course usually any author is free to publish in any other language than his native one. For works, there's original language of film or TV show (P364), which is extacly what you're looking for. However, it's not possible yet to perform queries like "get the original languages of all works published by Mark Twain", let alone that the Twain item could display such an information directly. --YMS (talk) 12:48, 28 January 2014 (UTC)

affiliations

Does anyone know where to find the discussion regarding the deletion of P160 (P160)? Danrok (talk) 13:46, 27 January 2014 (UTC)

A permanent link is given in the deletion summary. If you're looking for the live version (though not changed since), it's archived as Wikidata:Requests for deletions/Archive/2013/Properties/3#Property:P160 (such things can be found using the "What links here" function linked on each page). --YMS (talk) 14:03, 27 January 2014 (UTC)
Seems to have been deleted for no good reason. Affiliation is not the same as membership. In the same way marriage is not the same as friendship. Danrok (talk) 13:00, 30 January 2014 (UTC)

Users Pages

I think we must insert Personal Pages on Wikidata:Notability/Exclusion_criteria for this RFC Wikidata:Requests_for_comment/Inclusion_of_non-article_pages#User_pages --Rippitippi (talk) 05:47, 30 January 2014 (UTC)

It is already explicitly excluded on the main policy page. "To be valid, a link must not be a ... user page ..." The exclusion criteria IIRC is for things which would be allowed by the main page of the policy but which have been disallowed by specific vote. Ajraddatz (Talk) 05:49, 30 January 2014 (UTC)

Special:UnconnectedPages

This special page is very useful for finding pages on Wikipedias that aren't connected to Wikidata items. For some pages, though, you would never need the links, and those pages clutter the list and make it harder to find things that do need links. Is there a way a page can be exempted from being listed there? Some of the kinds of pages I'm thinking of are:

  • Redirected categories
  • Maintenance categories, such as "Articles needing additional references from July 2013"
  • Archives of various pages

There are probably others. Pages to be exempted could be determined by each Wiki. So, is there a way to keep a page off the list? --Auntof6 (talk) 08:15, 29 January 2014 (UTC)

I am not sure if there is any technical issue with this. But so we don't forget can you add it to bugs.wikimedia.org please? --Lydia Pintscher (WMDE) (talk) 13:38, 29 January 2014 (UTC)
Is there any api-version of this page? -- Lavallen (talk) 18:46, 30 January 2014 (UTC)
What is an api-version? --Auntof6 (talk) 07:35, 31 January 2014 (UTC)
I should file a bug, even though it's more of a feature request than a problem? Just verifying before I proceed. --Auntof6 (talk) 07:35, 31 January 2014 (UTC)
Yes please. We use bug reports also to track features. --Lydia Pintscher (WMDE) (talk) 07:40, 31 January 2014 (UTC)
OK, it's done -- Bug 60666. Thanks for the help. --Auntof6 (talk) 07:49, 31 January 2014 (UTC)

Removing deadlinks

I want to ask about Property:P856, should a claim to a link which is dead for a long time and doesn't look to be ever revived be deleted (like I did, see talkpage at Dewiki)? Ignatus (talk) 15:30, 29 January 2014 (UTC)

Why not working the same way as enWP? I would replace the link through "Official website (Archive)". If the archived website contains mainly advertising (de:Diskussion:Shirley Mills) the link should be deleted in WP first. (Otherwise a bot or an other user will add it again to WD.) --Kolja21 (talk) 16:08, 29 January 2014 (UTC)

I would expect the official link to always link to the official website, regardless whether the link is dead or not and regardless whether the link has information of particularly high quality or not. So no, I would disagree with your removal. Our third party re-users should be the ones to decide whether they want the official link or not, and better to lean to official dead link than to a unofficial live link.

Now, you might consider qualifying the statement with an accessdate or archiveurl and archivedate. --Izno (talk) 00:01, 30 January 2014 (UTC)

Sounds reasonable, but unfortunately professional spamers love former official web sites. I don't mind nude photos but I wouldn't like to click on a web site infected with a virus. --Kolja21 (talk) 00:21, 30 January 2014 (UTC)
I guess I would expect it to be the exception rather than the rule. --Izno (talk) 13:08, 31 January 2014 (UTC)

Julian/Gregorian calendar mixup in bot imported dates

Charlemagne (Q3044) had his death date imported from the Russian WP by a bot [9], however this bot set the date in the Julian calendar as date in the proleptic Gregorian calendar, for the display however the Julian calendar was chosen. Thus instead of January 28 it showed January 24, after correcting it the proletic Gregorian value saved is February 1, and the display in Julian now shows the expected 28th. I worry this must have been a systematic error for all the dates imported by this bot. Ahoerstemeier (talk) 09:12, 30 January 2014 (UTC)

Cf. Wikidata:Project_chat/Archive/2013/12#Dates in Julian / Gregorian calendar. I propose to delete by bot at least all dates before 1923 set by the concerned bots. Perhaps one of the bots that have produced this mess could clean it up again or any other bot. You may try to address them. As you can see I had no success in doing so. Good luck! In the hope that the mess won't stay long, --Marsupium (talk) 09:26, 31 January 2014 (UTC)

addition of data sets

Hi all. How does the coordination of the addition of large amounts of data from other sources take place? For example, I can see that a vast amount of people infobox data has been imported from some (which?) wikipedias, but that the data from many other infobox types has not. Can I assume that someone has this covered or is the general user (like me) supposed to take it on? I'm guessing that we're not waiting for people to add all such info manually, so is there somewhere where people are dividing up this task? I'm happy to help if so. Smb1001 (talk) 22:57, 31 January 2014 (UTC)

You can file a bot request if you want someone else to pull in a large amount of data, or you can build a bot yourself. I think the PYWiki support is generally decent at this point. --Izno (talk) 23:07, 31 January 2014 (UTC)
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat/Archive/2014/01&oldid=1102550675"