Wikidata:Project chat/Archive/2015/06

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

List of properties?

I'm looking at lists of properties at https://www.wikidata.org/wiki/Wikidata:List_of_properties/all or https://www.wikidata.org/wiki/Wikidata:List_of_properties and I find that head of government (P6) and head of state (P35) are not there. So I wonder is there a an up-to-date list grouped by some criteria other than by numbers? Looking something up in numbers lists is very hard since without knowing numbers upfront one has to look through a lot of pages.  – The preceding unsigned comment was added by Laboramus (talk • contribs).

The eight lists linked from Wikidata:List of properties are updated by users, so there are likely to be out of date. The "/all" list is updated by a bot but its source is those eight lists. I have created a query to see which properties are not listed there. Matěj Suchánek (talk) 08:01, 31 May 2015 (UTC)
It looks like that query return 744 items, which means out of about 1900 properties 40% is not described... --Laboramus (talk) 08:24, 31 May 2015 (UTC)
Special:ListProperties (by datatype). --- Jura 08:52, 31 May 2015 (UTC)
@Jura1: thanks, that looks useful! --Laboramus (talk) 06:51, 1 June 2015 (UTC)

Japanese translation

I'd like to ask for a Japanese translation. Mineral: Imayoshiite. Named after Ryuji Imayoshi (1905-1984), mineral collector (Google translation of [1]). Is the name of the mineral collector Ryuji Imayoshi or Takaharu Imayoshi ? Thx. Regards --Chris.urs-o (talk) 17:10, 31 May 2015 (UTC)

They give the romaji reading : "今吉隆治(Takaharu Imayoshi)", so I think it's Takaharu. According to this Japanese Wikipedia article of another person who have the same name,
隆治
can be read "Ryūji" or "Takaharu". Regards, Thibaut120094 (talk) 18:25, 31 May 2015 (UTC)
Thank you very much --Chris.urs-o (talk) 03:10, 1 June 2015 (UTC)

place of death (P20) - at sea

Hello everybody,

I just had someone who was declared to have died at sea. Could not find it. Could some please point me to the right item, if one exists ?

Thank you so much... --Hsarrazin (talk) 13:25, 30 May 2015 (UTC)

Pacific Ocean includes Captain Nemo (Q1046049). --- Jura 14:14, 30 May 2015 (UTC)
LOLLLLLL - Thanks for that one, the BNF AC does not say in which sea though - could be Mediterranea, Atlantic, Antarctic, Indian, or even Caspian or Black sea…
in fact, I just found here that it was near Accra, on the way from Congo to Belgium… died of illness, was immersed in the sea - probably a mention "dead at sea" would be useful for many sailors and explorators ;) --Hsarrazin (talk) 14:39, 30 May 2015 (UTC)
@Hsarrazin, Jura1: Just created World Ocean (Q20018609)     . I think that's the closest things we have to as see in the broader sense  .  – The preceding unsigned comment was added by TomT0m (talk • contribs).
@TomT0m, Hsarrazin, Jura1: Is that not a duplicate of World Ocean (Q715269)? Thryduulf (talk: local | en.wp | en.wikt) 15:00, 31 May 2015 (UTC)
@Thryduulf: I don't know, would Aral Sea (Q35883)      be a part of the world ocean ? TomT0m (talk) 15:30, 31 May 2015 (UTC)
Joseph Ingraham (Q1567975) uses simply "sea" sea (Q165). As it's a technical term, maybe we should probably make an item only for use with P20. As we tend to like more specific locations, "Atlantic Ocean" could go as a second value. BTW, for burial location, some items use burial at sea (Q630013). --- Jura 15:19, 31 May 2015 (UTC)
@Jura1: I don't see how it's useful to create an item. The technical term is just a denomination, it's deducable from the fact that the person died in a see or an ocean. Plus its meaning is "he died in some see or ocean", which is precisely said by
⟨ X ⟩ place of death (P20)   ⟨ World Ocean (Q20018609)      ⟩
(or World Ocean (Q715269)     ). TomT0m (talk) 15:30, 31 May 2015 (UTC)
Please do not use place of death (P20)+hospital (Q16917) and place of burial (P119)+cemetery (Q39614). Next question, what do we do when someone dies like Michael J. Adams (Q1122204)     ?--Giftzwerg 88 (talk) 19:17, 31 May 2015 (UTC)
Thank you everybody... in fact, I think that, burial location: burial at sea (Q630013) is the solution to my problem :) - I had seen it at least once, but could not remember how it was done... thanks Jura  – The preceding unsigned comment was added by Hsarrazin (talk • contribs).
@Hsarrazin: I tend to disagree, this is linking a property with a location range to a ... type of burial as value. This tends to be not really a good practice if we can avoid this by using a real location, even if it is a blurry one like the items we linked in this discussion. I think it's a better idea to use a proper location as value and in the infobox code a special case like if value == Q"oceans of the Earth", then display("at see"). TomT0m (talk) 13:21, 1 June 2015 (UTC)

List of properties?

I'm looking at lists of properties at https://www.wikidata.org/wiki/Wikidata:List_of_properties/all or https://www.wikidata.org/wiki/Wikidata:List_of_properties and I find that head of government (P6) and head of state (P35) are not there. So I wonder is there a an up-to-date list grouped by some criteria other than by numbers? Looking something up in numbers lists is very hard since without knowing numbers upfront one has to look through a lot of pages.  – The preceding unsigned comment was added by Laboramus (talk • contribs).

The eight lists linked from Wikidata:List of properties are updated by users, so there are likely to be out of date. The "/all" list is updated by a bot but its source is those eight lists. I have created a query to see which properties are not listed there. Matěj Suchánek (talk) 08:01, 31 May 2015 (UTC)
It looks like that query return 744 items, which means out of about 1900 properties 40% is not described... --Laboramus (talk) 08:24, 31 May 2015 (UTC)
Special:ListProperties (by datatype). --- Jura 08:52, 31 May 2015 (UTC)
@Jura1: thanks, that looks useful! --Laboramus (talk) 06:51, 1 June 2015 (UTC)

Patrolling changes

Hi, I'm wondering whether there are patrols in Wikidata and whether I can see what version of Wikidata entity is patrolled. --Andy pit (talk) 05:44, 31 May 2015 (UTC)

Patrolling pages in Wikidata is possible but not necessary. The only way to check it I know is Special:Log/patrol. Matěj Suchánek (talk) 08:11, 31 May 2015 (UTC)
Patrolling (Flagged revisions) is very necessary but Wikidata community is against it. Thus there is no way to see "last patrolled edition" or something. The Special:Log/patrol special page is about checking differences (changes), but it does not prevent vandalism to occur on the page (because older revisions could still have vandalism) -- VlSergey (gab) 09:34, 31 May 2015 (UTC)

Please give me some links, where I can read why flagged revisions are not necessary. Thank you, Conny (talk) 10:10, 31 May 2015 (UTC).

@Conny: I think it looks like Matěj Suchánek tells patrolling pages is not necessary. Edits cannot be patrolled after they have been ~30 days old. -- Innocent bystander (talk) 10:33, 31 May 2015 (UTC)
I meant that now you don't have to mark a change as patrolled to propagate it to all users and clients. Of course, having flagged revs would be nice. Matěj Suchánek (talk) 10:39, 31 May 2015 (UTC)
@Matěj Suchánek: Is there a discussion of flagged revs allready? Maybe we can make a poll, because MediaWiki should be ready for this at this time and for Wikipedia-Users it seems very important to have flagged revisions! Thank you, Conny (talk) 11:38, 31 May 2015 (UTC).
@Conny: It was started by Vlsergey, see Wikidata:Project chat/Archive/2014/09#Proposal: enable FlaggedRevs on wikidata.
My opinion on this: it would nice to have it enabled because WD influences many projects but it should not be the only way how to prevent vandalism, we could also make use of Wikidata development (like enforcing property constraints).
Also, I think I have once seen a Phabricator task where enabling on a WMF project couldn't be done because of WMF's disagreement. Matěj Suchánek (talk) 11:46, 31 May 2015 (UTC)
++1. Conny (talk) 13:57, 31 May 2015 (UTC).
Maybe Wikimedia disagrees because stats show, that flagged rev is a new editor killer... @all Maybe it is to early for such mechanisms? Conny (talk) 13:59, 31 May 2015 (UTC).
@Conny: how can any stat to show that invisible by default mechanism can do anything with new editors? Flagged revisions first of all is way to mark revisions as checked. And no, it is not too early, it's already late. We already have some "wikidata-protected" pages in ruwiki because of vandalism that we can't handle. -- VlSergey (gab) 18:36, 31 May 2015 (UTC)
@Vlsergey: Good to hear that you see advantage there. When it is time now, than let us initiate a poll for this. Conny (talk) 20:28, 31 May 2015 (UTC).

I have seen many cases in which Wikidata patrolling had been really important, because you can possibly see every change in the Wikipedias as well And, to be important, if a few editors try to do this, I think Patrolling is not so much work as you think: You only need some editors which sppek differnet languages to check this: [2] --MGChecker (talk) 11:10, 31 May 2015 (UTC)

There are users patrolling recent changes but they are not online everyday. For example, all edits of last weekend are patrolled. --Pasleim (talk) 11:28, 31 May 2015 (UTC)

Thanks everybody for answers. As I understand (after reading Wikidata:Project_chat/Archive/2014/09#Proposal:_enable_FlaggedRevs_on_wikidata), patrolling system will not be enabled in Wikidata in near future, so wikipedias are not able to trust data from here. --Andy pit (talk) 17:39, 31 May 2015 (UTC)

That's a weird conclusion. FlaggedRevs doesn't have a purpose on Wikidata without people that want to patrol changes. There are not much patrollers now and I don't think that will change with FlaggedRevs. FlaggedRevs will only hurt the good willing people, there is also a lot of good changes happening. There are a lot of other solutions possible. Sjoerd de Bruin (talk) 17:43, 31 May 2015 (UTC)
In what way should flagged revision make Wikidata more trustworthy? Almost all errors here are added by mistake by trusted users. Flagged revision would not prevent that! -- Innocent bystander (talk) 17:49, 31 May 2015 (UTC)
@Innocent bystander: Flagged Revisions protect not from errors but from vandalism. And I saw a lot of those already that was "distributed" further to ruwiki. -- VlSergey (gab) 18:36, 31 May 2015 (UTC)
+1 from me. No Wikipedia shall use Wikidata without Flagged Revisions or any alternative way to protect itself from vandalism. -- VlSergey (gab) 18:36, 31 May 2015 (UTC)
 
Vandalism from Wikidta in a Wikipedia article
Unfortunately we can't to anything against using Wikidata ihn Wikipedia, as you can see on the right. The global community hasn't any control about the app, which uses the description of the connected Wikidata item. In this picture you also can see why Flagged Revs would be important, if you speak a little bit German. If not: The subtitle from Wikidata isn't kind. --MGChecker (talk) 19:01, 31 May 2015 (UTC)
@MGChecker: there is a small tool that can help to use Wikidata by keeping Wikidata copy and updating it manually on demand. See here. -- VlSergey (gab) 19:58, 31 May 2015 (UTC)
Like I said,the community can't change anything of the app. --MGChecker (talk) 20:10, 31 May 2015 (UTC)
Sorry, didn't notice app mention. -- VlSergey (gab) 20:46, 31 May 2015 (UTC)

What about setting it up to how it is on enwiki, where only some pages have it turned on? Right now, there are some items that are semi-protected or even fullprotected for the long term because of repeated vandalism, some of it of a sensitive nature (i.e. items about famous people). --Rschen7754 20:41, 31 May 2015 (UTC)

I believe it shall be enabled on all pages, but each project community (ruwiki, enwiki) shall decide would it like to see any changes immediately or only patrolled one. Stabilization of some entries (not more than 1k of pages) may be used on Wikidata as well. -- VlSergey (gab) 20:44, 31 May 2015 (UTC)
Flagged revisions on the Russian Wikipedia are a complete disaster, with revisions not being checked for years. We have way less people on Wikidata, so that it would be a very bad idea to introduce flagged revisions here. Unchecked revisions are just going to pile up--Ymblanter (talk) 20:45, 31 May 2015 (UTC)
Pending changes as installed on en.wiki may actually work, as soon as we do not overdo them.--Ymblanter (talk) 20:46, 31 May 2015 (UTC)
Not, it is not. It is very useful element of anti-vandalism protection at ruwiki. We have a lot of unchecked revisions, but the number of such slowly going down. Anyway, it doesn't bother as long as page is not stabilized, and in case it is, there are always people to check and "approve". "Pending changes" is just another name for "Flagged Revisions", you know? -- VlSergey (gab) 20:50, 31 May 2015 (UTC)
How do we solve such things as users adding labels and descriptions to languages that are not familiar to any or only a few of us? In the items about the geography where I live, South Sapmi names are common. But I do not know if we have one single user here who understand a single word of South Sapmi. And since there is only ~500 speakers of that language today, it can be hard to find any who can verify such edits. Even in my own language, which has ~10 million speakers, we have maybe only a handful of active WD-users who edits anything else than sitelinks. And you have to count me out since I do not do countervandalism.
Another problem is that the reputation of Wikidata among the Wikipedians is in many cases not so good. We have added statements here based on Wikipedia, which according to almost everybody isn't a good source. If we starts to use "flagged revisions", we add something else that in many WP-users eyes does not rightfully belong in a "free" encyclopedia. If the lack of use "flagged revision" makes Wikidata look "untrustworthy" in the eyes of ruwiki, the use of "flagged revision" would make Wikidata look "unfree" in the eyes of svwiki. Dewiki have many potential users on svwiki, but many of them have left the building after the installation of "flagged revision". I'm afraid that a site-wide installation of Flagged revision would initiate an "opt out of Wikibase client"-discussion on svwiki. WMF could of course overrule such a decision, but that would only cause users to leave the project. -- Innocent bystander (talk) 04:45, 1 June 2015 (UTC)
@Innocent bystander: 1. First of all, we have autotranslation. If Google translate can't handle name and/or description, not even to show image, it's okay to assume that this is not a obvious vandalism. Obviously not many people will see it as well. My point that the obvious vandalism (in English, German, etc.) not only the most simple to handle, but most destructive as well. We don't need to verify edits and compare them with sources (we can, but that's completely another task), we need to make sure it doesn't break things at most popular wikis. 2. "Free" doesn't mean "you can do anything"... but i see your point. Contra-argument: wikis shall be able to ignore flagged revisions if they want to. If wiki would like to always see latest changes -- they shall do nothing. But there must be an option to say "hey, give us only checked revisions (if you have them for current entity), without obvious vandalism". -- VlSergey (gab) 09:23, 1 June 2015 (UTC)
@Vlsergey: Flagged revision demands every non-trusted users edits to be checked, even those done by users who edits in the smaller languages. The content of Sapmi statements about "original title" or "name of birth" will be used also in the larger projects, even those with flagged revision.
Autotranslation from and to Uralic languages gives you more headache than it gain. My feeling, when I tried to learn Russian, was that everything is done backwards compared with my own language. But at least, backwards is a clear direction. Sapmi has, compared with that, no sense of direction at all. Therefor Autotranslation or Uralic languages gives you extremly poor results.
I am not saying that I agree with all these opinions about a "free wiki". One of the reasons I did not feel comfortable with the sysop-tools on WP was that I could not "protect" as many pages as I thought should be protected. But the fact is, there are many users with these opinions, and if I want my project to use Wikidata, I have to make it look "good" even for those users. -- Innocent bystander (talk) 10:11, 1 June 2015 (UTC)
@Innocent bystander: 1. "demands every non-trusted users edits to be checked" -- we can add all "trusted" users to our trusted list. I.e. give trusted flags to all local sysops and, for example, all editors with 1 year after registration and more than 256 edits. 2. I think "original title" or "name of birth" shall be the simplest facts to check, much simpler than title or description. Usually I just check any reliable online source for it. (2 and 3). But this is not an argument to not enable Flagged Revisions. It's okay if we can't fight some non-obvious vandalism that we can't fight on local wikis as well (at ruwiki we don't have people who can speak Uralic languages). But we need a tool to fight the most obvious one. -- VlSergey (gab) 10:21, 1 June 2015 (UTC)
@Vlsergey: Most of the "vandalism" I have reverted here has been done by users who do not understand Wikidata yet. I know from talks on "my wiki" that not even wellknown trusted users fully understand why we separate articles about "Paris, Texas" from "Paris, France". "Not so well-matched interwiki is better than none interwiki at all" is their opinion. I am afraid that "Flagged revision" here can make us to self confident. "We have flagged revision, our data cannot be corrupted!" -- Innocent bystander (talk) 10:47, 1 June 2015 (UTC)
@Innocent bystander: those kind of changes is not vandalism, but mistakes. Of cource, Flagged Revisions does not make Wikidata error prune. But they do improve quality of data. Check this edit that shall never appear in any wiki: [3]. This is obvious vandalism. It should be reverted and shall not appear on any wiki. It's okay if some change "pass" Flagged Revision checks, it's okay if some mistakes pass this kind of "filter". Still it's better than nothing. -- VlSergey (gab) 11:57, 1 June 2015 (UTC)
@Vlsergey: That edit was successfully tracked by a filter, so we already have more than "nothing" in this very moment. -- Innocent bystander (talk) 12:38, 1 June 2015 (UTC)
@Innocent bystander: what kind of filter are you laking about? That change was visible for 5 hours. -- VlSergey (gab) 13:02, 1 June 2015 (UTC)
I read "possible vandalism" in the edit comment of that edit! -- Innocent bystander (talk) 13:05, 1 June 2015 (UTC)
Also, please note that most people of the European timezone's (the largest userbase, I think) are asleep at that moment. We need patrollers for every timezone. Sjoerd de Bruin (talk) 13:10, 1 June 2015 (UTC)
@Innocent bystander: it was visible by wikis for 5 hours, 5 hours Carl! Filter is nice, but it didn't prevent such obvious vandalism. With Flagged Revisions wikis can try to prevent such changes to be visible. -- VlSergey (gab) 14:17, 1 June 2015 (UTC)
What would be the number of changes to check per day? Currently, I mainly see an issue with eswiki making it too easy to stop by at Wikidata after vandalizing the articles there. --- Jura 06:03, 1 June 2015 (UTC)
It took 43 minutes to log the last 2K edits, the majority is in the item namespace, and now there is deep night in North America. Count for 50K edits per day.--Ymblanter (talk) 07:02, 1 June 2015 (UTC)
@Ymblanter:, you don't need to patrol all changes. First of all, bot and sysop changes shall be autopatrolled by default. Second, trusted users from other wikis (maybe wikisource, wikinews, commons as well) shall be also autopatrolled by default, may be have an ability to patrol as well. Third, we can even try to trust by default registered users from other wikis if they have, say, 1 year since registration and 256 edits accross all projects. It would be trivial bot task to check and assign "autopatrolled" flag to those users. So far only small fraction of edits will be notpatrolled automatically -- like the current number of unpatrolled revision in current system in Wikidata. -- VlSergey (gab) 07:28, 1 June 2015 (UTC)
A bot patrolling all changes caused by page moves on Wikipedia should be possible and cover a lot of currently unpatrolled changes. Do you think a RFP for that task is needed? -- Bene* talk 07:36, 1 June 2015 (UTC)
To be frank, page moves are not our problem. Such actions should be supervised by the editors in the clients, not by us. If a wellknown botowner can do this task, just let hir do it, without the red tape. -- Innocent bystander (talk) 08:36, 1 June 2015 (UTC)
I don't totally agree that page moves are not our problem, as an article is supposed to identify the subject of the item, especially if the item has a few statements. I agree Wikidatas community is too small to handle this, but this should be solved by closer relationships between Wikipedias and Wikidata. Not by community turning on themselves and throwing the ball at each other. But yes, looking at this discussion, it's a hard time making all this communities understand each other and cooperate :) TomT0m (talk) 09:02, 1 June 2015 (UTC)
Changing content in pages would then also be our problem. I, for example, often see changes from disambig to normal content. I also see a continuum from pure disamig-pages over semi-disambigs to complete "B & C"-articles, where the connection between B and C is not fully obvious, for example Lemek@svwiki. -- Innocent bystander (talk) 09:22, 1 June 2015 (UTC)
@TomT0m: fighting with vandalism and preserving logical consistency are different tasks. Rename-to-vandal the page is easier to handle at local wiki, thus we can assume that such renaming will be checked and handled at local wiki and not to check for vandalism here at Wikidata. -- VlSergey (gab) 09:28, 1 June 2015 (UTC)

It is still a long way off, but on Wikidata I can imagine that FlaggedRevs could be really useful, if there are enough users to patrol. We have to remember that we have a big reponsibility for the Wikipedias. But having too few users for patrolling does not have to be an obstacle, because it is possible to configure FlaggedRevs that versions that are not reviewed are shown by default. I have some ideas that would be good to make handling it more useful:

  • 50 edits shouldn't be enough fopr autopatrol.
  • Interwiki conditions (if not possible: make them possible!): If you have 250 Edits in all Wikis as a whole and are registered more than maybe 9 months, and weren't blocked until then you get autoreview.
  • Maybe validated revisions could be something useful as well: It woukd Wikidata make really more reliable.
  • Eventually we shouldn't concentrate on the revisions so musch, but more on the statements. In opposite to Wikipedia wie are a database.

So that are only a few ideas that eventually could be useful. --MGChecker (talk) 15:19, 1 June 2015 (UTC)

Japanese translation

I'd like to ask for a Japanese translation. Mineral: Imayoshiite. Named after Ryuji Imayoshi (1905-1984), mineral collector (Google translation of [4]). Is the name of the mineral collector Ryuji Imayoshi or Takaharu Imayoshi ? Thx. Regards --Chris.urs-o (talk) 17:10, 31 May 2015 (UTC)

They give the romaji reading : "今吉隆治(Takaharu Imayoshi)", so I think it's Takaharu. According to this Japanese Wikipedia article of another person who have the same name,
隆治
can be read "Ryūji" or "Takaharu". Regards, Thibaut120094 (talk) 18:25, 31 May 2015 (UTC)
Thank you very much --Chris.urs-o (talk) 03:10, 1 June 2015 (UTC)

Merge of given names "Gaetan" & "Cayetano"

Does anyone think this, by User:Vatekor, was a good merge; or should it be undone? It seems bad to me. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:27, 31 May 2015 (UTC)

I think it should be undone. --Yair rand (talk) 22:37, 31 May 2015 (UTC)
As merges can create lots of issues in the project, I think we need some kind of patrolling or restriction of merging. Either a four-eyes principle has to be added or a new userright (maybe attached to property-creator). What do you think? -- Bene* talk 18:11, 1 June 2015 (UTC)
@Ivan A. Krestinin:. Related to the discussion about solving redirects automatically. TomT0m (talk) 18:44, 1 June 2015 (UTC)
Hard to argue in terms of good or bad, but it's an understandable one. The chosen scheme for first names at Wikidata is somewhat different form the way interwikis used to be done and for this specific name the items weren't that close to what the scheme provides. The merge mainly lead to the messy situation being concentrated. Unfortunately the subsequent edits made it difficult to de-merge. When this topic was started, I think the situation was already improved. The "Cayetano" (and "Gaetan) now fit better into the general scheme, check:
Yesterday, we reached 2 million items for persons where the first or given name (P:P735) is defined. Hope this clarifies things. --- Jura 18:50, 1 June 2015 (UTC)

Item's top-level categories ?

Wikipedia's Portal:Contents/Categories list 11 major categories. Namely : Culture, Geography, Health, History, Mathematics, Nature, People, Philosophy, Religion, Society, Technology.

An other classification is Category:Main_topic_classifications (35C), but the issue is similar.

For a given entity, is there a way to know to which of these top-level category/ies it belong to ? Yug (talk) 13:44, 1 June 2015 (UTC)

@Yug: Rigth now classification is done mainly with a hierarchy of classes, using principle exposed in Help:Classification. This is something different from what you expose, entities are classified as island or geographic entities rather than geography. TomT0m (talk) 15:47, 1 June 2015 (UTC)

Arbitrary access enabled on Italian Wikipedia and all Wikisource projects

Hey :)

Arbitrary access is now enabled on Italian Wikipedia and all Wikisource projects as well. --Lydia Pintscher (WMDE) (talk) 14:13, 1 June 2015 (UTC)

What is the experience from the involved projects this far? -- Innocent bystander (talk) 14:39, 1 June 2015 (UTC)
My impression is that things are fine except for some quarrels because of people coming to nlwp who are not editors there and trying it out in main namespace in-line. I assume this is settled now or will very soon. Would love to hear more from other people. --Lydia Pintscher (WMDE) (talk) 14:43, 1 June 2015 (UTC)
@Lydia Pintscher (WMDE): those are great news! Is there any way of checking which pages are using arbitrary item access for a certain site? Thanks! --84.126.89.28 15:50, 1 June 2015 (UTC)
Not yet. We're working on getting the necessary database tables on toollabs as well as some way to get this info on-wiki. --Lydia Pintscher (WMDE) (talk) 15:52, 1 June 2015 (UTC)

Lists of public art on Wikipedia

Is there a way for items to be generated from Wikipedia articles which use machine-readable list templates, such as en:List of public art in the City of Westminster? I am keen to integrate such lists further with Wikidata and ultimately to generate the lists wholly or partly from Wikidata items.

I've been experimenting with Listeria and created a list of statues of British royalty in London here. One problem with that list is that the artists don't sort by surname. Is there an equivalent of defaultsort on Wikipedia for Wikidata? Ham II (talk) 08:36, 1 June 2015 (UTC)

No, but there is a property about "Surname". -- Innocent bystander (talk) 10:15, 1 June 2015 (UTC)
The templating (and embedded metadata in those templates) makes it easier for someone to write a parser, which will work across multiple articles. Perhaps doing so is project for a future hackathon? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:08, 1 June 2015 (UTC)

Use of special characters in titles

On Wikipedia, it's impossible to use certain characters, such as #, in article titles for technical reasons. Do the same restrictions apply here? For instance, this album is called Inspiration #1. Its Wikipedia article is titled Inspiration, but it's always called Inspiration #1 in the prose so the correct name is made clear. Is it possible to rename its record here on Wikidata to Inspiration #1 or will that cause problems? If not, does anyone have suggestions for how to show the album's correct name on its record? (alias maybe?) Thanks so much! Shinyang-i (talk) 20:48, 1 June 2015 (UTC)

On Wikidata you can rename it to Inspiration #1. It causes problem only on Wikipedia as it contains "#" in title, but here on Wikidata pages are named as ID's like "Q12345", so Inspiration #1 won't be shown in URL. --Stryn (talk) 20:53, 1 June 2015 (UTC)
Thanks so much for the info. That's exactly what I wanted to know! :) Shinyang-i (talk) 21:11, 1 June 2015 (UTC)

Catholic bishop (Q611644)

I'm adding Catholic bishop (Q611644) for example here Joseph Hii Teck Kwong (Q1707237) do you think is the right way? occupation (P106)->priest (Q42603) position held (P39)->Catholic bishop (Q611644) religion (P140)->Catholicism (Q1841) --Rippitippi (talk) 04:16, 26 May 2015 (UTC)

Catholic bishop (Q611644) is good for 'position held'. This can have qualifier (of (P642):Roman Catholic Diocese of Sibu (Q876489)).
Religion should be 'Catholic Church (Q9592)'. Catholicism (Q1841) is a wider term that includes people that are no longer part of the Roman Catholic Church. Look at other more famous bishops for additional properties to add.
Hope this helps. Filceolaire (talk) 06:52, 26 May 2015 (UTC)
I have a added the qualifiers for position held (P39) as well as adding his previous position as auxiliary bishop (Q75178) and therefore as titular see (Q15217609), as well as the properties Catholic Hierarchy person ID (P1047) and consecrator (P1598) which are also quite useful for Catholic bishops. Only thing I am not sure about - should we use the date of appointment or the date of consecration for start time (P580)?
Its the date of consecration that counts. He is not a bishop before consecration. If the person dies one dey before consecration, he is not a bishop no matter if he is appointed. A bishop can have several appointments, the day he begins his service in the new diocese (Q665487) counts. It is also a good idea to use diocese (P708).--Giftzwerg 88 (talk) 09:47, 26 May 2015 (UTC)

We have also Catholic priest (Q250867) it is not redundant Catholic Church (Q9592) Catholic bishop (Q611644)? --Rippitippi (talk) 10:41, 26 May 2015 (UTC)

In P140, you could hardly add Catholic priest (Q250867). --- Jura 11:40, 26 May 2015 (UTC)

In religion (P140), there are more items with Catholicism (Q1841) than with Catholic Church (Q9592). Should we convert them? --- Jura 11:40, 26 May 2015 (UTC)

@Jura1, Rippitippi, Giftzwerg 88, Filceolaire: A possible data structure:

Snipre (talk) 14:22, 26 May 2015 (UTC)

Catholic priest (Q250867) and Catholic bishop (Q611644) are not useful: we can cross religion (P140): Catholicism (Q1841) with occupation (P106):priest (Q42603) or occupation (P106):bishop (Q29182) to get the same information. Better to delete these items. Snipre (talk) 14:53, 26 May 2015 (UTC)
The items exist because the wikilinks exist. They should be used because of that. --Izno (talk) 16:41, 26 May 2015 (UTC)
It is not because an item exists that we have to use especially when the item is created for managing links and not based on a classification. Just take the example of bishop: we have auxiliary bishop and catholic bishop. How can I manage auxiliary bishop from the Catholic church ? Do I have to create an item catholic auxiliary bishop ? Snipre (talk) 22:42, 26 May 2015 (UTC)
@Snipre: I'd like to recall a class item could pretty much be associated to a query. It's not a problem to keep them in such a case, it's an opportunity to check if all instances are resuts of a query. TomT0m (talk) 16:59, 26 May 2015 (UTC)
You think with a technical point of view and not with a classification point of view: the problem is if I have a catholic priest and I want to use that level of the classification for person items I should be able to do the same for all religions which have a priest. So for example I should have an item for old catholic priest. This lead to the creation of items which are not necessary because they are just intersections of existing definitions. If I can define a catholic priest, why can't I have an item "dead person" for example ? Or French dead person ? Or French female dead person or perhaps French dead woman is better? For me this kind of question about classifications should be discussed in a help page. Snipre (talk) 22:42, 26 May 2015 (UTC)
@Snipre: I don't think you see things clearly : there will be entities for queries. So I don't see a point in this kind of item economy. Second I'm definitely talking of classification. My point is that technically we have no problem dealing with such classes. I think it's a chance. TomT0m (talk) 12:27, 27 May 2015 (UTC)

Why on earth have a catholic bishop? A bishop is an office, the person who has it is a priest and he or she is of a particular branch of religion.. WHY HAVE IT ? GerardM (talk) 20:18, 27 May 2015 (UTC)

Why not? Seems more specific and informative than not having it. --Yair rand (talk) 20:31, 27 May 2015 (UTC)
(edit conflict) @GerardM: Because if someone is a bishop there is a set of statements that can automatically added. For example. We know his religion, that he is a priest. For example. Because there may be a list of bishop page. because someone wanted it and as we can manage thiss we don't want to delete/restore fight. TomT0m (talk) 20:49, 27 May 2015 (UTC)
@GerardM, TomT0m: A bishop is not a function in the catholic religion, this is a state. Bishop receives special privileges which are not linked to his office but to his person. So priest and bishop items are correct, auxiliary and titular bishop too, but catholic bishop or catholic priest are a different way to classify items. We see the problem to specify that a catholic bishop has only an auxiliary position: we have two item "auxiliary bishop" and "catholic bishop". How do we have to use them ? If we accept combined items we should theoretically create "auxiliary catholic bishop" item or we have to specify that "auxiliary bishop" should be used only for position held (P39) and "catholic bishop" for occupation (P106).
The problem is not to economise items, the problem to provide the guidelines which will offer later the insurance that a query will be complete based on a similar way to describe items. Typical problem: if some items use occupation (P106)->priest (Q42603) with religion->catholic and other ones occupation (P106)->Catholic priest (Q250867) how we can get in one query all items for catholic priests ? The only way is to force people to use occupation (P106)->Catholic priest (Q250867). But now how can I get all priests from the Old Catholic Church ? I don't have an item "Old Catholic priest" so do I have to create it to keep a similar structure to Catholic priest (Q250867) ? Here we see the consequences of that system: we need a systematic creation of all combinations. This is not a problem just an organization to create and a management to follow. If we keep a "key words" system meaning we try to separate as much as possible the description criteria into different items and we combine these criteria only for query prupose we are getting out of the creation of combined items which is an endless work. Snipre (talk) 07:29, 28 May 2015 (UTC)
@Snipre: A class can be defined by a query. Theorically a reasoner could infer that an item that is in the result set of a query is an instance of (P31) the class, but also that an instance of the class has the properties that, if are not explicitely stated, would make the item in the result set of a query. As Wikidata is not that powerful, this means that we would have to do the work semi-automatically. Possible workflow: I know that our item is an auxiliary bishop. I make him a member of the auxiliary bishop class. A bot or a tool knows the class expression (or the query) that defines the auxiliary bishop class. After I or some other validated, he adds the missing statement without I having to do it manually. Note that anyway, if you want the auxiliary catholic bishop class, you still have to create an item for the query in the initial development plan as "on the fly" arbitrary query were not supported. This means it's no extra work actually wrt. explosive combination. The choice is which query do we make wrt. which class do we create. No difference. Because, no, we don't have to create every combinations. Eventually if there is a good query we could even delete the concrete statement that he is a catholic auxiliary bishop. I pretty much see in the future that a list of the queries an item is a result of can be exposed of the item page, maybe in the form of inferred P31, why not ? TomT0m (talk) 11:39, 28 May 2015 (UTC)

Respectfully, when we cannot even have the religion right, this difference about all kinds of bishop is not understood at all by the people we do it for. GerardM (talk) 15:20, 28 May 2015 (UTC)

so we can get to a guideline? --Rippitippi (talk) 17:54, 28 May 2015 (UTC)

@Rippitippi: My recommandation (for catholic clergy):
If you want to use Catholic priest (Q250867) or Catholic bishop (Q611644) add them after adding the previous data structure. At least like that we are sure to be able to query all catholic persons just using religion (P140).
Snipre (talk) 17:23, 29 May 2015 (UTC)
@Snipre: shouldn't P104 (P104) be religion (P140)? Mbch331 (talk) 17:28, 29 May 2015 (UTC)

Some of this is the same conversation that was happening over at the property proposal for religious rank. Please check that for some other editors' thoughts too. Note also that the word "consecrator" is not always canonically valid — a bishop is "consecrated" but a priest is "ordained" and a cardinal is "elevated," etc. When I asked for more properties around that (say, "ordained by" or "elevated by") nobody liked that, and I was told to use conferred by (P1027). (By that logic, consecrator (P1598) should be deleted.) Also, it's imperative to use start time (P580) and end time (P582) if everyone insists that "bishop" is a "position held" — see Joseph Ritter (Q1706916) for an instance where the same person was archbishop of numerous places. Sweet kate (talk) 18:05, 29 May 2015 (UTC)

Sweet kate You mix event and actor. A priest is ordained by a bishop but there is no special title for the bishop which is doing the ordination. So the title of consecrator is not totally correct but as there is no special title this is not totally wrong. And a cardinal is not elevated but created and a bishop is not a position in the catholic tradition. Snipre (talk) 19:08, 29 May 2015 (UTC)

@Snipre: @Filceolaire: So we must use Catholic Church (Q9592) or Catholicism (Q1841) ? --Rippitippi (talk) 20:51, 29 May 2015 (UTC)

@Rippitippi: Catholic Church (Q9592) is not a religion: this is an organisation. Use Catholicism (Q1841) for religion (P140). Snipre (talk) 11:19, 30 May 2015 (UTC)

A bishop like a cardinal and a pope is first and foremost a priest. That is his occupation. He can hold an office; bishop whatever. When you must make a difference between all kinds of bishops, you can use attributes. There is no argument I have seen and understand why it should be different. Oh and yes, The pope, (all bishops of Rome) are Roman Catholic. Thanks, GerardM (talk) 17:40, 1 June 2015 (UTC)

@GerardM: Pope, cardinal and bishop start as priest but a bishop receives the complete set of the Holy orders. For example a priest can't ordain another priest even if he want. A priest is considered as half a bishop in some way because at the origin there were no priest but only bishop. Then for example all cardinals (if below 75 years old) and all popes should be bishop. Priest and bishop are like bachelor and master degree in education. If you don't want to distinguish between priest and bishop so you use cleric which common to all ranks. Snipre (talk) 07:57, 2 June 2015 (UTC)
So what .. First and foremost a bishop a cardinal a pope is a position. The religious aspects are irrelevant. It is the same for a king who cares that he is coronated or not? GerardM (talk) 13:19, 2 June 2015 (UTC)

Use of TranslateThis?

What do you think about using Template:TranslateThis in places like discussion pages or WikiProject pages? I’ve used it on a few discussion pages where the item was very German-specific and I expected the majority of people visiting it to use German instead of English as a language. I also thought about using it on WikiProject sites, which look a bit weird with a mixture of English text and translated Template:Q labels… but then it’s unclear what editors who don’t speak all translated languages should do. Thoughts?DSGalaktos (talk) 22:25, 29 May 2015 (UTC)

Verstehe ich für Wikidata nicht, Wikidata:Forum kennst Du?--Oursana (talk) 01:07, 30 May 2015 (UTC)
My experience is that all use of templates in headers makes it difficult to understand the edithistory in pages. The editcomments does then not fit the headers and it can be difficult to identify under which header the last comment was added. -- Innocent bystander (talk) 07:10, 30 May 2015 (UTC)
We could probably fix that by adding a bit of JS to fill in the edit summary box appropriately. This might require the use of the anchor parameter to be helpful, though. --Yair rand (talk) 20:12, 31 May 2015 (UTC)
Done. --Yair rand (talk) 07:34, 2 June 2015 (UTC)

Using from Wikidata in the German Wikipedia

  Info: A Request for comment (in German "Meinungsbild") on the Geman Wikipedia starts in two hours, if Wikidata statements should be used in articles/article tempaltes, indifferent of direct in article text or in templates or tables. The reason for that is prevention of differnet values in infoboxes and articles:

de:Wikipedia:Meinungsbilder/Nutzung von Daten aus Wikidata im ANR

If someone wants that, I can translate the main points of this request. --MGChecker (talk) 17:08, 1 June 2015 (UTC)

I don't understand German. I think it is worth pointing out to the participants that dates before 1923 are at risk of being wrong due to problems with Wikidata's handling of Gregorian and Julian calendar dates. The further back in time you go, the worse the problem becomes. In addition, there is an uncertainty of one year for all dates before the year 1 AD, due to confusion about Wikidata handles negative years. Jc3s5h (talk) 21:44, 1 June 2015 (UTC)
I'll translate it tommorrow (later today). --MGChecker (talk) 22:24, 1 June 2015 (UTC)
The Meinungsbild doesn't care about detailed points, but about Wikidata usage in general. The first part is about if Wikidata should used in the main namespace without, with restriction or if it shouldn't be used. There are to questions:
  1. Without restrictions/With restrictions or not at all
  2. With restrictions/Not at all.
At the moment it looks like as "With restrictions" will be accepted. In the second part it's voted, which restrictions shouls be used:
  1. Local data shouldn't be replaced by Wikidara data.
  2. Only datas with refereneces (Wikipedia not included, also trivial data) schould be used.
  3. Wikidata shouldn't be used until there is really unrealistic "prevention against vandalism": 1. Only sysops and other new privileged users can modify data etc.
  4. Wikidara shold only be used in templates (which also can be used in article text.)
At the moemnt the majority likes the last and dislikes the first, while the other ones aren't so clear. --MGChecker (talk) 13:12, 2 June 2015 (UTC)

Testing Flow?

Hey folks :)

I'm currently at the MediaWiki hackathon in Lyon and talking to lots of people about Wikidata and other projects. One of the people I talked to is Danny, the product manager of Flow. There have been discussions here about using Flow on Wikidata here before. At that time Danny said he wants his team to finish a few more things before Wikidata could use it. Those blockers have now been resolved and I think it'd be great if we can help another project with our testing and feedback to make the discussion landscape around Wikimedia better. Danny set up a test page for you to go wild on and there is a page with the current features/benefits of Flow. If you like it and are willing to give it a try I think item talk pages are actually a nice test case since not too much is happening there and the discussions that do happen are not too complex. And depending on how that goes we can look at other spaces. What do you think? If you want to give feedback on the state of Flow you can use Wikidata talk:Flow.

Cheers --Lydia Pintscher (WMDE) (talk) 13:41, 25 May 2015 (UTC)

Does Flow handle templates in section headings better? (anchors, etc.) Sjoerd de Bruin (talk) 14:55, 25 May 2015 (UTC)
Flow doesn't support section headings yet, but you may be able to use separate threads to do what you want. Can you describe an example of what you'd like to do? DannyH (WMF) (talk) 16:25, 25 May 2015 (UTC)
I use a template in the section above this one, would that be possible in the thread name? Sjoerd de Bruin (talk) 16:30, 25 May 2015 (UTC)
I have a very bad impression of FLOW on the English Wikipedia. It is substandard and is being aggressively pushed through by WMF. Typical discussions of FLOW tend to go around the fact that it is not compatible with the current markup and can not be turned on in the opt-out regime.--Ymblanter (talk) 17:49, 25 May 2015 (UTC)
@Ymblanter: I think it would be damaging to take a position only wrt. the WMF. I don't get the incompatibilities in the markup you refer to because it seems based on the Visual editor libraries. Do you mean it would not be possible for old talkpages to be converted to Flow ? The only valid question is imho : can it fits our discussion patterns scheme ? I think it fits like a charm and we get better notifications and answers tracking. TomT0m (talk) 18:11, 25 May 2015 (UTC)
My understanding is that converting is not possible. The old talk pages will be archived, and the new pages will use these Facebook-style discussion tools.--Ymblanter (talk) 11:19, 26 May 2015 (UTC)
  Question What about talkpages that are used partly for documentation, like Talk:Q5 or Property Talk:P31 ? @DannyH (WMF): TomT0m (talk) 18:11, 25 May 2015 (UTC)
I don't think Flow is ready for Wikidata yet, and I've been following Flow since the beginning. Mainly per TomT0m.--Jasper Deng (talk) 18:27, 25 May 2015 (UTC)
@Jasper Deng: The question might not be is Flow ready for Wikidata but Is Wikidata ready for flow. Using talkpages for documentation seems to me like a workaround. TomT0m (talk) 18:32, 25 May 2015 (UTC)
So, addressing the broader issue of where to put property/item documentation then, where are we going to put it? Note that the English Wikipedia keeps a lot of internal information like WikiProject affiliations on article talk pages too.--Jasper Deng (talk) 18:34, 25 May 2015 (UTC)
That's a question for @Lea Lacroix (WMDE): :) What are precisely the issues on enwiki, so that we know what we're talking about ? TomT0m (talk) 18:39, 25 May 2015 (UTC)
I think the issue of "documentation" on talk pages is much smaller on item pages. That's why I suggested them to start. The stuff that is on property talk pages can hopefully soon be moved to a large extend to statements on the property page. The students' project to improve constraints reports is making good progress and can hopefully go live soon in a first version. That being said the Flow developers have this on their radar and I have seen a version that offers a space for this as well. If I understood Danny correctly this will be out in the next week or two. --Lydia Pintscher (WMDE) (talk) 18:46, 25 May 2015 (UTC)
I wonder if we should try to make editing items closer to the way talk pages are edited (with or without flow). It seems that some are at ease filling up talk pages, but seem to have difficulties adding statements to items. --- Jura 06:30, 26 May 2015 (UTC)
  Oppose at least until phab:T93883 is resolved. Wikidata is not a test site. --Ricordisamoa 10:04, 26 May 2015 (UTC)
The first point is fair. I'll bring this up with the Flow team. The second point: Please remember how much Wikidata has and still does depend on other projects being willing to try something new. Without projects being willing to test what we have we would not have gotten anywhere. --Lydia Pintscher (WMDE) (talk) 19:30, 26 May 2015 (UTC)
I like experimenting — and I've never archived any of my talk pages in the hope for a single pre-Flow archive! But I also recall the disappointing way the WMF has been treating the mediawiki.org community. --Ricordisamoa 12:56, 27 May 2015 (UTC)
@Ricordisamoa: I guess the questio now is How do we get out of this situation … We should get out of this chrisis by the up. TomT0m (talk) 13:50, 27 May 2015 (UTC)
@TomT0m: I'm sorry, I don't catch the metaphor. --Ricordisamoa 14:42, 27 May 2015 (UTC)
@Ricordisamoa: I just think being upset with the foundation is a poison in the long term. This can't last forever :) TomT0m (talk) 15:19, 27 May 2015 (UTC)
Indeed, the WMF should realize at some point that managing a company and managing a non-profit where content is created by unpaid volunteers are two very much different things. So far they did not demonstrate this understanding.--Ymblanter (talk) 15:39, 27 May 2015 (UTC)
@Ymblanter: Of course, and personally nobody will force me to do anything on any Wikimedia site … But we must realize a community of that size is not really easy to move. They try hard to make the experience better for everyone, but their always will be unhappy people on the moment (and there always has been). The risk is that things stays forever what they are, I understand that they want to speed up things a little sometimes, at the risk of making mistakes. Here I see nothing wrong in Lydia asking here first, why would we still refuse if they try to do better this time ? TomT0m (talk) 15:51, 27 May 2015 (UTC)
I have participated in the FLOW discussions in the English Wikipedia from the very beginning. The modality was always that the decision has been already taken, and adding FLOW to all talk pages is imminent, they just want to negotiate when and how it is going to happen. This is not the way of talking to the volunteer community.--Ymblanter (talk) 18:45, 27 May 2015 (UTC)
@Ymblanter: Well, community has always took care (mostly) of the content, and the foundation has always worked (mostly) to make hardware and software available … The discussion system has always been a pain (inconsistend indentations, bullet points, {{Outdent}}, no notifications of answer … Now that we have something better we will refuse it because we're grumpy ? Come on … That's exactly the reason the foundation had to act like this  . TomT0m (talk) 18:54, 27 May 2015 (UTC)
If I understand you right, TomT0m, the french community accepted all software changes powered by WMF without discussion? --Succu (talk) 18:59, 27 May 2015 (UTC)
@Succu: Sorry what ? I don't think you understand right. I don't speak for the french community. And yes, they're grumpy as well, this is what I'm telling. TomT0m (talk) 19:17, 27 May 2015 (UTC)
I personally find that the existing system is way more convenient than Facebook (and it has been made for discussion of serious business, for which Facebook is absolutely not suitable). But this is just my opinion and an opinion of several hundreds more long-time contrbutors. I do not oppose FLOW as such (though I am convinced it is hopeless), I strongly oppose the fact that the decision to install FLOW (and, for a reminder, there is no opt-out, unlike VE) was taken without any consultation with the community. If FLOW becomes the only way to communicate, I will leave the WMF projects.--Ymblanter (talk) 20:05, 27 May 2015 (UTC)
@Ymblanter: You would leave projects who aims to allow everybody to access knowledge freely for …_a change in the discussion system ? Yeah, that's as good as facebook, for sure :) Personnaly I just miss something to go to the parent comment in one click. What do you miss, seriously ? TomT0m (talk) 20:13, 27 May 2015 (UTC)
I miss a minimal standard of respect for the editing community. In 2008? 9? I participated heavily at the (now closed) Strategy Wiki, where we had to use Liquid Threads. This is not an experience I really want to repeat. And, yes, I have already heard that FLOW is conceptually different from Liquid Threads. I know. I have Facebook experience as well. Yes, I will leave because of the change of the discussion software.--Ymblanter (talk) 20:32, 27 May 2015 (UTC)
@Ymblanter: Please be more specific, what don't you like exactly ? As it it is not useful output … Imagine you're a flow developper, how would you answer to this ? I get they can't really answer something like this. TomT0m (talk)
It does not work like this. I tried to discuss this with Oliver when he was still in charge; indeed he wanted specific bugs and promised to fix them. No, I am opposed to the whole concept. I believe it is not fixable.--Ymblanter (talk) 20:56, 27 May 2015 (UTC)
@Ymblanter: Why not try to give feedback on the test page ? Feedback wanted and the page has already changed since yesterday thanks to our feedback … I don't know how it worked, but this is how it works now :) TomT0m (talk) 21:02, 27 May 2015 (UTC)
In my opinion, Wikidata is probably the single worst Wikimedia site for testing a new discussion system. Unlike the more public-facing projects, a lot of users who come here to post a question are already active Wikimedia contributors, who are frequently already confused by the parts of the interface that are different here by necessity, and had to come here to change something that impacts a page on their home project. My recommendation is that Wikidata should be the last project, or among the last projects, to have Flow enabled. This community is probably one of the more flexible communities when it comes to testing new software, and we can handle early tests here in a lot of areas, but the discussion system is not one of these areas. --Yair rand (talk) 20:49, 27 May 2015 (UTC)
  •   Strong oppose You don't fix things that aren't broken. The problems that Flow is supposed to fix aren't really even prevalent here, just on enwiki. I haven't seen an unsigned post by a newbie in a while, and people seem to communicate fine using the current system. Wikidata's not ready for Flow, and Flow is not ready for Wikidata. If Flow is used here, we need to fix several bugs stated above before the implementation. Wikidata, being only two years old, has enough problems to be sorted out, don't add Flow to that. --AmaryllisGardener talk 01:39, 28 May 2015 (UTC)

Hi everybody -- I'm sorry it took me so long to come back here to respond to all the questions and ideas; I was coming back home from the hackathon. I've read everything here, and on Wikidata talk:Flow and Flow tests. There's a lot to respond to all at once, so I'm going to do my best, and we'll see how it goes.

First, the general concerns about English WP, bad faith, aggressive rollout plans, Facebook and fixing things that aren't broken. I believe that wikitext talk pages are not the best tools for the job we expect them to do. I've been working on wikis since 2005 -- mostly on Wikia, where I started Muppet Wiki, fell in love with collaborative writing, and then allowed it to take over my life. I know what talk pages are for, and they're just not very good at this.

For example, look at the conversation that we're having right now. It's a sprawling mess of text, with overlapping subthreads running into each other. It's hard to read and hard to respond to, there's no way for me just to put this section on my watchlist, and I have to essentially reboot the conversation with an outdent in order to move the discussion forward. There has to be something better than this.

I've been working on Flow since I came to WMF, a little over a year ago. I know that the product designer who started the project a few years ago was kind of a loudmouth, and he burned a lot of bridges with a lot of people, saying all of the over-aggressive things that @Ymblanter: said. That was a dumb way of doing things, and it got this project off on the wrong foot. All I can say is that he's not at the Foundation anymore, and I'm not him. I don't think that Flow is an unstoppable force that we're going to roll out with a steamroller. I just think it's a good idea, and the communities that use it are going to benefit a great deal from it.

Right now, I'd say we have about half of the features that we're going to need in order to deserve full deployment on every talk page on every wiki. There's a lot more work coming up. But we've got a really strong core product by now, and the communities that are currently using Flow on real pages are the ones that we actually listen to. Trying out Flow on Wikidata is a chance for our team to learn more about what your community needs, and it's a chance for you to influence how the final product is going to work. That's what we're doing here right now.

Okay, moving on to specific questions and ideas:

@TomT0m: about space for documentation: We just released a change today that puts the header info into a side rail, where you can have all the important information at the top of the page and still have a clear space for the discussions. As an extra benefit, the side rail has a toggle so you can hide it and make the discussion full width if you want.

@Nikki: Thank you for the detailed feedback about the VisualEditor/source input, and the timestamps. If I manage to survive writing this post, I'm going to go make a whole bunch of bug tickets and questions for our UX designer based on your feedback. It looks like @Pasleim: and @Visite fortuitement prolongée: had similar problems recognizing what the "switch to source" toggle was supposed to be. That's very helpful.

@Sjoerddebruin: We treat topic titles as text-only, so putting a template in the title just posts as text, as you can see in this thread. One reason why we do this is that the topic title itself is used as a link in watchlist, recent changes and contributions, so even if we rendered that "Combat Flight Simulator" template as a link in the topic header, it would still have to be a link to the topic page when it's on the watchlist. Also, people can use templates (accidentally or on purpose) to get around the character count limit, so you could basically transclude an entire page into the topic header. People have done that, and it's super annoying. I'd suggest putting the link in the first message instead.

Another response for TomT0m: We've had a quoting feature hovering in the backlog for the last year, but people haven't been asking for it very much. It's hard to put a quoting feature into a wikitext reply field, because you have to use templates or divs to handle rendering the quotes, and if somebody messes up the div then the whole thing breaks. So basically, we're using the indentation for sub-threads and the mentions module for pings.

@Ricordisamoa: Speaking of indentation -- phab:T93883 is resolved. There's no way for a structured discussion system to precisely match the cultural practices around indenting in wikitext conversations, and as I said above, I don't think that's a worthwhile goal. The model that we have now has had a much better response overall compared to the old system. It's time for us to use what we've got, and move on.

@Jura1: Thanks for the note about Whatlinkshere; I'll file a bug ticket for that too.

And... there's probably a bunch of things that I've forgotten to mention, but this message needs to end sometime, so it might as well be now. Let me know what you think; I'm really happy to be having these conversations with you. DannyH (WMF) (talk) 01:37, 29 May 2015 (UTC)

Would it help to put it on test.wikidata.org? That would be a less drastic step and allow for people to get a better sense of what it would look like before everything is converted on a large production site, irreversably. --Rschen7754 04:34, 29 May 2015 (UTC)
@Rschen7754: We already have the two test pages I linked in my original post here. But if you think it'd make a difference I guess Danny can also enable it for (some) pages on test.wikidata.org. --Lydia Pintscher (WMDE) (talk) 06:32, 29 May 2015 (UTC)
@DannyH (WMF): The posts don't appear to be in namespace 2600 (at least, when looking at user contributions. @Innocent bystander: mentioned it.
Another issue with the user contributions page is that all posts appear if one ticks both "Only show edits that are latest revisions" and "Only show edits that are page creations". --- Jura 05:37, 29 May 2015 (UTC)
I cannot claim to have understood this discussion, but one thing is very clear: the documentation on Talk page cannot be moved as statements into the items in the forseeable future. We are still trying to get to the stage of being able to express basic properties in statements. The whole of Wikidata is still at the stage of being a workaround. - Brya (talk) 05:55, 29 May 2015 (UTC)
Template:Strong Support. Like Ymblanter I used Liquid threads on the Strategy Wiki and it was loads better than what we are using here. Automatically following threads you have contributed to in the past! Wouldn't that be great here when a newbie responds to a thread on an item talk page six months after and I get taken straight to that thread so the newbie can get a quick response? What about this page? Imagine seeing active threads? This would be so good. Filceolaire (talk) 06:54, 3 June 2015 (UTC)
I used Liquid threads and supported the installation of it on svsource. The problem was that the Extension was so full of bugs that we stoped using it. Almost every MediaWiki-update made the Village pump useless for some time, and we had to use the talk_page of the Village pump. One larger bug I remember was that the threads suddenly changed namespace. A few bugs are not a big problem, when you know that they will be fixed. The problem with Liquid threads was that it had very poor support from the developers. -- Innocent bystander (talk) 08:16, 3 June 2015 (UTC)
I remember that LT made my watchlist completely useless, since every new message in every single thread went there.--Ymblanter (talk) 09:05, 3 June 2015 (UTC)

Update: Danny started a tracking ticket at phabricator:T100844 for the issues/wishes we raise. --Lydia Pintscher (WMDE) (talk) 13:41, 3 June 2015 (UTC)

Mark as patrolled.js

I have just noticed that Matěj Suchánek suggested a month ago that User:Petr Matas/Mark as patrolled.js could become a gadget. If this is going to happen, I suggest the page to be moved from my user space directly into the MediaWiki namespace. Petr Matas 09:27, 2 June 2015 (UTC)

@Petr Matas: I've moved the gadget to MediaWiki:Gadget-Mark as patrolled.js. Old users of that gadget should either re-enable it in their preferences or you might consider to create User:Petr Matas/Mark as patrolled.js again and include the gadget in that page. -- Bene* talk 10:55, 2 June 2015 (UTC)
Uh, really nice and helpful. Are there some tools which adds exclamation marks and [Mark as patolled] links to the Contributions list and the Hiostory as well? --MGChecker (talk) 12:49, 2 June 2015 (UTC)
@MGChecker: I don't know about any such gadget. This would be a bit more difficult, because it seems that Contributions and History do not contain the unpatrolled flag. Therefore the gadget would need to do some extra queries to the server to get the needed information. It would be much easier to put all this directly into the MediaWiki. Petr Matas 20:07, 2 June 2015 (UTC)
@Bene*: Oops, I thought that a redirect would be created... Is it possible for scripts? For now, I recreated User:Petr Matas/Mark as patrolled.js and included the new location there. Petr Matas 20:07, 2 June 2015 (UTC)
Afaik this is the only way to "redirect" a script. Not very nice but still better than breaking existing uses of this tool. -- Bene* talk 20:10, 2 June 2015 (UTC)
Thanks :-) Petr Matas 20:25, 2 June 2015 (UTC)


Potential merge required

There are two items Auguste-Joseph Desarnot (Q867840) and Auguste-Joseph Desarnot (Q18507887), which might reference the same person. However, the first is used for interwiki and the second is externally referenced in VIAF. What to do in such case? Paweł Ziemian (talk) 16:04, 3 June 2015 (UTC)

Merge and redirect. --- Jura 16:05, 3 June 2015 (UTC)
  Done--Ymblanter (talk) 16:14, 3 June 2015 (UTC)

Statistics about Wikipedia content

I don't really know what to do with it, but many people seem to be interested about statistics, and I have done a few about Wikipedia content at User:Zolo/Wikipedia content. As it is mostly done by hand, I have mostly focused on two Wikis (fr and zh). --Zolo (talk) 13:40, 2 June 2015 (UTC)

Good work. Most interesting! --- Jura 13:50, 2 June 2015 (UTC)
  • (doing the numbers of Wikipedia_content#Living_people_by_nationality for Wikidata):
    Didn't know that France had almost twice as many biographies about living people compared to the US (on per capita basis). Or do they just lack P27? --- Jura 14:08, 2 June 2015 (UTC)
  • cebwiki is close to perfection as almost all their articles have P31 defined ;) --- Jura 14:08, 2 June 2015 (UTC)
  • gender is defined for all but 268 items of articles about people in frwiki (369904) --- Jura 14:19, 2 June 2015 (UTC)
  • of which I treated nearly half, and most were not human (Q5) but errors added by Dexbot/Kian… :( --Hsarrazin (talk) 23:57, 3 June 2015 (UTC)
I have added a few stats and modularized some of the so that you can now get statistics about a project this way: {{Site content|enwiki}}
(5,045,856)
human: 1,325,136 (26.3%)taxon: 315,294 (6.2%)administrative territorial entity: 563,681 (11.2%)architectural structure: 237,432 (4.7%)occurrence: 201,216 (4.0%)chemical compound: 19,682 (0.4%)film: 119,346 (2.4%)thoroughfare: 37,541 (0.7%)astronomical object: 26,683 (0.5%)Wikimedia list article: 105,702 (2.1%)Wikimedia disambiguation page: 302,981 (6.0%)artificial entity: 434,920 (8.6%)other P31/P279: 521,992 (10.3%)no P31/P279: 834,250 (16.5%) 
  •   human: 1,325,136 (26.3%)
  •   taxon: 315,294 (6.2%)
  •   administrative territorial entity: 563,681 (11.2%)
  •   architectural structure: 237,432 (4.7%)
  •   occurrence: 201,216 (4.0%)
  •   chemical compound: 19,682 (0.4%)
  •   film: 119,346 (2.4%)
  •   thoroughfare: 37,541 (0.7%)
  •   astronomical object: 26,683 (0.5%)
  •   Wikimedia list article: 105,702 (2.1%)
  •   Wikimedia disambiguation page: 302,981 (6.0%)
  •   artificial entity: 434,920 (8.6%)
  •   other P31/P279: 521,992 (10.3%)
  •   no P31/P279: 834,250 (16.5%)
Module:Statistical data/by project/classes, 2016-01-10

.

--Zolo (talk) 08:25, 3 June 2015 (UTC)

@Zolo: Is a div-tag or something missing in that template? The threads below this looks strange! -- Innocent bystander (talk) 08:33, 3 June 2015 (UTC)
@Innocent bystander: no idea... Apparently, putting the template on a new line fixes the issue (but not adding a new line directly in the template code). --Zolo (talk) 08:48, 3 June 2015 (UTC)
Forgot to say that if you wanted to add data, it can be done in Module:Statistical data. --Zolo (talk) 11:32, 3 June 2015 (UTC)

Wikidata-Gadgets on Phabricator: a central place for gadget bugs and ideas

Hey everyone,

Some may know the following problem: a gadget becomes broken due to changes to the software. You report it on the talk page of the gadget, then after some time without response you discuss it on our project chat. The time between reporting the bug and fixing the bug are weeks and sometimes months! Gadgets are a important part of the Wikidata experience and bugs with them should be fixed as soon as possible.

After some discussion with Lydia Pintscher (WMDE), I've decided to start a project on Phabricator. Here can we track bugs, post some wishes and even more. Here we can get a easy overview of the current state of gadgets on Wikidata.

You can find the project here. Feel free to add ideas or report bugs.

Greetings, Sjoerd de Bruin (talk) 17:02, 2 June 2015 (UTC)

@Sjoerddebruin: Please add a link and an explanation perhaps on Wikidata:Tools about your project and how to participate. This will be better to find your project in the future. Snipre (talk) 18:39, 2 June 2015 (UTC)
Yeah, working on it. I've already added a notice above the gadgets preferences pane. Sjoerd de Bruin (talk) 18:40, 2 June 2015 (UTC)
Yeah, one more Phabricator page !… just one small problem… just can't understand how this works :(
could you please provide a page explaining how this tool (Phabricator) works, please… this is just chinese to me… I guess it is mostly useful for devs, but if users have to use it to report a problem, it would be nice to understand how to
Thanks for all :) --Hsarrazin (talk) 21:19, 3 June 2015 (UTC)
PS : for example, weeks ago, I asked about ws badges, and was answered that task was on Phabricator. I cannot understand, by reading this page, where is the project, nor understand what's the "blocking task" ? could someone explain, please ?

Bot operators: breaking API change

Hey folks :)

MediaWiki Core is getting a breaking API change at the end of the month. This is not specific to Wikidata but since we have quite a few bots here I wanted to let you know. If you're running a bot please read https://lists.wikimedia.org/pipermail/mediawiki-api-announce/2015-June/000084.html. --Lydia Pintscher (WMDE) (talk) 13:44, 3 June 2015 (UTC)

@Lydia Pintscher (WMDE): Does this apply to wikidata changes too? I tried [5] and I see no warning, so this means nothing changed for this API? --Laboramus (talk) 19:54, 3 June 2015 (UTC)
It should apply to all our wikis, yes. And the list of bots given in the email had quite a few familiar names from here. --Lydia Pintscher (WMDE) (talk) 19:55, 3 June 2015 (UTC)

Temps/time & écart/gap

Bonjour. Je pose la question ici afin d'être certain. Est-ce qu'il existe une propriété similaire à start time (P580) qui me permette d'indiquer premièrement le temps (exprimé en heures, minutes et secondes) et deuxièmement un écart (lui aussi exprimé en heures, minutes et secondes). Je vais en avoir besoin pour mes classements de cyclisme.
[Google Translate] Hello. I ask here to be sure. Is there a property similar to start time (P580) that allows me to indicate first time (in hours, minutes and seconds) and secondly a gap (also expressed in hours, minutes and seconds). I'll need it for my rankings in cycling. Jérémy-Günther-Heinz Jähnick (talk) 15:20, 3 June 2015 (UTC)

The number with dimension datatype isn't available yet but is due to be with us soon. This will allow us to have properties with values in hours/minutes/seconds. See Wikidata:Property_proposal/Pending/2#Temporal_interval for properties which have been approved but are on hold waiting for this datatype. If none of these are what you need then you can propose other properties with this datatype and if approved they can be added to this page while we wait for the developers to complete this datatype. Filceolaire (talk) 01:39, 4 June 2015 (UTC)
Thank you @Filceolaire:. There is new progress with the use of Wikidata on the French Wikipedia. Jérémy-Günther-Heinz Jähnick (talk) 07:41, 4 June 2015 (UTC)

http or https?

Recently, formatter URI for RDF resource (P1921) was added (which is awesome, yay!) and the first instance I see of its usage is, naturally, Wikidata (Q2013) with the value of "https://www.wikidata.org/entity/$1". Now comes the question - should the canonical URL be http or https? I would prefer canonical URLs to be http, even though I use https while browsing, since not all clients can do https, yet less of them can do it right (up-to-date certificate stores, proper verification, SSL bugs fixed, etc.) and in general mixing canonical URLs with protocol issues doesn't look good to me. So, any thoughts on this? --Laboramus (talk) 19:26, 3 June 2015 (UTC)

Can't we omit these and leave it to the browser to negotiate whether to use http or https? Filceolaire (talk) 01:33, 4 June 2015 (UTC)
Not every tool is a browser. So there should also be a canonical URL. We can omit the http(s) prefix but then tools will have to decide, and if they decide differently same things get different IDs. Not good. So better to choose something. Note this is resource URL, which does not have to be real page URL (i.e., wikidata one redirects). --Laboramus (talk) 07:31, 4 June 2015 (UTC)

Maintaining the dearly departed

Hoi, for 2014 and 2015 I have spend a lot of effort on maintaining the date of death of people who died that year. From now on I will no longer do this. I do not have the time and I find that my time is spend more effectively in other ways. I find that sources like DBpedia do have that information available and we only need to make use of the service that is available to us. They will do a better job because they will not make the typos I do.

There is no reasonable argument why we cannot import their data after all their data is what Wikipedia tells the world. Thanks, GerardM (talk) 10:39, 4 June 2015 (UTC)

Because now Wikipedia tells the truth ? I miss something. And you never wonder why WPs were not using DBpedia to complete their data ? Snipre (talk) 20:39, 4 June 2015 (UTC)
Yes, you miss something. We have been importing Wikipedia data bigtime by hand and as a consequence add typos all the time. We are NOT updating Wikipedia here this is Wikidata. What we would do is two things. Add the data we do NOT have and compare the data we do have with the data they hold. This is imho a safe way to improve Wikidata easily and obviously. Choosing to do it by hand is not a testament of understanding our issues. GerardM (talk) 05:49, 5 June 2015 (UTC)

"instance of" alternative to draw a line between "wiki engine installation" and "wiki software"

I am trying to find all wiki engines, so I run query "instance of wiki software" and instead of getting the list of engines, such as MoinMoin and MediaWiki I get list of well known wiki installations. How to deal with this problem? --Techtonik (talk) 10:55, 4 June 2015 (UTC)

I think you're mixing up wiki (Q171) and wiki software (Q6686945)... --Yair rand (talk) 10:58, 4 June 2015 (UTC)
Aha! There is a bug in https://wdq.wmflabs.org/wdq/ that update Autolist query after you edit colored boxes on the page. And also a problem with Autolist that doesn't allow to see the text of the query. --Techtonik (talk) 11:07, 4 June 2015 (UTC)
Applied your patch (thanks!). --Magnus Manske (talk) 08:27, 5 June 2015 (UTC)
The patch is not enough to completely fix the issue if parser failed. It looks like the form uses the same method to update both query field and blocks from this.structure, but this.structure can be invalid if the query did not parse successfully. So, the solution may be to avoid storing broken this.structure if query parsing failed, and store error condition somewhere to be rendered. --Techtonik (talk) 10:23, 5 June 2015 (UTC)

next office hour on IRC

Hey folks :)

It's been a while since the last office hour. We'll do the next one on June 19th at 16UTC in #wikimedia-office on Freenode. See http://www.timeanddate.com/worldclock/fixedtime.html?hour=16&min=00&sec=0&day=19&month=06&year=2015 for your timezone. Looking forward to seeing many of you there!

Cheers --Lydia Pintscher (WMDE) (talk) 12:19, 5 June 2015 (UTC)

Pywikibot compat will no longer be supported - Please migrate to pywikibot core

Sorry for English, I hope someone translates this.
Pywikibot (then "Pywikipediabot") was started back in 2002. In 2007 a new branch (formerly known as "rewrite", now called "core") was started from scratch using the MediaWiki API. The developers of Pywikibot have decided to stop supporting the compat version of Pywikibot due to bad performance and architectural errors that make it hard to update, compared to core. If you are using pywikibot compat it is likely your code will break due to upcoming MediaWiki API changes (e.g. T101524). It is highly recommended you migrate to the core framework. There is a migration guide, and please contact us if you have any problem.

There is an upcoming MediaWiki API breaking change that compat will not be updated for. If your bot's name is in this list, your bot will most likely break.

Thank you,
The Pywikibot development team, 19:30, 5 June 2015 (UTC)

Wikidata and credibility

I blogged about the credibility of Wikidata. Wikidata is credible when it fulfils a few needs:

  • accurate
  • unbiased
  • timely
  • complete

I have argued it before, this is done by comparing our data with the data from other sources. This is done by completing our data where we have none. This is done by giving priority to keeping our data up to date.

I stopped adding the date of death because I was told that we should wait using the information from DBpedia because a tool might magically appear in the near future. We have seen how long it takes for basic functionality to appear (think query and units). I do not have the time for such nonsense. I feel it an insult to volunteer time; we are taken for granted.

The Wikipedia approach to sources if fine for Wikipedia. It adds no value to Wikidata. Researching the differences that exist between sources like VIAF, DNB, IMDb is how we use our community in a positive way. It makes a positive difference and it does not provide us with more drudgery. Thanks, GerardM (talk) 06:36, 6 June 2015 (UTC)

Patrol

I've been searching for a while, but didn't find anything. Is there a page, that documents the way the patrol system on wikidata works? -- Dr.üsenfieber (talk) 15:58, 2 June 2015 (UTC)

I set up a draft on Wikidata:Patrol --Pasleim (talk) 16:31, 7 June 2015 (UTC)

Linking bug with "instance of" vs "subclass of"?

I am trying to get all wiki engines with the query "is instance of wiki software". But along with wiki engine names I also get https://www.wikidata.org/wiki/Q638153 which is wiki software type, not a wiki software instance. Also, on its page there is no "instance of wiki software" relation, only "subclass of wiki". Why it appears in the results? --Techtonik (talk) 11:20, 4 June 2015 (UTC)

Race condition. =) User:Sjoerddebruin fixed the page while I was looking on the results page with old info. There is no problem now. --Techtonik (talk) 11:42, 4 June 2015 (UTC)

/ @Techtonik: Thread about this opened in Wikidata:Requests_for_comment/Adopt_Help:Classification_as_an_official_help_page. It's an attempt to put good foundations for this kind of issues. For example, an instance of (P31) Wikibase could be ... this production site. Wikibase as a software can be considered as the class of all such instances as well as the instructions (source code) that runs these sites ... Once we will have sorted that out we will have clear guidelines for users and you will have a query that will work (and will serve for users to see if they did correctly) :) Comment there appreciated. TomT0m (talk) 10:12, 7 June 2015 (UTC)

Get all Python versions (with dates)

I probably miss something obvious, but.. I know how to query for the instance of Python programming language, but how to get the list of all its versions? And also a list of versions with release dates? --Techtonik (talk)

  • Try to build one and will have a look at it. --- Jura 11:12, 5 June 2015 (UTC)
    • There already is an (incomplete) one in Q28865. --Mineo (talk) 11:44, 5 June 2015 (UTC)
      • I don't see where? I try to build it like CLAIM[31:28865] AND STRING[P348:""] meaning "find all Pythons Q28865 and get all properties named `version` P348 as a string", but the query is invalid and I don't see any other example that I can understand. --Techtonik (talk) 14:45, 6 June 2015 (UTC)

Switching items and migrating subjects

Just to share my amazement, it seems that sometimes the meaning of an item switches completely, while a subject can migrate several times from one item to another. Q3870479 was originally (January 2013) about a painting by Artemisia Gentileschi depicting John the Baptist just after his birth. On the other hand, Q6980151 (created March 2013) was about Johns birth. But in March 2014, someone moved ('merged') the sitelinks and labels of the latter item into the former, while at the same time creating a new item Q15940081 for the two sitelinks about the painting. In the next months a few people were busy with adding relevant claims to the new painting item, while these were kept in Q3870479 at the same time, making Artemisia Gentileschi the mother of Saint John, or something like that. In July, someone discovered the duplication and merged Q15940081 back into Q3870479. But in December, another person discovered that this was a mistake, a painting and a birth (or religious feast day) not being the same topic after all, and created yet another item for the painting, which now resides at Nativity of St. John the Baptist (Q18602467), although (s)he forgot to move the Italian sitelink. Bever (talk) 04:46, 7 June 2015 (UTC)

Yes I have seen similar confusing behavior as a result of good-faith edits and it takes time to repair it. Sometimes it is best to just untangle what is there and leave the history alone. Since Wikidata is so young I don't think there will be that many incoming links that will get messed up. This problem is a common one occuring for items that did not get many claims at the time of creation (probably only the sitelink). To prevent this from happening, the only solution is to add as many claims as possible so that people don't merge items incorrectly. Jane023 (talk) 09:55, 7 June 2015 (UTC)

About the relevance of "solving" redirects wrt. such informations

@Ivan A. Krestinin: These kind of big messes proves that these kind of wrong merging are not found quickly. One of the claimed properties of Wikidata is the stability of the IDs ... TomT0m (talk) 09:47, 7 June 2015 (UTC)

I do not see any redirect resolving edits in the discussed items history. IDs stability is good thing, but sometimes humans redefine items or properties. Often this happens while extending or specializing scope. So full IDs stability is not reached in current Wikidata stage. — Ivan A. Krestinin (talk) 10:23, 7 June 2015 (UTC)
You state the obvious ... It's still a claimed property and we should do whatever we can to achieve that. I think your automatic redirect solving is not a good idea in that context. It will amplify the mess created by such situations. TomT0m (talk) 10:34, 7 June 2015 (UTC)
Unresolved redirects creates bigger mess by current Wikidata engine implementation unfortunately. — Ivan A. Krestinin (talk) 10:59, 7 June 2015 (UTC)
I don't know any tool which would break by unresolved redirects. -- Bene* talk 11:06, 7 June 2015 (UTC)
Any information import bot for example, because it can not compare added value with existing value using simple way. — Ivan A. Krestinin (talk) 11:10, 7 June 2015 (UTC)
@Ivan A. Krestinin: would not that be solvable by an option to resolve redirect when you ask Wikidata for datas ? TomT0m (talk) 11:16, 7 June 2015 (UTC)
Yes, but please note that API is not the single data access point, Wikidata dumps have the same issue. — Ivan A. Krestinin (talk) 11:23, 7 June 2015 (UTC)
@Ivan A. Krestinin: That potentially also be done for dumps, and for dumps it's enough that the toolkits APIs does the work. @Lea Lacroix (WMDE): @Markus Krötzsch: Any comment about this issue ? TomT0m (talk) 11:50, 7 June 2015 (UTC)
We'll need to make this work better there as well, yes. --Lydia Pintscher (WMDE) (talk) 14:15, 7 June 2015 (UTC)
Also the UI does not fully support redirects. If a claim value is a redirect, I see only the Qid of the redirect instead of the label of the redirect target. And there is also WDQ which does not return results if a redirect is involved. --Pasleim (talk) 12:39, 7 June 2015 (UTC)
Yeah I noticed this the other day as well. Opened phabricator:T96553 for it. --Lydia Pintscher (WMDE) (talk) 14:15, 7 June 2015 (UTC)

Unknown margin of error

When a source's margin of error is unknown, how should this be represented on Wikidata? As ±0? --Yair rand (talk) 20:42, 1 June 2015 (UTC)

I don't think there is a general answer; it will depend on the particular property and what provisions the data model has for representing uncertainty in that property. If the value is stored as a number, such as integer or floating point, it would be impossible to enter an uncertainty unless fields are provided for that purpose. If the value is a character string, it would be possible to enter an uncertainty, but applications that are not expecting an uncertainty would not parse the string correctly.
I think "±0" would be a poor way to express an unknown uncertainty; I would interpret an uncertainty of "±0" as meaning the value is exact, as in "1 international inch = 2.54 cm ±0 cm". Jc3s5h (talk) 21:50, 1 June 2015 (UTC)
±0 would be bad. It is OK to have a too large margin - so maybe use the same order of magnitude as the value itself. If the number is 1000, use +- 500, if the number is 1120, use +-10 or something. Note that we do not have to be as exact as the reference supports, but a wider answer as long as it is supported by the reference. Or, put differently, correctness beats accuracy. --Denny (talk) 22:07, 1 June 2015 (UTC)
Suppose the statement in question is the CIA World Factbook saying that the population of Mexico is 120,286,655, without any specific margin given. (I probably should have waited for an answer here before adding that...) There isn't any margin that's obvious from the number, and adding any particular margin would probably be quite arbitrary, and as a result likely inconsistent from statement to statement. --Yair rand (talk) 22:26, 1 June 2015 (UTC)
Don't we have any property for the method of counting people? I guess the numbers of censuses often are exact based on the method used. Here for example, censuses are almost always based on the resident registration (Q254213). resident registration (Q254213) can be something different than the number of people actually living in a place. There is illegal immigrants, people (illegaly) choosing registration based on prefered tax-levels, people living in a care-centers for old people, but still registred in their house, military on international service, weekend-commuters etc etc. The method in such case have flaws, but the number is still exact. -- Innocent bystander (talk) 06:38, 2 June 2015 (UTC)
@Innocent bystander, Yair rand: See determination method (P459). Snipre (talk) 07:35, 2 June 2015 (UTC)
phab:T68580 is relevant to your question as at least one comment (mine, though there are others agreeing) there should be an "unknown" precision. --Izno (talk) 18:54, 2 June 2015 (UTC)

For a population number like 120,286,655, I would understand that the number is indeed meant to be exact. How else should it be interpreted? If it was a scientific source, it would offer a margin of error, but for such a popular source like the CIA fact book, I don't see any other interpretation being likely.

I was wrong in what I said before: I didn't mean the same order of magnitude as the number, I meant the order of magnitude of the last significant digit. This seems a pretty decent guess, if no other margin of error is given. --Denny (talk) 04:22, 3 June 2015 (UTC)

Agree, but it is difficult to know if the number 120,286,000 has 6, 7, 8 or 9 significant digits. When it occur in a list like: a = 123,456,789, b = 234,567,893, c = 345,678,901, d = 987,654,000, e = 456,987,123, it's a decent guess that d is exactly as exact as a, b, c and e. But when you see numbers like: Norway = 4,123,456,789, Belgium = 10,584,534, Mali = 14,517,176, then I would not make that assumption, since those numbers, obviously, are not measured in the same way. -- Innocent bystander (talk) 05:09, 3 June 2015 (UTC)
  • Further question: If a source is obviously rounding all of the values to the nearest thousand, should the margin be ±500, or ±1000, or what? --Yair rand (talk) 20:38, 7 June 2015 (UTC)

Wikidata weekly summary #161

How to extract events/role information form WikiData ?

Hi,

I would like to know, how can I extract events/ role information from WikiData.

Regards, Raj

What kind of information do you want to extract ? Please be more accurate and provide examples. Snipre (talk) 07:59, 8 June 2015 (UTC)

Merge

I merged two items and it seems as if the links did not move. What did I do wrong.. Thanks, GerardM (talk) 10:13, 6 June 2015 (UTC)

To me the merge looks fine. If you wanted to move the content to Q20049277, you would have had to uncheck "Always merge into .." when merging from Q3323451. --- Jura 10:53, 6 June 2015 (UTC)
Is it possible to do this later ? GerardM (talk) 12:38, 6 June 2015 (UTC)
Not without reverting both edits. --- Jura 08:59, 7 June 2015 (UTC)


Ok so I removed them and added new ones. Thanks, GerardM (talk) 13:30, 8 June 2015 (UTC)

alphanumeric listing software

I would like to find the alphanumeric software like on the music page in order to speed up my searches for either music or document files. All help is appreciated. Thanks, Paul

??? Please explain better what you want. Snipre (talk) 07:58, 8 June 2015 (UTC)

Global Replace will now update file usage on Wikidata under individual user accounts

JFYI and in case you experience any trouble, let me know. Here is the article I wrote for The Commoner:

The procedure of replacing one file name where the file is used with another file name across all Wikimedia Projects is called Global Replace. There are various reasons for replacing usage globally. One of them and probably the most important one is to keep articles updated when files have been renamed, until the server software, MediaWiki, which powers Wikimedia Commons and all its sister projects is able to accomplish this task on its own. Almost all gadgets and scripts at Wikimedia Commons use an open-source JavaScript library for globally replacing files, c:MediaWiki:Gadget-libGlobalReplace.js. For the curious, there are a Java-Tool source code and a bot source code, written in PHP, too.

The JavaScript library has been recently updated to support replacements at Wikidata. -- Rillke (talk) 17:15, 8 June 2015 (UTC)

"Instance of Wikinews article" doesn't work so well

Some items are claimed as instance of (P31) = Wikinews article (Q17633526), and this doesn't really make much sense.

For example, Solar eclipse of March 29, 2006 (Q934366) can't really be both a Wikinews article (Q17633526) and a solar eclipse (Q3887), can it?

In any case, we can assume there is a news article, if the link exists under the Wikinews links.

So, should these instance of (P31) = Wikinews article (Q17633526) claims be removed? Danrok (talk) 19:56, 6 June 2015 (UTC)

I think these items should be splitted. Too much errors like this happen. Sjoerd de Bruin (talk) 20:24, 6 June 2015 (UTC)
With a few exceptions, I wonder if it wouldn't be better if one couldn't add Wikipedia and (Wikinews or Wikisource) pages to the same item. --- Jura 20:29, 6 June 2015 (UTC)
The Wikinews article is about the same topic as the Wikipedia article, so they should be on the same item. In general, all claims instance of (P31) = Wikinews article (Q17633526) should be replaced by instance of (P31) = event (Q1656682) or a subclass of it. --Pasleim (talk) 21:03, 6 June 2015 (UTC)
I agree with Sjoerddebruin and Jura: these items should be split. (and to be honest, I think it was a mistake to add Wikidata items for Wikinews pages) Pichpich (talk) 01:10, 7 June 2015 (UTC)
No, Danrok and Pasleim are right, they should not be split but the claims about Wikinews articles should be removed since they make no sense. There are only rare occasions where a Wikimedia page is rightfully the subject of an item. People who add claims like that are confusing sitelinks, administrative claims, and topical claims. Perhaps this is caused by Wikidata having several goals in one project: being a repository for interwiki links, and a general database with information about the world at the same time. Bever (talk) 04:54, 7 June 2015 (UTC)
Surely there could be any number of news items about a particular topic, in which case there is no good solution? And, yes, this is problem for Wikisource as well (try translations of "The Raven"). - Brya (talk) 04:50, 7 June 2015 (UTC)
It looks like the bug who limited the number of interwiki from one project to another finally is solved. Then we maybe can start to use sitelinks to Wikisource as we originally intended to? -- Innocent bystander (talk) 05:58, 7 June 2015 (UTC)
The bug was not solved, it was a local bugfix I saw in svwikisource. -- Innocent bystander (talk) 10:03, 7 June 2015 (UTC)
Not sure what was intended, but isn't that going to make it worse? You would have lots of sitelinks to Wikisource/Wikinews on an item and you wouldn't be quite sure what it was about. --- Jura 08:53, 7 June 2015 (UTC)
I know very little about Wikinews and how they use interwiki. But in this version of the page Vårt land@svwikisource you see multiple interwiki to fiwikisource. There is no sitelink to fiwikisource in this item here. The interwiki is made by the help of edition or translation of (P629) and has edition or translation (P747). All sitelinks are in "cousins" to the item related to that page. These pages should not share items, since they have different metadata, but they used to be manually interwikilinked before Wikidata was initiated. -- Innocent bystander (talk) 10:03, 7 June 2015 (UTC)
Other than in the sense that there are also multiple Wikipedia articles about one topic, no. Each news article deals with one topic, and no two news articles in the same language have the same content or precise topic as each other. --Yair rand (talk) 12:53, 8 June 2015 (UTC)
I think Wikinews contributors don't understand how WD is working. They mix an event (the eruption in this case) with a wikinews article. We should have two articles one for the eruption and anoter one for the wikinews article with a link between these two items, typically statement is subject of (P805). Snipre (talk) 15:31, 7 June 2015 (UTC)
Maybe it's just ICCU who has a different view on it, looking at the sitelinks moved at Q76. --- Jura 15:39, 7 June 2015 (UTC)
@Snipre: I'm not a Wikinews contributors but still I think wikipedia:ru:Солнечное затмение 29 марта 2006 года and wikinews:ru:Солнечное затмение 29 марта 2006 года are both describing exactly the same event and therefore belong to the same item. Creating a separate item for the Wikinews article would mean that items do no longer represent real world entities but pages of the MediaWiki software. Of course we have to be careful if the Wikinews article doesn't completely match the Wikipedia article, e.g. Barack Obama elected 44th President of the United States (Q17595309) isn't about the same as 2008 United States presidential election (Q45578) and therefore two items are needed, however, all claims on Barack Obama elected 44th President of the United States (Q17595309) should be changed. --Pasleim (talk) 16:09, 7 June 2015 (UTC)
What is the proportion between the two cases at Wikinews? What is it for Wikisource? --- Jura 16:16, 7 June 2015 (UTC)
In Wikisource: Universal Declaration of Human Rights (Q7813) should not have any page in Wikisource connected to itself, except for s:sv:FN:s deklaration om de mänskliga rättigheterna who is a typical "Wikisource-disambig" about one single work (not to be comfused with other disambigs). Instead each Wikisource-page should be treated as "editions" of the text. Each such items should have statements about that specific edition. Since some of these texts are "original" in some of the "official languages" of UN, they should have statements describing them as that. Other editions should have statements describing who has translated them and when they were published. The svwikisource has two versions of this text in different pages and it should be possible for any project to link to them both, if they like to, just like svwikisource link to all four Finnish versions of the Finnish anthem. -- Innocent bystander (talk) 16:35, 7 June 2015 (UTC)
Observe that a page like s:sv:Korpen (Raven) not should be regarded as a "WS-disambig", but rather as an ordinary "WP-disambig", since the only thing the pages linked to in the page, has in common, is the title. If they all would have been related to "Poe's Raven", it should have been a WS-disambig and would have been linked to the WP-article about poem. -- Innocent bystander (talk) 17:04, 7 June 2015 (UTC)
So for Wikisource, the proportion would be 99 (no sitelinks to WP, only WS):1 (sitelinks to WP and WS "Author:", WS-disambiguation).
For Wikinews, is it similar? --- Jura 16:39, 7 June 2015 (UTC) --- Jura 16:54, 7 June 2015 (UTC)
In my opinion, Wikinews articles should be linked to Wikipedia articles about the exact same topic. That's what sitelinks are for. No items should have instance of (P31) Wikinews article (Q17633526). --Yair rand (talk) 20:35, 7 June 2015 (UTC)
I have just removed the two offending properties from Q934366. MSGJ (talk) 09:58, 8 June 2015 (UTC)
That doesn't fix the problem, a bot or user will add it again soon. Sjoerd de Bruin (talk) 10:04, 8 June 2015 (UTC)

I think that it doesn't really even make sense to link Wikinews pages to Wikidata. I think it's kind of like Wiktionary, nearly impossible to link. Sometimes there are multiple stories about one event. P.S. I think that Wikinews is a failure. Someone created a story titled "What is causing Australia's slower Internet speeds?", but I couldn't find a story about American Pharoah winning the Triple Crown. Makes no sense. --AmaryllisGardener talk 17:57, 8 June 2015 (UTC)

I never thought it would make any sense to try to use Wikidata for Wikinews articles. I was skeptical about using Wikidata for Wikinews topic categories, because different-language Wikinewses use very different topic hierarchies, and the results have been iffy imho but the big failure in that regard has been due to a mismatch between Wikidata's strict ontologies and the practical purposes of interwikis, which failure also causes problems for Wikipedia (but I've seen no sign that anyone at Wikidata wants to hear that, possibly because they have no idea what to do about it).

Realistically, the biggest problem with Wikidata — and I most certainly don't hold any Wikidatans personally responsible for this, it's just human nature — is that once Wikidata starts doing something, no matter whether it does it well, the local projects stop doing it, resulting in a lowering of standards across the board and corresponding long-term damage to the morale of all the projects.

Btw, somebody "created" a page titled "What is causing Australia's slower Internet speeds", or something close to that, but that's not coverage by Wikinews. They submitted it for review and it was assessed "not ready" with a detailed review comment explaining what was wrong with it. Afaict it was probably submitted by a university journalism student, part of a class of them who have been sent to Wikinews to learn because Wikinews has a serious review process. (I was kind of hoping somebody would cover the Triple Crown, even though Wikinews's function is not particularly to cover all the "big" stories; but, nobody chose to write about it.) --Pi zero (talk) 20:39, 8 June 2015 (UTC)

What do you mean by multiple stories about one event? Could you give an example of a situation where having multiple articles on the exact same topic would make sense? --Yair rand (talk) 23:22, 8 June 2015 (UTC)
@Yair rand: Actually we did have a case a while back with two Wikinews articles deliberately written about the same game, one from the perspective of the winning side and the other from the perspective of the losing side. (I'd have trouble finding it.) But mainly I've observed the phenomenon with topic categories on Wikinews, and with articles on Wikipedia; of course, Wikinews topic categories correspond to Wikipedia articles, and link from the Wikidata items of Wikipedia articles. Alas, the only examples I can conjure up are Wikinews topic cats, because I haven't been collecting examples where the same thing happens on Wikipedia. But here are a few examples from en.wn:
  • n:en:Category:Guantanamo Bay — one category encompasses the Bay, the US territory, the US naval base in the territory, and the detention center on the base. Wikidata would presumably want us to choose just one of those things to link to; let's say it's the bay. Suppose, plausibly, that everyone relied on Wikidata to do all their interwikis. We would have outgoing interwikis only to pages on other projects that those projects have chosen to link to the Guantanamo Bay Wikidata item, so if a given project happens to have made a different choice we wouldn't provide any interwiki for them (even though they might have a page for some other one of the four items involved). And we would have incoming interwikis only from those pages on other projects, even though it would be more useful all around to send interwikis to our page from pages on other projects associated with any of the four items — it might be appropriate to send interwikis to our category from multiple pages on a single project.
  • n:en:Category:Donetsk — covers both the city, and the oblast containing the city.
  • n:en:Category:Palestine — covers both the Palestinian territories and the Palestinian state. Does not cover the geographical area called Palestine. I gave up very quickly on even attempting to sort out how this really ought to be handled on Wikidata, because sorting it out would entail systematically doing damage to the various Wikinewses by depriving them of interwikis.
It's driving me up the wall that I encountered an example of this sort of thing happening on English Wikipedia, where they had a single article that covered multiple things that are ontologically separable, and I've forgotten what the example was. --Pi zero (talk) 04:39, 9 June 2015 (UTC)

Not sure how to add something "safely", part 2

Pinging Izno, Kolja21, and Gymel, who participated in the first discussion.

Going back to the Hilbert spaces discussion at Wikidata:Project chat/Archive/2015/05, I now wonder if we can add LCSH #85060803 to the Q190056 page. It's described at http://id.loc.gov/authorities/subjects/sh85060803.html, which can easily be found by going to http://id.loc.gov/authorities/subjects.html and searching for <Hilbert space>. I'm at an academic library, so it's possible that this is subscription-only. Could we add this link to Q190056? Nyttend (talk) 14:27, 9 June 2015 (UTC)

We could, but KrBot would spot it as a duplicate and remove it ;-) Actually, User:Kolja21 has already added the property during our first discussion. http://id.loc.gov and friends (http://lccn.loc.gov/ , http://authorities.loc.gov/ ) are public interfaces without restrictions, English Wikipedia has learned since how to suppress the non-working Worldcat links for LCSH numbers and here on Wikidata consensus has deepened that Library of Congress authority ID (P244) should be the appropriate place for LCSH numbers, not only the LCNAF ones.
IIRC the original discussion was focussed to the viability of the use of specifically the OCLC "ARN" numbers visible in the Connexion client and the conclusion was that they indeed represent LCSH records, but since they are neither publicly verifyable nor can be derived from some "known" number, and futhermore are of very limited use (keyword searches for that number in Worldcat yield mostly relevant results), recording them here (necessarily in the context of a property different from Library of Congress authority ID (P244)) seems not particularly thrilling. -- Gymel (talk) 15:41, 9 June 2015 (UTC)
Yes, the original was a request to add the authority record number, but I wasn't attempting to suggest that this time. I simply was asking for something like the "added the property" diff that you gave; I forgot to check to see if it were already there. Thanks! Nyttend (talk) 16:00, 9 June 2015 (UTC)

bilateral relations

Hi, there is some questions to be solved about the bilateral relation (Q15221623) items. UserːPopcorndude proposes to change country (P17) for participant (P710). This imply to enlarge the scale of participant (P710) to all relation (Q930933). It is also proposed to change the constraints of diplomatic mission sent (P531). If interested in this field, it is discussed here. Louperivois (talk) 16:48, 9 June 2015 (UTC)

Catholic Church (Q9592): organization or religion?

Catholic Church (Q9592)

@GerardM, Jane023, Ayack, Filceolaire, Giftzwerg 88, Olivier LPB: Sorry to start again an old discussion but we should be coherent especially when we protect again any edition like this one using a bot. How can Catholic Church (Q9592) be a religion if this item is defined as a subclass of Christian Church which is defined as a religious organization ? And why can't we use Catholicism (Q1841) as religion and at the same define this last item as a suclass of Christianity which is defined as a subclass of Abrahamic religion ? A person can't have Catholic Church (Q9592) as religion but can be member of Catholic Church (Q9592). And following the Greek definition Catholicism (Q1841) means "universal doctrine" which is more adapted to the concept of religion. So please stop the bot protection. Snipre (talk) 18:26, 8 June 2015 (UTC)

  • I suppose the question is mainly of interested in regards to Property:P140 used on items for persons and labelled "religion", but that property seems to be primarily for church affiliation. --- Jura 18:37, 8 June 2015 (UTC)
    @Jura So if your supposition was correct, we shouldn't have a constraint for Property:P140 limiting the value for P140 to items labeled with instance of: religion. A church is not a religion, for example the Catholicism as religion is composed of 23 churches. Then from classification point of view we have to distinguish between a church as organization and a religion as theory/doctrine/philosophy. We can't have mixed concepts in one item. Snipre (talk) 07:56, 9 June 2015 (UTC)
  • Snipre: I totally agree with you and I don't understand these edits. — Ayack (talk) 20:03, 8 June 2015 (UTC)
  • I am agree to Snipre (talkcontribslogs) Olivier LPB (talk) 07:37, 9 June 2015 (UTC)
    People are a member of a church not of a religion. Also doctrinal differences between your churches is such that each of them do not want to be mistaken for the other. Your insistence on this fallacy does exactly that. GerardM (talk) 09:01, 9 June 2015 (UTC)
    What is a religion and what is a church ? The problem is the definition. From a very simple approach a religion is a religious doctrine/belief and a church is an organization. In the normal language we usually mix these two concepts but this isn't correct. Typical examples: Catholic Church (Q9592) is composed of 23 different churches (one western church and 22 eastern churches and all recognized the pope as chief but have the official title of church) having the same doctrine/religion. Same for some protestant churches which are independent from organization point of view (name, distinct buildings or independent hierarchy) but share common belief/doctrine. Here we see we have to split between church as organization and religion as doctrine in order to be able to link correctly items. A religion defined as religious doctrine can be shared by different churches, so we have to split the concepts. Snipre (talk) 09:31, 9 June 2015 (UTC)
    By the way, GerardM, please comment this fabulous proof of absurdity: the religion of Catholic Church (Q9592) is ... Catholic Church (Q9592). Snipre (talk) 10:01, 9 June 2015 (UTC)
    The Catholic Church is both a religion and an organisation/church. I don't know if this extends to other religions/churches but it is perfectly legitimate in this case. It has a specific, unique set of doctrines (religion) and it is a hierarchical organisation (church); one feeds into the other and both intersect in one entity. From a practical point of view, I don't care what data item it is; I just want to have one — currently Roman Catholicism (the largest sect of the largest faith on Earth) does not exist on Wikidata. Catholic Church (Q9592), the data item linking to all the Wikipedia articles for Roman Catholicism, has recently been purged of all references to Roman Catholicism with no replacement. - AdamBMorgan (talk) 10:40, 9 June 2015 (UTC)
    @AdamBMorgan: In that case you have to create two items because WD structure doesn't allow mixing of concepts. I don't want to enter in the definition of what is the catholic church, I just want to solve the problem of mixing two concepts, an organization and a religious doctrine, in one item. This completely confuses our contraints system for properties (please have a lot to the constraint reports of P140). Then for maintenance point of view, an doctrine has a proper set of properties and an organization another one. So we have to split concepts in order to facilitate connections work.
    By the way saying that catholic church is a doctrine this is an language abuse. Even the bishops and cardinals speak about "doctrine of catholic church" and never literally "catholic church" to say doctrine. Just have a look at the Vatican web site to see that we call the doctrine of the catholic church "doctrine" and not Ctaholic church: the congregation for the doctrine of faith and not the congregation for the catholic church of faith.
    Currently Catholic Church (Q9592) is defined as an organization (please look at the value of property instance of (P31) if this is not clear) so if we don't agree about Catholicism (Q1841) as doctrine of Catholic Church (Q9592) we have to create an new item. Snipre (talk) 11:32, 9 June 2015 (UTC)
    @Snipre: I disagree about the "mixing of concepts"; I would call the Catholic Church an intersection of both religion and organisation. I think it could be adequately covered with these statements:
    Catholic Church (Q9592) instance of (P31) Christian Church (Q34651)
    Catholic Church (Q9592) instance of (P31) religion (Q9174)
    Catholic Church (Q9592) subclass of (P279) Catholicism (Q1841)
    I don't think religion (P140) makes much sense for this item, the religion is not separate to the organisation, but religion (P140) Catholicism (Q1841) would be OK.
    However, if you want to create a new item for "Roman Catholicism" and state that as the religion of the Roman Catholic Church (and by extension all people that follow that religion) that should be OK too. I just want a data item to use. (Actually, it might be easier to create a new item for the church and re-purpose this item as the religion, if going down that road.) - AdamBMorgan (talk) 13:05, 9 June 2015 (UTC)
    Your proposition is not based on data structure reasons but on personal view. If you recognized that two concepts exist you have to split them in the database in order to provide adequate descriptions and avoid inconsistency like the one I pointed (religion of Catholic Church (Q9592) (as organization) is ... Catholic Church (Q9592) (as religious doctrine)). So even your proposition doesn't avoid this "illegal" relation in term of data structure.
    Then we agree about the fact that the slitting is a reasonable solution. Snipre (talk) 14:31, 9 June 2015 (UTC)
    @GerardM: where have you seen that the property religion refer to the church and not the religion : the property is religion so we put a religion and not a church. Olivier LPB (talk) 13:47, 9 June 2015 (UTC)
  • For political subject we separate ideology and organization. Just have a look at French Communist Party (Q192821) to see that we don't mix the parti with its ideology describes by communism (Q6186). We have to use the same structure for religion and using religion (P140) to link organization with doctrine like we do for political partis using political ideology (P1142). Snipre (talk) 11:39, 9 June 2015 (UTC)
    One problem is that there is a continuum of both religion and political ideologies. The 30-year religon war in Europe was mainly between catholic France and catholic Austria, so if you want to involve doctrine, you have to split also the catholic church since they not always have agreed. There have been communist-parties in almost every country in the world. But there is more than one kind of communism. The politics in Sweden have a history of Lenin-communism, Stalin-communism, Mao-communism, nationalistic communism, elitistic communism, Tito-communism. Communism is not what puts them together, it's what tears them apart. It's the same thing with religion. -- Innocent bystander (talk) 11:51, 9 June 2015 (UTC)
    We can't avoid this step of splitting until we arrive at the level one item = one concept. We have Maoism (Q167651), Stalinism (Q179121), Leninism (Q155059),... this shows that the splitting is not an exception but something which is well established in WD. If we have to do the same for religions, I think this can be only an improvement in order to use the correct terms instead of using confusing concepts. Even in religion we have already a lot of accurate terms to indicate small differences in doctrine: arianism (Q83922), homoiousia (Q11863577), Q3139908, Anomoeanism (Q1194032), Gallicanism (Q740931) ... So this is not a new thing. Snipre (talk) 12:43, 9 June 2015 (UTC)
    @GerardM, Jane023, Ayack, Snipre, Giftzwerg 88, Olivier LPB: Who say that religion (P140) refers to a philosophy rather than an organisation? The constraints are there to reflect the usage of the properties, not to dictate them. As the requirement is that 'religion' is claimed by the subject or documented we are pretty well bound to use it to link to organisations rather than philosophies since membership of organisations is in general better documented. If a person claims an organisation as their 'religion' then we should respect that claim. The constraints for religion (P140) need to be changed so it can link to religious organisations as well as to religious philosophies. Filceolaire (talk) 14:15, 10 June 2015 (UTC)
    @Filceolaire
    If a person claims an organisation as their 'religion' then we should respect that claim
    No ! It's to us to decide if the property intended usage map the usage in common language. Common language is full of ambiguities, that's why we create several items. We should do the same with properties. In religion, that may be complex because someone can say he is a catholic, for example, but is not baptized (yet) or the Catholic Church don't want to baptize him for some weird reason (this happens), or was baptized, does not believe anymore but did not make the effort to officially said Roma he does not want to be a catholic anymore ? If there is two usages, we can use two properties as well: says he agree or believes the dogma, is a member of the organization, and so on. TomT0m (talk) 15:44, 10 June 2015 (UTC)
    @Filceolaire: Please look at the constraint of religion (P140) to see that value of P140 should be labeled "instance of: religion" so next time you want to fix a value, please do the job until the end: delete the constraint or modify Catholic Church (Q9592) to respect the constraint. Then according to your reflexion we don't need the property P140 because membership is defined with member of (P463). And finally please explain how you can accept that for religion (P140) of Catholic Church (Q9592) we have as value Catholic Church (Q9592) ? So even if your proposition makes sense, in fact this leads to a non sense in term of data structure. And in case of Buddhism (Q748) or Confucianism (Q9581) as religion of a person, can you give us the organizations which are in charge of those religions ? Snipre (talk) 16:26, 10 June 2015 (UTC)

I agree with AdamBMorgan that religion=church in this sense. A church (architectural structure) is a specific place where the organized church meets and this church (organization) is an instance of a religion. We don't need to separate the idea of religion and church (organization) because these are the same. To be clear, the United States deliberately decided to separate church and state, because the founding fathers realized (quite correctly) that the state cannot be democratic otherwise. Jane023 (talk) 18:54, 10 June 2015 (UTC)

Instances of historical country (Q3024240)

There are some 2000 items with instance of (P31) historical country (Q3024240). (A large percentage of these are inaccurate, but that's besides the point.) Being a historical country (Q3024240) would, I think, be implied by having a dissolved, abolished or demolished date (P576) statement or end time (P582) qualifier. Furthermore, having instance of (P31) historical country (Q3024240) statements without qualifiers indicating time is simply confusing, as when they were countries, they were not also historical country (Q3024240)s. Would there be any objections to removing all of these? --Yair rand (talk) 00:54, 9 June 2015 (UTC)

If it's inaccurate: yes, please remove it. If has to be derived in a complex way by an unknown algorithm from non-existent qualifiers, no. --- Jura 07:41, 9 June 2015 (UTC)
@Yair rand: I see no problem into let them stay as historical state. Knowing they are historical state with no end date can point us to a potential problem, or to missing information that has to be filled up just as you found them. Actually as beeing in a class implies some properties or values, this is true for a lot of other classes. TomT0m (talk) 08:58, 9 June 2015 (UTC)
I guess those statements can stay, but a related question is if such things as Swedish Empire (Q215443) should have any statement like instance of (P31): country at all. From my point of view, the article is about the history of Sweden, not a country in itself. Union between Sweden and Norway (Q62589) is more difficult to define, if it was a country or just two nations sharing a king, like UK and Belize does today. Sweden-Finland (Q3279296) was definitly not a country, it is today a name for a period when the Eastern parts of Sweden was located east of the Baltic sea. -- Innocent bystander (talk) 09:12, 9 June 2015 (UTC)
Please do not remove instance of (P31) historical country (Q3024240) when it is missing a start/enddate as only reason. Many items, such as buildings, have a 'is a' claim and missing a date qualifier. (They are not wrong but missing information in my opinion). Michiel1972 (talk) 12:19, 9 June 2015 (UTC)
Please improve, don't remove. So yes, I object to removing all of these. Multichill (talk) 18:50, 9 June 2015 (UTC)
Okay, so suppose the item does have the appropriate end/dissolution dates. Then can the historical country (Q3024240) statements be removed? It really seems like redundant data; we're not adding "dead human" to every page on a dead person. --Yair rand (talk) 23:27, 9 June 2015 (UTC)
@Yair rand: : suppose a vandal remove an end date statement and this remains unnoticed for some reason. Then the item could reappear into the to check items list. I don't really see how this kind of redundancy could be bad. TomT0m (talk) 10:09, 10 June 2015 (UTC)
(What is the "to check items list"?)
The reason this is bad is that by having two independent ways of saying something, we're unnecessarily creating a fork of the data, which will not be synchronized to each other. If a country becomes a former country, and this data is added by only one method, then any user querying by the other method will not have accurate/complete data. Possible vandalism removing a statement is not a good reason for adding redundant forms of saying the same thing. --Yair rand (talk) 10:20, 10 June 2015 (UTC)
@Yair rand: It's a mechanism to keep things synchronized (query, constraint report, ...) . There is no forks if we know that the instances of a class should have some informations, it's just basic consistency check. The kind Krbot already does. There is also no fork if we use subclass and learn to use Wikidata Query. Class is also a way to compress things : We at first sight can see if the country is a former one just by reading its class. TomT0m (talk) 10:54, 10 June 2015 (UTC)

How do you add references to pages

I tried to add references to the batman page I created, but they did not appear. What do I do? Super duck (talk) 18:00, 9 June 2015 (UTC)

Why did you make a new page for Batman? It already existed. I merged it with Batman (Q2695156). Mbch331 (talk) 07:43, 10 June 2015 (UTC)
Mbch331Sorry. Super duck (talk) 18:01, 10 June 2015 (UTC)

Filling in edit summary section headings with User:Yair rand/EditTranslatedSections.js

When a header uses the {{TranslateThis}} template, the edit summaries resulting from edits to that section either link to the anchor containing the full wikitext of the template use (if it's all on one line) or don't add a section link at all (if it spans multiple lines). These don't really allow easy following of discussion. I wrote a script, User:Yair rand/EditTranslatedSections.js, to fix this issue, and I propose that it be turned on by default. --Yair rand (talk) 05:40, 10 June 2015 (UTC)

Thanks! Can you extend it to also support proper linking to {{Q}} inclusions in headings? -- Bene* talk 11:37, 10 June 2015 (UTC)
Probably not, unfortunately. For those, the anchor would always vary by language, unlike {{TranslateThis}}, which has an anchor param to set which language is the canonical one for anchoring purposes. (When anchor is not set, the script will default to user language, which will make it similarly difficult to track discussions on occasion.) Maybe there could be something similar created for {{Q}}, though... --Yair rand (talk) 12:04, 10 June 2015 (UTC)

Wikimania talk input

Hey folks :)

I'll be giving a talk about the state of Wikidata at Wikimania in Mexico City. Is there anything you think I should definitely talk about? Some great story? A useful project? Some nice numbers? --Lydia Pintscher (WMDE) (talk) 07:03, 10 June 2015 (UTC)

  • Perhaps a word about the future implementation of the numeric datatype with unit and the current implementation of the arbitrary access: we have often discussion in WP talkpages about WD and we have to mention that WD can't solve the problems of the contributors because of the lack of these two features.
Then we need more people to take part to our projects: in order to progress we need to create properties, to define how items should be classified or which items should be create and with which definition,... and too few persons are available to perform this kind of organization tasks, or simply to discuss and to propose solutions. Snipre (talk) 07:53, 10 June 2015 (UTC)
  • Mention how internationalisation is baked into wikidata just waiting for localisation to be done and how even a little localisation goes a long way. Filceolaire (talk) 09:54, 10 June 2015 (UTC)
Well, maybe not completely baked in... --Yair rand (talk) 10:23, 10 June 2015 (UTC)
  • Where we stand .. well I guess we should try to do better in assessment than .. (let's stop discussing that).
    I'd attempt to detail (A) what currently can be done, how well that works; (B) What can be done in the next 6 months (to be sure that at Wikimania a year later it has actually been done), and then (C) what are the options (and resources) beyond that.
    For all points, personally, I'd attempt to do a status content wise, but your focus might be technology. Wikipedia is probably more interested in content.
    Topics to be covered (D) could be sitelinks and what they can tell us about Wikipedia (D1), authority control for people (D2), infoboxes (D3), images (D4), coordinates (D5), how to build another cebwiki/itwiki BioStubs/gazetteer .. not that I think it's a good idea (D6). Main groups for items (E) are probably people (E1), taxons (E2), geographical areas (E3), architectural structures (E4), works (E5), categories, disambiguation and other MediaWiki interna (E6), WS/WN (E7), the unknown (E8), one could probably lists series of sets of thousands items that are being developed and can provide a good structure for the field (E9). Some might be interested in a short general comparison to Freebase and Dbpedia, maybe not the knowledge graph (F). List a few uses external parties already make (G). Given the other assessment, maybe a clarification about descriptions and labels is needed - one doesn't need most of them for D6 (H). --- Jura 11:07, 10 June 2015 (UTC)

--Atlasowa (talk) 12:34, 10 June 2015 (UTC)



The risk is if you do the stats on an abstract level, that they may not necessarily mean much to Wikipedia contributors and we might end up with research projects about descriptions in unused languages. What can it do for biographies (E1), what can it do for film articles (E5)? What can it do for infoboxes (D3). I think User:Zolo/Wikipedia content is worth developing, not necessarily just for this presentation. (BTW, you might want to display the slides on a separate page). --- Jura 12:45, 10 June 2015 (UTC)

cawiki issue with Module:Wikidata

Would someone look into the issue cawiki has at ca:Plantilla Discussió:Categoria per any de defunció or ca:Tema:Siwj5t9zxhszv4qg. Their Wikidata LUA modules outputs incorrect dates. --- Jura 11:51, 10 June 2015 (UTC)

  • @Jura1: they works with time as with strings, without parsing them according to ISO 8601 (Q50101). The core problem, i assume, in the following lines:
84 if #d>26 then
90 if #d>26 then
  • due to recent change in Wikidata data time string does not have leading zeroes anymore, thus size of strings is never 26 anymore. They should migrate to appropriate ISO 8601 (Q50101) parsing. And yes, welcome to changing API hell. -- Vlsergey (talk) 15:40, 10 June 2015 (UTC)

Unnoticed vandalisms :

Today :

The second one should have been flagged easily. What the status of our filters ? TomT0m (talk) 15:57, 10 June 2015 (UTC)

First edit is not possible to track with a abuse filter. For the second one, you could ask for a blacklist with some words. Sjoerd de Bruin (talk) 16:10, 10 June 2015 (UTC)
To the devteam, an admin, ? TomT0m (talk) 16:15, 10 June 2015 (UTC)
On Wikidata talk:Abuse filter, I think. Sjoerd de Bruin (talk) 16:19, 10 June 2015 (UTC)
  • I don't get the title of this thread. It was noticed and reverted. Congrats to TomTom! Keep up the good work! --- Jura 16:17, 10 June 2015 (UTC)
    • @Jura1: If I replace the text of an article with haha or caca in frwiki, it gets reverted by a bot in a heartbeat. Here I reverted it in time that has nothing to do, just because the items were in my watchlist ... TomT0m (talk) 16:20, 10 June 2015 (UTC)
      Same is in enwiki.--Ymblanter (talk) 16:30, 10 June 2015 (UTC)
It's a bit more difficult to do that here since there actually can be items that legitimately have a label "haha", whereas no Wikipedia article would have that as its contents. As a result, we need to do a lot more manual patrolling. This is made substantially more difficult by the fact that we don't have any good way of filtering the recent changes to changes we actually understand... --Yair rand (talk) 16:39, 10 June 2015 (UTC)
I blocked the IP from the first edit for a year since all of their edits are vandalism, and they have been previously blocked for a month.--Ymblanter (talk) 16:29, 10 June 2015 (UTC)
I made Special:AbuseFilter/history/11/diff/prev/756 to prevent what's shown under the second link. I won't open the first one since I may not. Matěj Suchánek (talk) 16:47, 10 June 2015 (UTC)
@Matěj Suchánek: It's not that bad, you can click, just turn the sound down and prepare to close the window :) TomT0m (talk) 16:51, 10 June 2015 (UTC)
I thought I shouldn't open the revision difference (stupid me)... I think there are URL blacklists directly in some extensions, someone could test if they work with Wikibase, and populate them. If they didn't work, we could use AbuseFilter which also can (and it does in Wikidata) block spammers. Matěj Suchánek (talk) 16:58, 10 June 2015 (UTC)
Little more tricky to do good filters here, since kaka is something I ate together with a cup of coffee earlier today. And when I told my last name to people I met in Russia, they started to laugh, I still do not know why. Maybe I have the same problem, as a Yugoslavian friend of mine. His last name meant "pee" in Swedish, so he changed it. -- Innocent bystander (talk) 16:59, 10 June 2015 (UTC)

Query Tools?

Are there any (online) tools available to query wikidata? For example, I want to get "all wiki software written in Python". How to do that? --Techtonik (talk) 08:36, 4 June 2015 (UTC)

Autolist works pretty well. The query you're looking for is "claim[31:6686945] AND claim[277:28865]", which currently returns no results. --Yair rand (talk) 09:06, 4 June 2015 (UTC)
Thanks. I also found https://wdq.wmflabs.org/wdq/ which is much easier to use and edit. I see there is a problem with linking wiki installations and wiki software. I'll ask next question about it right away to keep it search friendly. --Techtonik (talk) 10:48, 4 June 2015 (UTC)
@Techtonik: You may also wanna try the beta SPARQL endpoint which will once become the official query endpoint for Wikidata. -- Bene* talk 18:39, 5 June 2015 (UTC)
Interesting stuff. Looks like SQL. It would be cool if the current wdq interface which maps query to graphical outline, also mapped to this SPARQL. The prefixes are surely confusing. For the starter, I'd prefer an SQL like query interface that operates solely on wikidata. ---Techtonik (talk) 14:54, 6 June 2015 (UTC)
@Techtonik: You can also use http://tools.wmflabs.org/wdq2sparql/w2s.php to translate WDQ queries to SPARQL. Be careful though as big queries like TREE/WEB can result in timeouts for the endpoint, it's now restricted to 30 seconds per query. --Laboramus (talk) 21:23, 7 June 2015 (UTC)
@Techtonik: see also {{WDQ}} and {{Query}}. TomT0m (talk) 10:15, 7 June 2015 (UTC)
@Bene*: What do you mean by "official query endpoint"? Is SPARQL going to become the standard method of writing queries? --Yair rand (talk) 09:52, 11 June 2015 (UTC)
The service you query against will be a SPARQL endpoint. But that doesn't necessarily mean you will have to write your queries in SPARQL. Bene and Tpt have been working on a tool to convert natural language questions to SPARQL for example at the last hackathon. There is also a tool to convert the query language used by Wikidata Query to SPARQL. For good integration of all this in Wikidata itself and on the sister projects we'll still have to see what we'll end up using. --Lydia Pintscher (WMDE) (talk) 12:41, 11 June 2015 (UTC)

Coming to a decision for usage instructions in descriptions

Hey folks :)

In the Wikipedia app on iOS and Android you see the Wikidata description below the article title. The developers at WMF want to roll this out further now on the mobile website. One of the blockers for that is usage instructions that are sometimes added to the description of an item here. They're imho fine for Wikidata but are pretty bad when shown as the description of the topic of an article on Wikipedia. There were already several discussions here, in phabricator:T97566 and on some item talk pages. I'd like to get us to a conclusion now so we have this solved instead of kinda discussing it over and over ;-)

The options I see for us:

  1. Ban usage instructions in descriptions. I don't think that's acceptable.
  2. Find a special syntax to allow consumers like the Wikipedia app to strip out the usage instructions. We could for example tell them it is ok to strip out everything in brackets.
  3. Add a special field for them. This would be the cleanest solution but I don't think the additional development time it'd need from the development team and the additional maintenance effort for the editors is justified at this point. We might want to consider this later however.

Opinions? Comments? --Lydia Pintscher (WMDE) (talk) 16:27, 4 June 2015 (UTC)

Is it not an option to have a monolingual text property for it that then gets shown in the search results? That would seem like the most obvious solution to me. I don't think a dedicated field would be the best solution because it's not something that most items are going to need. The idea of coming up with a special syntax for it so that people can extract it out of a chunk of text seems bizarre to me... isn't the whole point of Wikidata to have structured data so that people don't have to try and extract the things they want out of chunks of text? Maybe as a short term option, but ultimately it seems to me that descriptions and usage instructions are two separate things that don't belong in the same field. - Nikki (talk) 16:55, 4 June 2015 (UTC)
Yeah we could go with a property for it as well. And you're right about it not being good to have both in the same field. However we need a short-term solution. --Lydia Pintscher (WMDE) (talk) 17:05, 4 June 2015 (UTC)
Dedicated text property would be useful as qualifier in image (P18) to describe its contents sometimes. Otherwise the simple syntax that trim usage hint from description could be a first instance of comment marker like "//" (from C, java, etc.) or "--" (from Lua) in the text. Paweł Ziemian (talk) 20:33, 4 June 2015 (UTC)
I like option 2, but not brackets specifically, which are occasionally needed for actual descriptions. We should probably also use this to separate property descriptions from usage instructions. Is there any chance we could have the usage instructions displayed differently in the suggestions box? --Yair rand (talk) 21:13, 4 June 2015 (UTC)
Would Description <!-- Internal help --> break something? --Succu (talk) 21:21, 4 June 2015 (UTC)
I really like Nikki's proposal of a dedicated property. However, it should be of multilingual text which is not available yet. The property should be called "usage instructions" and can be used on items and properties. Relevant popups in the UI should then be configured to show the instructions in the user's current interface language. However, this requires the multilingual text datatype. -- Bene* talk 21:24, 4 June 2015 (UTC)
Bene, I take your point, but think maybe getting it done quickly, even with a monolingual text property, would be a good start. StevenJ81 (talk) 22:38, 4 June 2015 (UTC)
I think a dedicated text property would be a good solution. If we include hidden usage instructions in the description (option #2), the semantics are still polluted and there's a good chance the instructions will be accidentally deleted when someone edits the description through a 3rd party interface (as already exists on the iOS app). Wikibase could then have a configuration variable that let's you add a specific property to the auto-suggest hints and this property could be listed there. Even if it is only mono-lingual to start with, it would be a usable solution for the short-term. Kaldari (talk) 22:42, 4 June 2015 (UTC)
Migrating from monolingual to multilingual text is indeed very easy and can be done later on. However, using the monolingual text, all values in all languages will be shown in the interface and it will get cluttered too much. This is the advantage of the multilingual text; it will work like the labels/descriptions and is able to only show the value in the user's language and some fallbacks. -- Bene* talk 06:28, 5 June 2015 (UTC)
You ask for comments.. We do know that the manual descriptions are often flawed. They are much worse than the automated descriptions. Why are they not considered. They do not have the issues you mention. GerardM (talk) 05:45, 5 June 2015 (UTC)
Lydia you omitted two other options which I proposed on fabricator
4. mobile wikipedia ignores the wikidata descriptions and generates the descriptions automatically from the statements
5. Pids and Qids in descriptions are automatically converted into text links. This would allow all 'usage notes' to be rewritten into a form which is useful for others and would make it much easier for users to find the right item or property or mobile WP article
Option 4. seems like a very quick win since much of the work on auto descriptions has been done here already. Option 5. can't be that hard surely. Parsing text into html is something wikis have been doing for years. Mobile WP could even just implement option 5. in their article descriptions and leave the wikidata descriptions as they are.
Could you give us a comment on why these options are not practical? Filceolaire (talk) 06:16, 5 June 2015 (UTC)
Automatic descriptions are not (yet?) good enough for Wikipedians to accept them. For linking: How would you rewrite Q503 for example to make it suitable for the usecases at hand? --Lydia Pintscher (WMDE) (talk) 10:30, 5 June 2015 (UTC)
As far as I am concerned, I would be OK with 2, 4, and 5, but I agree that generating automatic descriptions is unlikely to work in so many cases that it is not a viable solution (not across the board, perhaps in some areas?). - Brya (talk) 10:57, 5 June 2015 (UTC)
Prefer #5 > #3 > #2. Not #4, #1 --- Jura 11:10, 5 June 2015 (UTC)
Got the numbers mixed up. It's prefer #4 (autodesc) > #3 (separate field) > #2 (format). Not #5 (wikilinks in desc), #1 (drop without replacement). --- Jura 22:03, 6 June 2015 (UTC)
For every example you post there are more that prove the opposite. Banana currently is fruit.. It is as descriptive as the one in English and less confusing. Your argument fails badly. Sorry. GerardM (talk) 13:23, 5 June 2015 (UTC)
I don't like the idea of putting usage instructions in a new field: they should be where they are noticed. I would prefer the reverse, having a definition in a new field. It may be possible to populate this, for starters, with automatically-created definitions, and take it from there, correcting as needed. - Brya (talk) 18:21, 5 June 2015 (UTC)
As a quick solution, a marker like // between description and usage hint is okay for me. --Pasleim (talk) 21:19, 6 June 2015 (UTC)
Using -- like Paweł Ziemian suggested above might be preferable. It's hard to tell whether the average editor is more likely to be familiar with any of the languages that use // or with Lua. A more important consideration would be if either would be used in any descriptions for other uses. I can't think of any, but maybe some entity makes use of sequential slashes or dashes? The Wikipedia search function seems unable to check for these adequately. --Yair rand (talk) 20:32, 7 June 2015 (UTC)
[6] and [7] list all descriptions with // and --, respectively. --Pasleim (talk) 22:10, 7 June 2015 (UTC)
Thank you. So, that's 359 for //, mostly URLs, with some hundreds of items with descriptions "Listed company on the Polish Stock Exchange - details at http://www.gpw.pl/karta_spolki_en..." generated by User:Reinheitsgebot, and 51 for --, mostly either odd mistakes, attempts at making pseudo-redirects, and odd HTML-style comments. Hmm... --Yair rand (talk) 23:09, 7 June 2015 (UTC)
I would prefer // (with a space before and after, both for readability and to avoid matching URLs) over -- because -- seems like something more likely to be deliberately used in natural language, e.g. as a replacement for an en-dash or em-dash or in Min-nan to represent a syllable with a neutral tone (en:Pe̍h-ōe-jī#Hyphens). - Nikki (talk) 23:45, 7 June 2015 (UTC)
Unfortunately, double slash is also used on occasion in natural language. --Yair rand (talk) 02:06, 8 June 2015 (UTC)
If we additionally put a space between it as suggested by Nikki that should be ok though? I don't think we will find something that is never used in any language. But we can find something that is rare enough :) --Lydia Pintscher (WMDE) (talk) 16:45, 8 June 2015 (UTC)
It seems like // (with a space) should be usable. There are currently only 3 instances of that being used on Wikidata (all in Russian). In all of those cases, it seems to be used as a type of punctuation in a book citation, which doesn't really belong in a description anyway. Automated descriptions have a few issues: (1) It isn't easy to provide an interface to edit these outside of Wikidata. (2) Until Wikidata is more fleshed out, this will give rather haphazard results. (3) Automated descriptions are not always going to help with disambiguation. For example, there are two different French films named "Love". Kaldari (talk) 17:58, 8 June 2015 (UTC)
I like Nikki idea with two spaces, one before and one after ( // ). Paweł Ziemian (talk) 19:22, 8 June 2015 (UTC)
+1. --Yair rand (talk) 23:34, 9 June 2015 (UTC)
I like a new field for descriptions, auto filled based on statements but with the provision for the auto fill algorithm to improve over time and provision for editors to improve or replace auto descriptions. Filceolaire (talk) 09:42, 10 June 2015 (UTC)
In answer to Lydia's question about option 5. and banana I would rewrite the description as
"the fruit (for the best-known species, see Musa × paradisiaca (Q10757112); for the genus, see Musa (Q8666090))"
ok? Filceolaire (talk) 12:14, 10 June 2015 (UTC)
What do we do if there is no article on the Wikipedia for the linked items? --Lydia Pintscher (WMDE) (talk) 12:43, 11 June 2015 (UTC)
the same thing you do when there is no description in a language (I hope) add automated descriptions they are typically more informative. GerardM (talk) 13:27, 11 June 2015 (UTC)
Lydia Pintscher (WMDE) This will happen a lot. In many cases the commentary in the wikidata description will serve to distinguish between multiple wikidata items which are all described on the same wikipedia article. This is not a bug. It is after all what the Wikidata descriptions are for. This is another reason why WMF should not use these descriptions but should auto-generate their own descriptions instead, as my option 4 above. Filceolaire (talk) 11:09, 15 June 2015 (UTC)

Collapsing data from what are conceptually two separate things (description, and usage instructions) into a single field using a delimiter doesn't make sense to me at all; you're taking structured data and making it unstructured. Just like you wouldn't do that in a database, it shouldn't be done here. Third parties who reuse Wikidata will all then have to implement logic to parse the description and remove the superfluous information, so we'd be making it harder for third parties from using Wikidata, which I see as the entire point of Wikidata in the first place. I'd really prefer Wikidata actually stay true to its commitment to be a structured data repository, and use separate fields for separate concepts. --Deskana (WMF) (talk) 01:40, 12 June 2015 (UTC)

@Deskana (WMF): The main purpose of the description-field, as far as I have understood, is to help us add statements here at this site here at Wikidata. In some cases we obviously have needed both a "description" and a "usage instruction" to fullfill that purpose. That was maybe not possible to see when Wikidata was originally designed, that was a need we discovered on the road. Now the WMF-developers want to use these description-fields for something completly different, for the mobile interface in the clients. I cannot see how that become our problem, and I am not so sure that is a very good idea at all. -- Innocent bystander (talk) 04:45, 12 June 2015 (UTC)
The problem is that people already include usage information in descriptions, because it's information that people need and and we don't have anywhere else it can go. I completely agree that they should be separate (see my original comments) but something is needed until they can be split properly, because descriptions and usage information are currently combined in the description field in a completely unstructured way which is causing some issues for some people trying to use the data, and standardising how they're combined is at least an improvement for now even if it's not ideal. - Nikki (talk) 06:41, 12 June 2015 (UTC)
 #4 seems much the better option for re-users. descriptions are not meant to be a substitute or a replacement or place of storage for the information in statements. --- Jura 09:08, 12 June 2015 (UTC)

So we're at an impasse here. I think it's being made much more than it actually is. The next things that we need is data about how many descriptions does this actually affect. I'll try to get that. --Lydia Pintscher (WMDE) (talk) 12:36, 22 June 2015 (UTC)

Telescope properties

Hello. I'm currently having a go at updating en:Template:Infobox telescope so that it can use wikidata properties directly, using en:Module:Wikidata. The main issue seems to be finding relevant properties here to link to - it seems to be a bit tricky to find existing properties, and it's not clear if all of the needed properties exist yet. There's also some more complicated issues: in particular, some properties might need to have multiple entries, and also it's not obvious how to include wikidata coordinates on enwp (simply passing them into en:Template:Coord doesn't seem to work). I was wondering if anyone would be able to help me figure out what to do here? I've put my current understanding of the parameter-property connections at User:Mike Peel/sandbox - feel free to edit it! Thanks. Mike Peel (talk) 19:03, 6 June 2015 (UTC)

Wikidata does not yet support quantities with units. That means, we don't have properties for altitude, diameter, area etc. --Pasleim (talk) 21:06, 6 June 2015 (UTC)
Thanks @Pasleim:, that's unfortunate, is there an ETA for that support? Also, is it possible to give a range of values using wikidata (e.g., range in wavelength or frequency)? Thanks. Mike Peel (talk) 21:30, 6 June 2015 (UTC)
@Mike Peel: What is ETA ? Snipre (talk) 16:51, 7 June 2015 (UTC)
Estimated Time of Arrival? -- Innocent bystander (talk) 16:53, 7 June 2015 (UTC)
No date defined. The dev team is working on that feature. Better follow this to have a better information about the progress. If I have to give an estimation I would say September/October. Snipre (talk) 21:06, 7 June 2015 (UTC)
Mike Peel, Quantities (and dates) can have an uncertainty but to define a range you need two properties and two statements as start time (P580) and end time (P582) for example.
To search for properties put "P:" in front of your search term (yes this should be better documented)
Properties using the "quantity with units" datatype which have been approved but are on hold waiting for this datatype to be available are listed at Wikidata:Property proposal/Pending/2. If the property you need isn't there then you can propose others to be added to this 'pending' list.
Properties with multiple entries (or 'values') is not a problem. There is even a mechanism where one of the values can be marked as the preferred value. This is done to identify the current value where we also list historical values - for population for instance.
Multiple values can also be distinguished with qualifier properties like applies to part (P518) to distinguish diameter, diameter2, diameter3.
For "built" and "first light" dates you can use the significant event (P793) property.
For "style" you can use instance of (P31), creating new items for each style as required or you can create a new property linking to these items. I would be against using free text as it makes it more difficult to localise the style info into other languages.
Can't help you with Template:Coord. Sorry.
OK? Filceolaire (talk) 12:44, 10 June 2015 (UTC)
@Filceolaire: thanks! That's really useful information, although I'm not sure how well that will integrated with Module:Wikidata, particularly where there might be multiple entries for different things using the same property number (ping @RexxS: about this). The Template:Coord issue is now sorted, and wikidata coordinates are now being automatically included in the wikipedia articles through the template nicely. :-) Thanks. Mike Peel (talk) 18:49, 12 June 2015 (UTC)
The 'Preferred' ranking is very obtuse! After a bit of googling, I found Help:Ranking#Preferred_rank, which explains it, but I have no clue how I'd find that while editing a wikidata entry! Thanks. Mike Peel (talk) 20:17, 17 June 2015 (UTC)
@Mike Peel: Help:Ranking#How to apply ranks. Matěj Suchánek (talk) 10:28, 18 June 2015 (UTC)
@Matěj Suchánek: thanks, figured it out now, but I would never have noticed it on a wikidata entry page if it hadn't been mentioned here! Thanks. Mike Peel (talk) 20:24, 18 June 2015 (UTC)
@Mike Peel: Multiple values should be passed to something like en:Template:Plainlist or en:Template:Flatlist. See also my comments at en:Template talk:Infobox telescope. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:12, 25 June 2015 (UTC)

next round of arbitrary access rollouts

Hey :)

Rollout of arbitrary access is going well. So we're continuing on more projects. Here's the schedule for the next projects to get it:

  • 11 June 2015: all Wikiquote, all remaining Wikivoyage
  • 15 June 2015: arwiki, cawiki, eswiki, huwiki, kowiki, rowiki, ukwiki, viwiki


Cheers --Lydia Pintscher (WMDE) (talk) 14:08, 7 June 2015 (UTC)

@Lydia Pintscher (WMDE): would you mind to publish whole schedule for optimistic scenario, or at least some "all projects by the end of June"? -- VlSergey (gab) 19:39, 7 June 2015 (UTC)
We'll be doing those I just announced and then we'll have to take a break because of travel and Wikimania. We'll continue with the rest at the end of July. By the middle of August I hope we're done with the Wikipedias. Then the remaining sister projects. --Lydia Pintscher (WMDE) (talk) 19:50, 7 June 2015 (UTC)
@Lydia Pintscher (WMDE): it is very sad to read, but let's hope for the August at least. -- VlSergey (gab) 20:18, 7 June 2015 (UTC)
Is there a project you're particularly interested in (except Commons)? Maybe we can move it forward a bit. --Lydia Pintscher (WMDE) (talk) 11:53, 8 June 2015 (UTC)
@Lydia Pintscher (WMDE): i'm very interesting in ruwiki and a bit less in ruwikinews. We have some plans for ruwikisource-ruwiki integration (dictionary links), but they require arbitrary access to prevent information duplication at Wikidata. -- VlSergey (gab) 17:05, 8 June 2015 (UTC)
OK. Let me talk to Katie who's doing the deployment if we can push that a bit. --Lydia Pintscher (WMDE) (talk) 17:31, 8 June 2015 (UTC)

By popular demand I've just squeezed in ruwiki and cswiki for June 23rd. @Vlsergey: Please make sure you're carefully scaling it up on ruwiki to see how performance and so on goes. --Lydia Pintscher (WMDE) (talk) 12:53, 12 June 2015 (UTC)

Thanks a lot! We will start using AA very carefully. -- Vlsergey (talk) 12:57, 12 June 2015 (UTC)

Evaluation of WD by WMF

Sorry, but was this evaluation of WD by the WMF discussed in the project chat ? Thanks Snipre (talk) 20:50, 7 June 2015 (UTC)

  • Very interesting text. I can hardly disagree with conclusion, except "Proposed Wikidata improvements":
    • It's not possible to introduce datatypes except instance of (P31). We can create API to create some filters like "select only humans", but nothing more.
    • Focus shall be made on trustworthiness, not localization of some descriptions.
    • Database like Wikidata does not need any kind of "Importance levels for properties".
  • -- VlSergey (gab) 21:09, 7 June 2015 (UTC)
Will try to read this tomorrow. I already don't understand the pertinence of the research questions. Seems they didn't get to Zolo's item group analysis either. --- Jura 22:04, 7 June 2015 (UTC)
Actually, I like research questions 4 to 7. It would be interesting to do a study on these. --- Jura 04:54, 8 June 2015 (UTC)
It is difficult by pure statements describe the difference of importance between being honorary scout and king of Sweden, Q45068 is both. How should we describe the scale of importance between them? Depending of your POV, its importance is different. Compared with any other person being king of Sweden, Q45068 has probably not been any better or worse than anybody else could have been. But as an honour scout, he has made a big difference.
The example mentioned in the report tells another problem with pure databases. That Barack Obama got the Nobel peace prize has made a lot to the reputation of the price, and not in a good manner. But that is hard to describe in a statement, and it's my opinion that we should restrain us from trying to describe that into our database.
I do not like the emphasis added to images in the report. Choosing good images is something that I prefer to see humans and article-editors do. Why should every Wikipedia use the same picture of John Doe? And how should we solve the caption problem? File:Skule Tostesson.JPG depicts Skuli Tostisson Kongsfostre (Q6180419), but does he look that way? No, it's just a wooden scupture of somebody we do not know how he looked like. It can still be used as an illustration, but it needs a caption which tells it's not him.
Almost all of the descriptions added to items which do not have any sitelink in the related language are made by automatic tools. It makes good statistics, but that is all. Almost every item related to a category have a sv-description. But that description tells nothing more than it is a category. Since that is visible both in the label and in the sitelinks, almost nothing of value is added. -- Innocent bystander (talk) 09:16, 8 June 2015 (UTC)
The weekly update gives a different perspective on the numbers about images. Have a look also at Zolo's numbers about Wikipedia content, especially "cebwiki". --- Jura 10:07, 8 June 2015 (UTC)
I'm surprised that they did not considered instance of (P31) and subclass of (P279) in the hierarchical properties. That's kind of weird. TomT0m (talk) 11:08, 8 June 2015 (UTC)
Indeed. Furthermore, it's a complete mess to advice to focus on descriptions. Descriptions don't contain any valuable data. We should rather focus on adding more and more statements to our items and then generate descriptions based on these statements. I fear that the author of that evaluation does not have the best understanding of what we need at Wikidata. However, we have to remember that they mainly focused on how Wikidata can help the Foundation's projects and not how Wikidata can be improved in general. -- Bene* talk 11:37, 8 June 2015 (UTC)
Yes, note that they want us to start eliminating English and German labels, to restore balance. There is a lot about descriptions, but the seem to feel these are intended to define the items. Maybe they are confused with Freebase? Anyway, the way forward is clear enough, by and large. - Brya (talk) 16:37, 8 June 2015 (UTC)
When there is an English description, I see that. It's enough for me and most likely anybody else Swedish-speaking person who dare to come here. It maybe works to have automatic tools adding description based on statements, but I'm afraid that would look even worse than Google translate. Google translate is good at spelling, but put together a sentence that makes sense with a grammar that isn't confusing, is it less successfull with. -- Innocent bystander (talk) 16:56, 8 June 2015 (UTC)
Long term, it will become possible to producing descriptive text, given enough statements. After all, this is what bots have been doing for awhile now. But it will need work (what these bots are writing, all too often, indeed looks a bit strange). And the right way to do this will differ per topic. - Brya (talk) 17:47, 8 June 2015 (UTC)
Perhaps this shouldn't be done by bots but by Wikibase itself because the statements may change and even better automated descriptions will become available. Only where automatic descriptions don't work humans should enter one to override it. In all other cases no description should be set and Wikibase should dynamically create one. -- Bene* talk 17:51, 8 June 2015 (UTC)
Or .. showing boths :) I proposed a solution for this, associating a Wikitemplate in each languages to a class/query. If the item is in the result of a query (or an instance of a class, said in a different way), then the template can be used to generate the description. Maybe it's in the result set of several queries, then we should know which query are the most specific (or said differently, the deepest in the subclass tree). TomT0m (talk) 17:57, 8 June 2015 (UTC)
itwiki generates the first line of text for biographies with a template, see it:Category:BioBot --- Jura 18:11, 8 June 2015 (UTC)

Ironholds, please publish the dataset of your study. --Succu (talk) 21:58, 8 June 2015 (UTC)

One thing Ironholds asks for is a way to find any date, any image, any coordinate for an item. This is what the "search (including sub-properties)" function was meant to deliver but Lydia and the dev team don't want to work on that yet. Sorry if that sounds snarky but the lack of this function means that we will have to create unproductive workarounds which will be baked into the structure to such an extent that by the time we get this function it will be too late to unravel the mess. If the WMF would like to work on delivering this function sooner then they will have my eternal gratitude. I promise I will never again complain about them. Filceolaire (talk) 14:01, 10 June 2015 (UTC)

What he's asking for as I understand it is getting an image, date or coordinate for an item no matter what property it has. That's already possibly when filtering by datatype. --Lydia Pintscher (WMDE) (talk) 14:12, 10 June 2015 (UTC)
Yes Lydia Pintscher (WMDE), you can get a long way towards what he is looking for by filtering on datatype but if you filter on "Commons file" you might get a picture or a map or a recording or a video. Being able to filter on just the pictures or just the graphics type files (pictures plus maps) etc. would, I believe, be more useful for what he is trying to do. Filceolaire (talk) 11:24, 15 June 2015 (UTC)

After reading the analysis, I would suggest that we take the general idea of the report and provide our own analysis. I am rather confident that with our better understanding of Wikidata we can provide an improved alternate report. Thoughts? --Denny (talk) 15:56, 10 June 2015 (UTC)

Among the questions raised in the report, I think the most important, and complex, one is "place holder articles". I can only imagine it to be a stub article with a bot-generated short text and a 100% wikidata based infobox. The point that should be investigated is the one the report avoided: infoboxes.
We can do a lot with Wikidata. fr:The Octagon House and fr:Wang Hui seem to work alright, and will work better will arbitrary access. However, they require much more client-side coding than a few #property calls here and there. Writing and maintaining everything in every small wiki does not seem very realistic. My impression is that the one really-needed feature is a central repository of Lua modules. --Zolo (talk) 16:56, 10 June 2015 (UTC)
See Wikidata:Article placeholder input for the article placeholder. --Lydia Pintscher (WMDE) (talk) 16:58, 10 June 2015 (UTC)
Ok thansk, so my concerns about template complexity may not really apply to placeholder, but they still apply to infoboxes and other complex templates, which are arguably the most important use of Wikidata in Wikipedia. --Zolo (talk) 19:58, 10 June 2015 (UTC)
A Template central repository would really be great, importing a template to another wiki, chase the bugs and correct it is a pain, especially for complex template. Lua dev tends to look like PHP development, yet compare Mediawiki coding infrastructure, ticket tracking, test environment, merges with wikitemplate and lua development ... A cross language tool for this would be good. Not sure how to implement this efficiently though :) I even begun a set of script to download/upload versions of a lua module on my computer to run tests, and fake module interface to Wikidata (unfortunately my hard drive died and it seems I had not pushed them on github :/) TomT0m (talk) 20:16, 10 June 2015 (UTC)

Connecting items to s:de:Paulys Realencyclopädie der classischen Altertumswissenschaft

Hi, I am working with the project s:de:Paulys Realencyclopädie der classischen Altertumswissenschaft ("RE") in the German Wikisource. How can we connect our articles to the accurate Wikidata/-pedia items?

Take two examples: Q20000000 and Q1534902. I wanted to connect the Wikipedia items to the Wikisource items. This makes sense on the ground that the RE is an encyclopedia, and every article in it can possibly have an equivalent in Wikipedia. So Q20000000 and Q108429 should be a single item, linking to the Wikipedia pages as well as the Wikisource site s:de:RE:Bogenschiessen. However, somehow I can't put a connection in place. Why not? What is the Wikidata policy here?

Thanks in advance, --Tolanor (talk) 21:58, 8 June 2015 (UTC)

Okay, now I understand - it was a stupid editing error on my side. However, now I want to merge Q19996427 and Q1534902. Is there any policy against this? And can we somehow do this on a larger scale, automatically? We want to integrate our RE project into Wikidata, as the project includes a lot of very useful raw data. Can anyone help with this? --Tolanor (talk) 22:04, 8 June 2015 (UTC)
@Tolanor: see this discussion: Wikidata:Project chat/Archive/2015/05#Items for Wikisource. No, you should not merge Wikisource encyclopedia items and main Wikidata items ("main topic") because even at dewikisource you have multiple encyclopedias. -- Vlsergey (talk) 05:44, 9 June 2015 (UTC)
I don't really see a consensus about this in the discussion you mentioned. Isn't it possible to link more than one de-wikisource article? --Tolanor (talk) 11:01, 9 June 2015 (UTC)
You cannot (at least should not) link them by sitelinks, but it's still possible to link them by the help of the statements in the items. Here I have tested adding interwiki between WS-articles by the help of properties, instead of sitelinks. Since fiwikisource has four relevant pages, they cannot share one item. They also have different authors and other kind of metadata, and should therefor not share items. -- Innocent bystander (talk) 11:08, 9 June 2015 (UTC)
@Tolanor: A Wikinews article, as a news article, is imho better with an item on his own, like a scientific article and so on. We then can use the same properties (subject, author, ...) on any item. Something in Wikinews that could be linked to items about a topic could be, for example, pages that lists the articles about the topic ... TomT0m (talk) 11:33, 9 June 2015 (UTC)
You can use described by source (P1343) to connect them. --Infovarius (talk) 18:22, 9 June 2015 (UTC)
Okay, now I see the problem. Is it possible to somehow connect the RE-items on Wikidata with their subjects automatically? Do you maybe have a bot for that? We have some 17.000 RE articles on Wikisource. Tolanor (talk) 00:07, 12 June 2015 (UTC)
There is no way to do so without a huge margin of error. RE article names are, in 999 out of 1000 cases, not coherent with Wikipedia page names. We would need to set links by hand, I'm afraid. Jonathan Groß (talk) 07:33, 12 June 2015 (UTC)
In Bogenschiessen (Pauly-Wissowa) (Q20000000) there is statement, main subject (P921): archery (Q108429). You can design a template at dewikisource that uses that information to create a link to "w:de:Bogenschießen". -- Innocent bystander (talk) 07:57, 12 June 2015 (UTC)
It should be possible to extract and link automatically the articles already defined in the template. As P1343 is for offline references, you might want to create a dedicated property. --- Jura 08:03, 12 June 2015 (UTC)
@Jura1: we are using described by source (P1343) for both online and offline sources. It just more convenient to have separate property for each online source, but not required. And encyclopedia article is obviously P1343 case, even if it has URL or Wikisource link. See Arthur Conan Doyle (Q35610) for example. -- Vlsergey (talk) 10:02, 12 June 2015 (UTC)
Well, you are not supposed to do this. If you think it's not convenient, it would even be preferable to not suggest it. --- Jura 10:09, 12 June 2015 (UTC)
@Jura1: actually, described by source (P1343) was created especially for this purpose. How can you tell that i'm not supposed to do this if this is exactly designation of this property? -- Vlsergey (talk) 11:33, 12 June 2015 (UTC)
In English, the description (on property talk) reads "better to avoid described by source (P1343) for online resources". --- Jura 11:35, 12 June 2015 (UTC)
Feel free to fix it. The better description would be something like "better to avoid using this to link item to online databases, would they have their own distinguish properties. Social network links use website account on (P553) / website username (P554)". -- Vlsergey (talk) 12:56, 12 June 2015 (UTC)
@Tolanor: yes, it is possible. It quite simple bot work if you have link between Wikisource article and Wikipedia article somewhere. -- Vlsergey (talk) 10:02, 12 June 2015 (UTC)
@Tolanor, Vlsergey: Please be cautios. Adding main subject (P921) with appropriate qualifiers will require careful control. For example, the RE article RE:Abactores was linked to a Wikipedia page that does not match the articles scope [8]. This has since been removed, but there are many other cases that we can't list automatically.
I, too, see that it would be beneficial to link the RE article items with corresponding Wikidata items. For example, the RE project on Wikisource uses only Wikipedia as an external reference, whereas Wikidata incorporates items from all Wikimedia projects and makes it possible to make way more references.
If a bot were to add main subject (P921) to the RE article items, we must manually check all statements afterwards and then (a) somehow document that main subject (P921) is correctly used or (b) correct the statement or (c) remove it where there is no appropriate Wikidata item to link. Jonathan Groß (talk) 17:03, 12 June 2015 (UTC)
@Jonathan Groß: I think incorrect main subject (P921) of Wikisource dictionary article is not actually a big issue. It has the same value in Wikisource for a long time and they were okay with that. We are just about to move that from Wikisource to Wikidata. -- Vlsergey (talk) 17:15, 12 June 2015 (UTC)
I think having to check thousands of instances will be a notable issue. Jonathan Groß (talk) 21:22, 12 June 2015 (UTC)
There is no big difference checking Wikisource articles or Wikidata items... Wikidata can be faster actually. We can create summary sheet for items, for example. -- Vlsergey (talk) 22:45, 12 June 2015 (UTC)
What do you mean by "summary sheet"? For what items exactly? How will that sheet be layed out? Jonathan Groß (talk) 07:44, 13 June 2015 (UTC)
I'd also say that correcting the links on Wikidata would be faster than on Wikisource, as the crowd here is larger. Also, I don't think that there are many wrong links left - I just removed all links to "Rechtswesen im antiken Rom". Are there more of this kind, Jonathan? --Tolanor (talk) 19:07, 15 June 2015 (UTC)
@Tolanor: yes there are. I don't know specifics, but Roman gentes (Iulius, Aelius etc.) and people named Alexandros are prone to that. I didn't take the time to check all hundreds of articles, but you're welcome to do so ;) Thank you for your effort, BTW. Jonathan Groß (talk) 20:07, 15 June 2015 (UTC)
Ah, yes, I remember that faintly. I have gone through all Iulii now and the Aelii until s:RE:Aelius 100, there weren't many bad links. Will continue another time. --Tolanor (talk) 01:34, 16 June 2015 (UTC)

Come to think about it, after the links are established it would also be nice to have a bot compare the connections between the Wikipedia and the RE items on Wikidata and in the Wikisource project on a regular basis (say, every three months), so that we can add or correct links that have been corrected either on Wikidata or on Wikisource, and so that everything is synchronized. Would that also be possible? --Tolanor (talk) 19:12, 15 June 2015 (UTC)

I think so. BTW Tolanor: Check out Claudianus 10 (Pauly-Wissowa) (Q20158337). I linked it to a Wikidata which (currently) has no German Wikipedia article. Of course, I did it by hand, but it is a good example how Wikidata is useful in linking the "main topics" with Wikipedia content. Jonathan Groß (talk) 15:05, 19 June 2015 (UTC)

merging Fregata and Fregatidae

Hi folks, not been here before - the interwiki links are two mutually exclusive lists, which I think has arisen by happenstance. The genus and family are both used as terms for "frigatebird" so it'd be good to merge the items. cheers, Casliber (talk) 11:47, 9 June 2015 (UTC)

Well, perhaps the fact that the genus and the family each have a scientific name of their own should have given you the hint that these are two distinct entities? - Brya (talk) 16:18, 9 June 2015 (UTC)
Yes, but on all wikis, the two taxa are treated in the same article. The difference is just that some Wikipedias use fregatidae as the scope, with Fregata redirected, while the others do the opposite. Both taxa have the same content, and this is reflected on all Wikis, therefore they should be merged. FunkMonk (talk) 10:47, 10 June 2015 (UTC)
@FunkMonk: Wikidata is not Wikipedia, different entities, different defintions, needs different items. See Wikidata:WikiProject Taxonomy for an introduction to taxonomy in Wikipedia. Such cases should bo solved by either traditional interwikis, or by redirects in Wikipedia (although atm it's not well supported by Wikipedia). Of course in this case it's due to a limitation of the model of traditional taxonomy :) TomT0m (talk) 11:24, 10 June 2015 (UTC)
Think of it like this Casliber : What is the parent taxon (P171)? Fregata and Fregatidae will have different answers to this question so they need different items to store those different answers. The easiest way to decide if you need one item or two is to think about the statements we can make. Can those statements be added on the same item (maybe with a start time/end time qualifiers)? If not then we need two items. Filceolaire (talk) 14:58, 10 June 2015 (UTC)
Well, "Both taxa have the same content" cannot be true: taxa do not have content. And "on all wikis, the two taxa are treated in the same article" is not true, and wouldn't really be an argument if it were true ("All my friends jumped off the bridge, so ..." - Brya (talk) 16:27, 10 June 2015 (UTC))
@Brya: I think he meant the same instances :) Unless if there was another extinct taxon in the tree, he's probably true. TomT0m (talk) 16:40, 10 June 2015 (UTC)
As far as I see, Wikidata does not have a page for every named higher level taxon, only those that have Wikipedia articles. English Wikipedia has a combined genus and family article about frigatebirds. Other Wikipedias do too. The difference is just whether they focus on the family or genus. Which creates the problem of incomplete interlinking. FunkMonk (talk) 23:26, 12 June 2015 (UTC)
Yes FunkMonk, Wikidata does not yet have an item for every named higher level taxon but it will, eventually, have all of these. Filceolaire (talk) 11:32, 15 June 2015 (UTC)
We have nearly 40,000 taxa without a sitelink. --Succu (talk) 12:09, 15 June 2015 (UTC)
Well, the problem will probably not be fixed even then. I will try to explain the issue again: Fregatidae includes a single genus, Fregata. Therefore, the species content of the two is the same. This has led to the English Wikipedia article using the genus name as "the scope" of the subject "frigatebird" with Fregatidae as a redirect, whereas for example the French Wikipedia has used "Fregatidae" as the scope of "frigatebird", with Fregata as a redirect. Both articles are about the exact same subject, and there will never be a separate fregatidae article on English Wikipedia, like there will never be a separate Fregata article on the French Wikipedia, since they are covered in the same articles, as they should be, as they have exactly the same species content. When a family contains a single genus, it is redundant to create an article about each. But the problem is, that now they cannot link to each other on Wikidata, simply because the English article is anchored to the genus name, whereas the French article is anchored to the family name. How will this be fixed? FunkMonk (talk) 13:01, 25 June 2015 (UTC)
@FunkMonk: There is a project page devoted to such problems: Wikidata:WikiProject Cross Items Interwikis. I think one possible solution is to add a module to the Taxobox with an algoritm that looks if the taxa is monotypic, and if it is, look for potential missing interwiki in the higher (or lower) level taxa, and transclude that interwiki into the page if it is needed. It's therefor solvable. But it is my opinion that it's the decision of the local projects, if they want to implement such a solution or not. If the frwiki/enwiki do not want such interwiki, we should not force them to have it. -- Innocent bystander (talk) 13:17, 25 June 2015 (UTC)
Whether something is "monotypic" is a taxonomic question, and taxonomy can change without notice (and often enough does change). - Brya (talk) 16:39, 25 June 2015 (UTC)

Ordering of statements

Has there been any discussion on perhaps fixing the order of statements in items? It would be nice, for example, if the place of birth (P19) could be next to the date of birth (P569). MSGJ (talk) 13:26, 10 June 2015 (UTC)

We used to be able to move statements around through the HMI. That was removed because the code was bad, IIRC. I think maybe one can still do it via the API. There's at least one gadget/script that sorts things into similar locations but not precisely sure of the name. --Izno (talk) 20:13, 10 June 2015 (UTC)
@MSGJ, Izno: It's User:Soulkeeper/statementSort.js. — Ayack (talk) 06:59, 11 June 2015 (UTC)
I'll try the script, thanks. But probably better if the software automatically displayed statements in some defined order? MSGJ (talk) 08:11, 11 June 2015 (UTC)
Is there a screenshot of the tool? Would be great for Wikidata:Tools/User scripts#StatementSort.
See also:
and

Monolingual text vs. multilingual text

I just stumbled over the newly created taxon common name (P1843) which is of the datatype monolingual text. But the way I understood that datatype it is supposed to be used for statements which has exactly one (or very few) combination of language and text, like in native label (P1705). But a common name could be in any language, so this would have been a property for the multilingual text datatype, which sadly is still under development. Was there any consensus to stop waiting for that datatype and create properties in the wrong datatype first and move them to the right one by bot later? Or was that property created to hastily without thinking of the correct datatype? Ahoerstemeier (talk) 14:08, 10 June 2015 (UTC)

See the creation discussion at Wikidata:Property_proposal/Archive/30#common_name. Filceolaire (talk) 14:43, 10 June 2015 (UTC)
Maybe the label should be more obvious that it is supposed to be used for taxa? The common names for taxa are constantly changing, but normally only in one language at the time. -- Innocent bystander (talk) 15:00, 10 June 2015 (UTC)
Done boldly. --Izno (talk) 20:14, 10 June 2015 (UTC)

Bonny and Clyde ; arbitrary access

Hi, Arbitrary access is (almost) on.

Time to think of what we can do with lua and interwikis. So I'll start a discussion with a few ideas.

99of9
Achim Raschka (talk)
Andrawaag (talk)
Brya (talk)
CanadianCodhead (talk)
Canley
Circeus
Dan Koehl (talk)
Daniel Mietchen (talk)
Enwebb
Faendalimas
FelixReimann (talk)
Hyperik (talk)
Infomuse (talk)
Infovarius (talk)
Jean-Marc Vanel
Joel Sachs
Klortho (talk)
Lymantria (talk)
Magnefl (talk)
MPF
Manojk
MargaretRDonald
Mellis (talk)
Michael Goodyear
Mr. Fulano (talk)
Nis Jørgensen
Oronsay
PEAK99
Peter Coxhead
PhiLiP
Andy Mabbett (talk)
Plantdrew
Prot D
pvmoutside
RaboKarbakian
Rod Page
Strobilomyces (talk)
Stuchka (talk)
Succu (talk)
TiagoLubiana (talk)
Tinm
Tom.Reding
TomT0m
Tommy Kronkvist (talk)
Tris T7 TT me
Tubezlob
William Avery
Minorax
Culex
Koala0090
Mike Krüger
Friesen5000
Salgo60
TED
GoEThe (talk)
Estopedist1
Leptospira
  Notified participants of WikiProject Taxonomy : If a taxon have a parent taxon, and no wikipedia article in some language, but its parent is, is it a good idea to show a link to the closest Wikidata item with the most interwikis ? Especially interesting for monotypic taxon.

  • Bonny and Clyde : If a band member or group of person member has an article and an item, but no interwikis in most language, is it a good idea to provide interwikis to the item of the band or the group ?


More generally, if an item has no interwikis, but an item he is linked to by a statement has a lot of interwikis, is it a good idea to provide a link to Wikidata for this many interwikis ?

Time to be creative :) TomT0m (talk) 18:02, 10 June 2015 (UTC)

  • I don't get why arbitrary access matter for this. Would this be some kind of JS, any project can decide on it's own policy, and JS don't need arbitrary access enabled. Would this be part of out-of-the-box MediaWiki setup, arbitrary access isn't required as well. -- Vlsergey (talk) 18:07, 10 June 2015 (UTC)
    @Vlsergey: Because we could also code a template for this an insert it in articles, say {{Interwikis}}. This would be far less extensive to Wikimedia servers.
    @TomT0m: Well, yeah. But that template would obviously have per-project policy. -- Vlsergey (talk) 18:27, 10 June 2015 (UTC)
    Solving this in Mediawiki is not really in the projects of the devteam, so yes it would require a per project policy. And to do this generally and do not depend of specific hardcoded properties, we would have to have some kind of hook in mediawiki so that we can in-wiki be populated by for example a template or in an old interwikis like fashion. Maybe the template could gererate old style interwiki links ... I don't know if it's possible. Then a bot would just have to insert the template in articles. TomT0m (talk) 18:35, 10 June 2015 (UTC)
  • Personally I would like to merge parent (and only single one) links using part of (P361) property. -- Vlsergey (talk) 18:07, 10 June 2015 (UTC)
    I don't get it, you mean merge the items ? Can you provide an example ? Merging items is off topic. TomT0m (talk) 18:21, 10 June 2015 (UTC)
    I mean merge interwikis of course. For example we have Bony. Bony have aa-sitelink and bb-sitelink. But Bony&Claid have en-sitelink, de-sitelink, etc. Thus Bony on AA-site should have bb-intewiki (directly from Bony item) and en-, de-, etc-interwiki from parent entity (linked by part of (P361)). I.e. if we have interwiki for particular project in item itself, we taking it. If we don't -- go to parent (part of (P361)), check sitelink there. But only once (don't go to grandparent). -- Vlsergey (talk) 18:27, 10 June 2015 (UTC)
    @Vlsergey: OK, I get it, possible maybe if generating old style interwikis by a template is possible. TomT0m (talk) 18:37, 10 June 2015 (UTC)

I have already done some experiments with interwiki with arbitrary access. See here (Raven (Poe)). It is here done with the help of P629 and P747. -- Innocent bystander (talk) 06:59, 11 June 2015 (UTC)

@Innocent bystander: (an all others) : I'll create a Wikidata:WikiProject Cross Items Interwikis to help coordinate the efforts. TomT0m (talk) 15:39, 16 June 2015 (UTC)

Myxogastria (Q46309) and Myxomycota (Q10334205)

There are links right? --Rippitippi (talk) 18:09, 10 June 2015 (UTC)

And Mycetozoa (Amoebozoa) (Q378789) is related too. Need a botanist to sort these out! en:Myxomycota is a disambiguation page so I've detached that one from Q10334205. MSGJ (talk) 22:40, 10 June 2015 (UTC)
The main problem is that nobody put in references to document the various taxonomic positions, in a field that has seen many changes proposed. And perhaps ranks have been mixed up in the linking. But a headache, yes. - Brya (talk) 06:02, 11 June 2015 (UTC)

Some Questions - Analyzing Wikidata for Business Use & Potential Alternative UI

Hi Wikidata contributors,

I'm pursuing what may be an alternative interface to Wikidata focusing on a spreadsheet style tabular interface for statistical analysis that has access to mass miscellaneous data (including a focus on two-way access, ie, having users incorporate many ontologies/databases in their project and then optionally contributing the additional data back to Wikidata after appropriately translating it). Point is - it's something that may be of value to Wikidata, resulting in increased contributions from and much increased use of the data. I'm still doing early feasibility analysis, and I have a lot of questions about Wikidata. A lot of this stuff may be noobish/basic, I need education! :) I would love some help, if anyone knows, or cares to help me find the answers –

  1. First of all, and most importantly, is anyone aware of any reports on the completeness/comprehensiveness/cohesion of Wikidata's data, as well as info about cross-pollination between Wikidata, Freebase, DBPedia, etc? I've seen a lot of raw disorganized quasi-valuable data, but I don't believe Wikidata is a business-usable product without these kind of status reports readily available so I'm surprised at what little I've found, but, if necessary, will be procuring it myself (in fact, I have already started.) As Wikidata nears higher levels of completeness, these reports can also direct future work.
  2. What is policy (& goals?) on approximations, confidence levels, and the related and easily-confused topic of value ranges? I know Wikidata clearly accepts approximations already, and some cases relating to this I've noticed:
    • Identifying Earth's shape as an 'Oblate Spheroid' shape Q2 (it is not, it is very bumpy, oblate spheroids are smooth and can only be an approximation of Earth's shape)
    • Ranges used with DataType:Quantity, but can find no documentation on what that is supposed to mean. Is it a confidence interval, with an assumed 100% confidence level? I've also seen "precision" on global coordinates. I've also seen
    • Multiple statements for the same property with different ranks. Surely there is a better method of displaying confidence than a three-tiered grading system, but I see it used for this task as well as others (such as highlighting the most recent release of some metric.)
    • I've noticed qualifiers being used to identify some statements as "estimates", with no further info.
    I can't figure out any sort of policy tying this stuff together ('this stuff' being (a) expressing probability distributions (confidence levels at various values)(aka 'how reliable is this data?') and (b) expressing distributions of values (such as how to express the location of London, considering it is distributed over a wide area and not a single point)). What is the purpose of ranks, ranges, and precision? What is the policy on approximates and estimated values?
  3. What is the policy on bad/wrong data? The project goals [9] state "Wikidata will not be about the truth, but about statements and their references. These can be contradictory." - is this part of the current policy for Wikidata in its current stage? If so, why is so little data sourced?
  4. What is the policy on database normalization? For example, I notice World Ocean has the property 'part of' and statement 'Earth', yet Earth independently maintains the property 'has part' with statement 'World Ocean'. I see that 'part of' is identified as being an 'inverse of' 'has part'. I am very tempted to edit one to determine if any bot has tasked itself with synchronizing these relationships, but do not wish to be flagged as a vandal. I've also noticed Constraints, but can find no meta info about them or how they are supposed to work, be created, or be enforced.
  5. Why do people contribute to Wikidata? What is the main driving factor behind the main volume of individual edits? Some guesses to show the level of detail I'm looking for:
    • People working on Wikipedia who find themselves having to also edit Wikidata because of transclusions?
    • People trying to increase use of data they or their organization publishes. A more business-understandable initiative would be people trying to increase exposure to the fact their org publishes such data, but wikidata has poor uptake as of yet, and no CC-BY, etc, so I'm not so sure.
    • People interested in the fundamental proposal of a huge, broad, open ontology, and want to see it happen and find themselves doing minor edits? (my guess is that this is most of the energy on wikidata right now, but it doesn't materialize as statement/item edits, it instead involves managing properties, wiki pages, bots, etc)
    • People documenting their personal knowledge/interest areas? (ala wikipedia)
    • People somehow browsing the data and just fixing what they see is wrong? (is ANYONE browsing, except via Wikidata Game?)
  6. Is there any kind of Wikipedia-style 'Userspace' for data on Wikidata? The use-case is to ease importing large amounts of data which may or may not be configured correctly, providing a purgatory to validate it first.
  7. Was schema.org ever looked at in the development of Wikidata? I think Wikidata already has a superior and undeniably larger schema, but I'm curious as to hear the perspective from someone more knowledgeable and closely connected to Wikidata.

Thanks. I'm sure most of this has been discussed before, I just can't find it. :) I look forward to doing the data report (if still needed), then continuing to pursue the alternative UI.

-- Pcg01 (talk) 20:05, 10 June 2015 (UTC)

  • Quite a large set of questions:
    1. As far as I know we have only statistical view of the database (see WD:Statistics, Special:Statistics and Wikidata:Database reports )
    2. There is no policy about the use of the interval. Qualifiers can be used in a more accurate way but this is the task of the contibutors to create the items to present the methods of calculation, the problem is more a question of habit to look for that information and to consider that methods and confidence are integral part of the values (we already have difficulty to raise awareness among contributors of the need to have for each value the corresponding reference). Most of the features (rank, uncertainty,... ) were developed but not really fixed by rules for their uses, some discussions started about the need of those features but never finished on a properly description of their final uses.
    3. For policy about bad/wrong data, the idea is to use the rank feature to distinguish between the different statements.
    4. For the policy about database normalization, I have no idea.
    5. More a personal point of view: we handle a large quantity of data in WP (especially in infoboxes) and WP is not appropriate to store and to maintain this data. WD is the answer to that need of handling large data sets. The general ontology is not a priority because as international community WD has more difficulties to reach consensus.
    6. I don't understand the question.
    7. I don't know. Snipre (talk) 07:56, 11 June 2015 (UTC)
Thanks for the insight, Snipre, lot of useful info. Is there any more discussion I can read regarding WD & the international community reaching consensuses? I've seen this mentioned in journalism pieces about WD, but have not seen internal discussion. It just feels like something with a technical solution, and I want to see how much that has been explored. --pcg (talk) 06:49, 13 June 2015 (UTC)
  • I think question 6 is about a place where data imports can be tested. You can do that here. Sjoerd de Bruin (talk) 08:00, 11 June 2015 (UTC)
Thanks, I'll use the test install --pcg (talk) 06:49, 13 June 2015 (UTC)
  • Re #4: Constraints can be added to a property by adding the relevant template to the property's talk page. Lists of constraint violations are linked to from the talk pages, and can be repaired manually. I believe the developers are working on a system to make constraints more visible to editors, along with a system to semi-automatically add/change/remove inverses where appropriate.
  • Re #6: There is no equivalent of the Wikipedia userspace for items, but there is a Test Repository, as well as a Sandbox here on the project itself. --Yair rand (talk) 08:09, 11 June 2015 (UTC)
A constraint that is manually maintained isn't much of a constraint ;) I look forward to the further developments.. --pcg (talk)
@Pcg01: It's not worded correctly. By design, when the datatype of a property is item, there is no constraint on the kind of item that can be used. Moreother there is no hard-wired type system in Wikibase below the Wikidata Data Model (Q16354757)     . You can read this blogpost from Denny to understand the design decisions. As far as I understand, even with deeper constraint integration this still be possible to enter values that violates the constraints : we found a kitty mayor :) TomT0m (talk) 10:28, 13 June 2015 (UTC)
Wow, thank you for that link to the WMDE blog. The article you linked + the other articles answer so many areas of confusion I had. :) The decisions regarding the constraints seem very reasonable. Wish this information was more easily accessible. --pcg (talk) 15:36, 13 June 2015 (UTC)
The most useful statistics are here. Thanks, GerardM (talk) 16:41, 14 June 2015 (UTC)

Wikidata Community User Group?

Dear all, I was wondering if there would be interest in creating a Wikidata user group as other communities like Wikisource or Wikinews have. The advantage of having a semi-informal volunteering organization is that it helps to build the sense of community even further by helping to organize ourselves as volunteers IRL. It might be useful as a first step towards bigger Wikidata community events, or as a help to the development team to prioritize, or just to have fun :) Anyone interested?? --Micru (talk) 21:45, 10 June 2015 (UTC)

user:Micru It is a good idea. I would watch the activities and join as a member but would not volunteer for any leadership role. I also think that it is better to form these things early to begin discussion so that after 1-2 years if there is a reason to develop the group then there is already some base established. In the current user group scheme, it takes three people to volunteer as founders to establish a group. I would not be one of those three but I would be a fourth supporter. Blue Rasberry (talk) 14:12, 11 June 2015 (UTC)
I'm happy to support such an effort. Let me know if you need me. --Lydia Pintscher (WMDE) (talk) 14:18, 11 June 2015 (UTC)
I would initially be interested (though eventually I might not be able to travel if needed).--Ymblanter (talk) 17:28, 11 June 2015 (UTC)
I have some formal flags there, if people think this would create conflict of interests, this is fine with me.--Ymblanter (talk) 17:30, 11 June 2015 (UTC)
I'll be happy to participate too (but I would not take any very time consuming leadership role). Tpt (talk) 21:44, 11 June 2015 (UTC)
@Bluerasberry, Lydia Pintscher (WMDE), Ymblanter, Tpt: I have started the gathering page here: meta:Wikidata Community User Group. Feel free to join :) Btw, we might need a logo, any ideas about the concept? --Micru (talk) 10:26, 12 June 2015 (UTC)
I really like the idea of strengthening the Wikidata community. Do you know what the legal side of such a group is? What are the rights and duties of such an organization? Would it be possible to use Wikimedia logos and merchandising? What about the legal responsibility of the leaders? -- Bene* talk 11:55, 12 June 2015 (UTC)
The requirements for the user groups are very low, only three members required plus the signature of a code of conduct which can be found on that page too. The whole process is described here. Normally the user group has a logo, but it can be asked for permission to use other logos according to the trademark policy. The legal responsibility of the promoters are as other volunteers, nothing special. I think it can help to give the community a visible face, or at least gather the volunteers who want to give the community a visible face, which in turn will strengthen the community and its relationship with other organizations in or outside the wikimedia movement.--Micru (talk) 20:40, 12 June 2015 (UTC)

QuickStatements

Hello everybody, I just tried to add the newly created Database of Classical Scholars ID (P1935) to all eligible items (about 250) with QuickStatements, and it gave me the following error message:

   Processing Q1738885 (Q1738885 P1935 2)
   ERROR (set_quantity) : Invalid snak (Can only construct StringValue from strings)

As it seems, QS doesn't let me add number values as strings. But Database of Classical Scholars ID (P1935) demangs a string, even if it is in the form of a number. How can I persuade QS to let me do that? Jonathan Groß (talk) 08:04, 11 June 2015 (UTC)

All strings in QuickStatements have to be in double quotes. Have you tried doing that? - Nikki (talk) 08:13, 11 June 2015 (UTC)
No, I didn't. Silly me, it says so in the QS help. Thank you for reminding me. It's working now. Jonathan Groß (talk) 09:04, 11 June 2015 (UTC)

Primary Source filled with some data from Freebase

Hey everyone,

As you may already know I am currently employed by Google (Q95) to work on the possible importation of Freebase (Q1453477) content into Wikidata.

I have just uploaded into the Primary Source Tool a first set of 1 million of statements containing only some simple statements (e.g. without qualifiers) with as object an other Wikidata item (more data are coming soon).

Feel free to test the tool and to do bug reports or suggest improvements on Wikidata talk:WikiProject Freebase. I have created a list of known issues. More data are coming soon. Tpt (talk) 21:36, 11 June 2015 (UTC)

Change described by source (P1343) qualificator for Wikisource articles

Repeating @Infovarius: proposal from Wikidata:Project chat/Archive/2015/05#Items for Wikisource: replace recommended qualifier for described by source (P1343) property for Wikisource linking from stated in (P248) to statement is subject of (P805).

with qualifier stated in (P248) with value BaEED / New-York (Q16947036)
=>
with qualifier statement is subject of (P805) with value BaEED / New-York (Q16947036)
(+) statement is subject of (P805) is much better for such link
(-) bot work is required to change all existing qualifiers of property from stated in (P248) to statement is subject of (P805) for all described by source (P1343)
(-) need to change LUA modules and templates at single project (ruwikisource)
(-) need to change gadgets code to use and populate new property

-- Vlsergey (talk) 13:26, 12 June 2015 (UTC)

In statement is subject of (P805) I expect to find a typical WP-article if you know what I mean. But I guess this maybe works when it's obvious that it's a WS-article. -- Innocent bystander (talk) 13:39, 12 June 2015 (UTC)
You shouldn't use statement is subject of (P805) in this case, as it doesn't make sense. The above example is stating that the dictionary is a subject of the linked article item, which is not what is intended. Hazmat2 (talk) 22:00, 21 June 2015 (UTC)
@Hazmat2: what qualifier should one use then? (Not a dictionary is a subject, but a fact that entity is described in dictionary is subject of the linked article item) -- Vlsergey (talk) 22:57, 21 June 2015 (UTC)
It seems to me that the best way is to not use a qualifier at all, but to use the statement, described by source (P1343) and link directly to the article item. In terms of data, I believe this makes the most sense. The article is the most specific source (which is exactly what "described by source" refers to) and its item already states where the article is located. How would you use described by source (P1343) for a book that is a biography on a single subject? I personally would link directly to the book. A dictionary/encyclopedia is a special case where entries are often unrelated, or indirectly related, and therefore we have separate Wikisource pages and separate WD entries. They are compilations of hundreds or thousands of "tiny books." Instead of linking to the entire dictionary, if we have the ability to link directly to the item, I think we should.
However, since you requested another alternative, I think another option is to create a property specifically for entries/sections/etc. that have their own items, though I'm on the fence about this. The important thing to remember is that these are not "Wikisource items" but simply dictionary entry items that have a transcription on Wikisource. I apologize for my scattered thoughts but I'm just trying to get it out quickly as I have to go. Hazmat2 (talk) 23:55, 21 June 2015 (UTC)

Redundancy problem (offtopic)

  • There are several problems with this:
    (1) the item Brockhaus and Efron Encyclopedic Dictionary (Q602358) is redundant as it's already present on item for the Wikisource page (BaEED / New-York (Q16947036)). Besides it makes it needless complicated for users to add this property manually. It gets even worse if you compare this with some of the solutions for texts that are also hosted elsewhere but Wikisource. The Wikisource version is much more complicated to maintain and build.
    Suggestion solution: use a dedicated property that links directly Wikisource.
    (2) described by source (P1343) is meant for offline resources. Not because it has "source" in its name that it's for Wikisource
    Suggested solution: see (1).
    --- Jura 13:51, 12 June 2015 (UTC)
Re(2). BaEED is an offline source, the only difference here is that it can be found on Wikisource. -- Innocent bystander (talk) 13:59, 12 June 2015 (UTC)
Well, once it gets online, it's an online one. Looking at Wikidata:Property_proposal/Archive/22#P1343 it seems to have been proposed for random works in mind. Supposedly some that didn't have dedicated properties. The description on Property_talk:P1343 seems to evolve. In any case, I don't think this is a solution for online sources with a large number of uses. --- Jura 14:04, 12 June 2015 (UTC)
@Jura1: manual work with Wikidata shall be only for fixing problems. Usually Wikidata should be available via gadgets or third party UI. Thus it's not a problem. -- Vlsergey (talk) 14:54, 12 June 2015 (UTC)
Not sure if the guys working on Paulys' understand it this way. --- Jura 15:22, 12 June 2015 (UTC)
  • (2 edit conflicts in a row, I have to type faster...)
Also, lemmas or headings are usually recorded as section, verse, paragraph, or clause (P958) and neither with stated in (P248) nor statement is subject of (P805). But Jura's advice to directly link to the article in presence of a dedicated item is stronger. Thus should suffice. (Optionally provide a page or column number if the article is rather long and describes several New Yorks like state and city).
Innocent bystander: As long as the item's label reflects the offline resources lemma I don't see a real problem. If you insist you would have to provide two different references (imagine the print edition you want to cite is different from the Wikisource one) or create an additional item for the abstract encyclopedia article as a work for which the printed and the Wikisource versions are different editions or manifestations. -- Gymel (talk) 14:08, 12 June 2015 (UTC)
One reason I maybe have doubts here, is that I normally never use the word "article" about pages in WS. -- Innocent bystander (talk) 14:47, 12 June 2015 (UTC)
The problem with omitting original source that we can't select by dictionary anymore. Currently it is possible to select article from dictionaries and ignore all others. In proposed case we need to enumerate all linked sources, check their "part of" and select only ones we are interested in. Additional problem is losing unification. Currently we have the same structure for both online and offline sources, including available on Wikidata. -- Vlsergey (talk) 14:52, 12 June 2015 (UTC)
You can always say . However this article has interesting additional info like an autor of its own, a volume and page number, a publication year of the specific volume, peraps a certain edition of Brockhaus and Efron Encyclopedic Dictionary (Q602358) for all that to make sense and additionally some URLs of places where you can read it online, including Wikisource (we have reference URL (P854) for that but nothing like "reference item" to provide a coarse statement with a confinement to two items simultaneously). So you cannot prohibit Wikidata from having a specific item for that encyclopedic item/article to bundle these properties (I mean, that's all what Wikidata is about...) And you cannot prohibit that more specific items are preferred for linking against broader ones. So you can only hope that there is some uniformity in those items and they all state . -- Gymel (talk) 15:20, 12 June 2015 (UTC)

The reason behind P1343 is not to provide references (this is very interesting side effect), but to provide a way to obtain links to per-project predefined list of dictionaries. For example, ruwikisource need a way to select links to other dictionaries (incl. Britanica 1911, Brockhaus and Efron Encyclopedic Dictionary (Q602358), etc) from main topic item. Thus we need a property that: 1. accepting references to dictionaries (not to articles itselfs) and accepting references to articles itselfs as qualifiers. Nobody prohibits anyone from creating specific items. Go on. But if one need those articles to be listed as "references to dictionaries" this item need to be present in described by source (P1343) and be specified as qualifier of value like Brockhaus and Efron Encyclopedic Dictionary (Q602358), Great Soviet Encyclopedia (1926–1947) (Q20078554), Bible Encyclopedia of Archimandrite Nicephorus (Q4086271), etc. See ru:s:Модуль:Другие источники for implementation example and ru:s:МЭСБЕ/Рерих for example (line "Другие источники: БСЭ1 : ЭСБЕ") -- Vlsergey (talk) 16:11, 12 June 2015 (UTC)

And no, you can't move dictionary inside item (you still better to copy it). Because this will lead to "cycle all items from property, access them, check what's inside, and throw away 90%". This will obviously bad pattern. -- Vlsergey (talk) 16:13, 12 June 2015 (UTC)
I wouldn't say so: WDQ should be able to confine a query to those items which are connected by published in (P1433) to a certain dictionary. On the other hand to give a reference with described by source (P1343) to a resource 10.000 times more coarse than the specific item and providing a redundant link via statement is subject of (P805) - how do you intend to enforce that (almost everyone will spot the redundancy in that construction and somehow try to "resolve" it)? I have been trying to understand statement is subject of (P805) from the few examples given and came to the following conclusion: Consulting the object behind BaEED / New-York (Q16947036) will positively answer the question, whether New York City (Q60) is indeed described by Brockhaus and Efron Encyclopedic Dictionary (Q602358). That's not wrong (but perhaps the conclusion is more implicit than the inventors of statement is subject of (P805) had in mind?) but an impressingly convoluted way to express some rather linear association "X is described by (Y part of Z)" (which IMHO implies "X is described by Z")... Its like describing a needle like this: "Throw it into a haystack and then there will be one object attracted by a magnet: that's a needle!". -- Gymel (talk) 16:54, 12 June 2015 (UTC)
@Gymel: it's not possible to use WDQ from LUA/templates. LUA/templates access to items must be straightforward. Filtering items (not claims) from LUA code is no-no. 2. "how do you intend to enforce that" -- well, that's a big problem at Wikidata, because a lot of people make changes that they think is good, but actually breaks things at other wikis. But regarding P1343 I saw no bad changes in my watchlist so far. ruwiki & ruwikisource users are making changes using special gadgets, thus they will not made a mistake. But other users are problems. -- Vlsergey (talk) 17:03, 12 June 2015 (UTC)

Should we use stated in (P248) as a qualifier?

Moved to this thread by Vlsergey. --Succu (talk) 21:03, 21 June 2015 (UTC)

I made this change at Charles Darwin (Q1035) which was reverted by Vlsergey. In my opionion references should be added to the reference section and not added as a qualifier. --Succu (talk) 20:34, 21 June 2015 (UTC)

Maybe statement is subject of (P805) should be used. Sjoerd de Bruin (talk) 20:36, 21 June 2015 (UTC)
I agree with succu that stated in (P248) is a source property and should only be used in references. The redundancy problem which arised above should be solved using programming code and not by introducing more and more redundancy. -- Bene* talk 20:44, 21 June 2015 (UTC)
It's really a bad idea to mix online and offline sources in one property, whatever the qualifier being used. --- Jura 05:03, 22 June 2015 (UTC)

Domain of described by source (P1343)

The english property description is: „The enumeration of sources where item is described. Dictionary, encyclopedia, etc. Qualificator stated in (P248) shall be used for links to articular article in Wikisource and reference URL (P854) for external reference. Recommended only for printed dictionaries and encyclopedias (including resent at Wikisource or their own website). For sites like Internet Movie Database (Q37312) there are additional properties like IMDb ID (P345)“.

It seems to me that the original description was Google translated and extended. We should clarify the domain of described by source (P1343) before we are talking about bot requests. --Succu (talk) 20:13, 24 June 2015 (UTC)

Redundancy

I feel that I need to clarify something. Redundancy. Redundancy itself is not bad. Redundancy is a property of data when we have some duplication. It has pro and contras:

  •   Oppose Redundancy obviously occupies more storage space. As we all know, this is not our problem at all (don't worry about tech. stuff until Maksem some IT guy knocks in your IM)
  •   Oppose Duplicating means we need to update 2 or more copies of data simultaneously, otherwise we have discrepancy. Thus we need additional bots and tools to check and resolve such discrepancies.
  •   Support Redundancy means faster access. We don't need to go to separate item if the same data already stored in current one. We don't need arbitrary access to another item, thus saving precious "one of 500 expensive parser function call". Redundancy means less complexity for client LUA modules and templates.
  •   Support Redundancy means data are more simple to observe, edit, check and make vandal-prone. Single incorrect change in "very main item" means we have one Wikipedia article broken and one new line in discrepancy report. Also for Wikipedia article editor it is much simple to edit one specified item to change incorrect data than to understand LUA/templates/Wikidata logic and find a place where he should actually fix data.

One shall never throw away something just because or redundancy. One shall always provide an argument that redundancy is not necessary in the particular case. In most cases there is a simple way to get the same data. More over the person who introduce redundancy need to explain, why it was introduced, why he does not see a simple way to do without it. But first of all everyone need to accept the possibility that sometimes such redundancy "pro" is more than "contra".

I'll give you single example where redundancy "pro" is much better than "contra", from my point of view. Birth place. In ruwiki a lot of articles have line in infobox with something like "City, Region, Country". One can say "it's very simple to use arbitrary access and get to the country from specific item". Well, actually, it's not so simple:

  • World changes. The countries centuries ago are not the same as today. Every country has "life span". We need to find a parent item applied to specified timespan.
  • Time span is defined by birth date, that can have 2 or more possible variants
  • Did I mention already that birth date can be something "at the end of XIV century", leading us to problems with "birthplace" LUA module?

Personally I believe that birthplace and deathplace redundancy should be preserved. At least until some very smart guy (hi Bene) write some very-very smart LUA module, some very handsome girl (hi, Lydia) write down detailed description for usual Wikipedia editor how to fix "this-city-was-not-part-of-something-at-this-date" problem (pages 50-100 I assume), and some very-very stupid project from TOP10 (not ruwiki!) will put it in action in their infoboxes.

Obviously redundancy is very bad for data that updates regularly. One shall never duplicate URLs, statistical data, etc. But let you mind not to be fixed on incorrect assumption that "redundancy is bad". -- Vlsergey (talk) 16:45, 12 June 2015 (UTC)

Thats way too abstract for me. We have redundancy by inverse properties, which in a sense may be the worst one because it's spread out to different items. Then we have redundancy of expression, some more complex statements can only be made by qualifying a claim with property P by something with property Q but the same can be expressed by qualifying a claim with property Q' by property P. We also have the case you may be alluding to, where parallel claims exist where one could be (sometimes unambigously) inferred from the other. But I have to admit the specific zig-zag redundancy you proposed above I've never encountered before... -- Gymel (talk) 17:05, 12 June 2015 (UTC)
To clarify, I the story started several days ago, when a Russian user complained at the Russian version of the Project Chat that they are being reverted by @Sjoerddebruin: (see Q191848 for example). The point was that the user tried to add a qualifier to the iem on a football player. The property was showing the city of birth, and the idea was to have a qualifier which would also indicate the country of birth. Sjoerd reverted saying that this info is redundant. The user has chosen to edit-war, and won in most of the articles. However, I agreed with Sjoerd, that the info is redundant. The only motivation of the user was that in Russian Wikipedia they read the country directly from Wikidata in the infobox, and since they are only have the Arbitrary access in a month, all info they need for the infobox NOW needs to be in the same item. I tried to explain that this is not a good approach, without much success, got attacked and decided to unwatch the page, so that I do not know what they decided there. I still oppose this approach, but I might be in the minority, I do not know.--Ymblanter (talk) 17:38, 12 June 2015 (UTC)
Note that arbitrary access for the Russian Wikipedia has been moved forward to the 23rd of June, see what Lydia said in #next_round_of_arbitrary_access_rollouts above. - Nikki (talk) 17:56, 12 June 2015 (UTC)
I know, but in any case I do not think that the absence of an arbitrary access at a project should be a reason for us to add information readily available elswere. At best this must be negociated with the Wikidata community.--Ymblanter (talk) 17:59, 12 June 2015 (UTC)
I didn't mean it as something for/against what you said, I just wanted to add an update for anyone reading this who wasn't aware that the date had changed. :) I do agree with you though, I don't see a good reason to duplicate the data just because arbitrary access isn't fully rolled out yet. - Nikki (talk) 18:15, 12 June 2015 (UTC)
I made some discoveries some days ago when I tried to edit 2014 European Parliament election in the United Kingdom (Q5412950). My approach to the problem here was that the sum of all UK-constituencies would describe the UK-election in the whole. I intended not tp add very much in the UK-item, but instead in the constituency-items. I did this to avoid Redundancy. But a BIG problem I discovered, was that the sum of information I had in sources for each constituency would not fit the sum of all information for the whole country. That one constituency (Northern Ireland) has a completly different voting system, was not the only problem. The sources for each constituency was not as detailed as the sources for the whole nation. -- Innocent bystander (talk) 18:28, 12 June 2015 (UTC)

Motî?

I'm working on reducing the number of items with no claims and while making a list of items without claims, but do have a ':' in their title (this list) I ran into quite a few on the Walloon Wikipedia that look like dictionary entries, some examples:

Any suggestions on what kind of claim(s) to add to these kind of items? Multichill (talk) 15:41, 13 June 2015 (UTC)

Add them instance of (P31) dictionary page in Wikipedia (Q20088085). See also w:wa:Special:PrefixIndex/Motî:. Visite fortuitement prolongée (talk) 20:21, 13 June 2015 (UTC)
That was my plan B :-)
Running now. Multichill (talk) 10:14, 14 June 2015 (UTC)

Kevin Bacon challenge

Hoi, I have a challenge for those people who know how to use Wikidata data... It is the Kevin Bacon six degrees of separation challenge.. Can this be done? My challenge would be to expand networks to Mr Bacon. Thanks, GerardM (talk) 19:36, 13 June 2015 (UTC)

Challenge accepted! I am currently running a program (code here) to find the shortest path from Kevin Bacon (Q3454165) to Genghis Khan (Q720) by brute horse, recording everything else it encounters. The other end point you suggested, Julius Caesar (Q1048), has a WDPBN (WikiData Psuedo Bacon Number) of at most 90, since July (Q121) is currently holding steady at 89 (Julius Caesar (Q1048) itself has not yet been added to the list, but probably will be soon).
I have no idea how long it will take, but Autolist says there's a familial connection of some sort between George Washington (Q23) and Genghis Khan (Q720) (and, for that matter, 32000 other people) (George Washington (Q23)'s number, incidentally, is 68, via head of government (P6) from United States of America (Q30)) Popcorndude (talk) 03:03, 14 June 2015 (UTC)
Due to an unfortunate error, my program was not actually recording the shortest paths (Bacon's own WDPBN is currently listed as 4). I shall now run the program again. Popcorndude (talk) 03:09, 14 June 2015 (UTC)
If anyone's interested:
Popcorndude (talk) 03:29, 14 June 2015 (UTC)

Thank you Popcorndude !!. I blogged about your effort the challenge is also to make it possible to use your routine with "live" Wikidata data. WDQ counts as such for me. Thanks, GerardM (talk) 06:46, 14 June 2015 (UTC)

I've been trying to make a live version, it's not working. The best I can do for now is say either download the code I linked to and run it yourself (it gets up to 6 degrees in about an hour or 2), or contact me and ask (I may be able to upload some of them in a useable manner somewhere). (By the way, every village in China probably has a WDPBN of 7) Popcorndude (talk) 21:29, 14 June 2015 (UTC)
In a few hours, when the program runs out of connections, I'll see about sharing them somehow (I think I have a way). Until then, here's one that made me laugh:
Popcorndude (talk) 23:28, 14 June 2015 (UTC)
For WDPBN: 1 -> 13 items, 2 -> 265 items, 3 -> 3889, 4 -> 24000, 5 -> 60969, 6 -> 166099, 7 -> 705344
That's 960580 items with WDPBN < 8
Here's all the items with a WDPBN of 2 not linked from United States of America (Q30) (uploading this in a meaningful way is quite impractical for any higher numbers, unfortunately). Popcorndude (talk) 14:23, 15 June 2015 (UTC)
This is almost certainly the shortest path, but I only checked familial properties after Elizabeth the Cuman (Q265467), so there might be sorter one, though it's rather unlikely.
An alternate path goes through only people starting from George Washington (Q23) (WDPBN 2), and has Elizabeth the Cuman (Q265467) as being 20 away from him.
Another random thing I found: Arnold Schwarzenegger (Q2685): Bacon Number: 2, WDPBN: 5, distance from George Washington (Q23) using only parent-child connections: 42, using parent, child, sibling, and spouse: 37
And I'm now reasonably confident that I will not be making a "live" version of this program anytime soon, though I am willing to answer requests for specific people's paths. Popcorndude (talk) 19:59, 15 June 2015 (UTC)
Mary of Hungary (Q614258) was missing links to her siblings. Popcorndude (talk) 20:17, 15 June 2015 (UTC)

Pie charts about people items

A few pies can be found at User:Jura1/People charts. --- Jura 09:00, 14 June 2015 (UTC)

Help

Hello,
I don't know how to link this page > nl:Giffard-Luchtschip and fr:Dirigeable Giffard. Anyway, it's the same article. Thanks Mike Coppolano (talk) 11:26, 14 June 2015 (UTC)

Mike Coppolano Thank you for the report. I merged the corresponding Wikidata items. -- Gymel (talk) 11:51, 14 June 2015 (UTC)

Wikidata weekly summary #162

New rfc: Help:Basic membership properties

Hi, the project guideline status of Help:BMP has been challenged, although its been a reference since the beginning. So I opened a RfC to adopt it formally.

@Andreasmperu, Snipre, Zolo, Emw, Succu, Izno:@Bene*, Kopiersperre:, ...

It's already open for comment at Wikidata:Requests for comment/Adopt Help:Classification as an official help page. --- Jura 17:47, 14 June 2015 (UTC)
@Jura1: No, it's another page ... TomT0m (talk) 17:49, 14 June 2015 (UTC)
The currently open RFC is at Wikidata:Requests for comment/Adopt Help:Classification as an official help page. Opening another one is just confusing. If you need help contributing, maybe there are specific problems you encounter when adding statements and you want us to look into. --- Jura 17:55, 14 June 2015 (UTC)
@Jura1: OK, I guess you have another idea then. What do you suggest ? TomT0m (talk) 18:00, 14 June 2015 (UTC)
When trying to add statements, what problems do you encounter and feel you need help? --- Jura 18:01, 14 June 2015 (UTC)
@Jura1: I'll think about it the next time I'll have a problem adding a statement and I'll ping you, I promise. TomT0m (talk) 18:05, 14 June 2015 (UTC)

Mediation needed with Andreasmperu

See for example this diff : https://www.wikidata.org/w/index.php?title=Q15709880&curid=17344953&diff=222473528&oldid=222473145 . The previous edit is a revert of mine with an explanation as a comment. Several of his edit popped up I wanted to talk about, pinged him several times today on item talk pages, absolutely no answer. Not easy to deal with ...

(Notifying Andreasmperu) TomT0m (talk) 18:41, 14 June 2015 (UTC)

PS: see Talk:Q24925‎‎ and Talk:Q2882221‎. TomT0m (talk) 18:51, 14 June 2015 (UTC)

Hi, TomT0m. Aren't you supposed to contact me first on my talk page if you have anything to discuss? I tend to check my watchlist often that my notifications when I am editing. I only received a ping for this talk page, but I did not understand what you were referring to (since no specific items or editions were mentioned). Also, I am a "she" not a "he" (not that it is such a big deal). Andreasm háblame / just talk to me 18:57, 14 June 2015 (UTC)
@Andreasmperu: Sorry, I'll know that, now, I did not know Andreas could be a shea :) Can I ask you why you added
⟨ radio program ⟩ part of (P361)   ⟨ radio program or tv program ⟩
but
⟨ radio program ⟩ subclass of (P279)   ⟨ work ⟩
 ? That's inconsistent. We just have to say
⟨ radio program ⟩ subclass of (P279)   ⟨ radio program or tv program ⟩
and
⟨ radio program or tv program ⟩ subclass of (P279)   ⟨ work ⟩
. part of (P361) is for objects that are made of other objects, like car/wheel, not per classes.
And why you did not try to know why I reverted you ? I put something in the revert message related to the previous question : Wikidata:Property proposal/Generic#subclass. I also don't understand why you did not answer and ignored me when you could have asked more details ... I still don't know what the plan is about genre, I don't know if that make sense if a genre can be a subclass of another one. I know works of a subgenre are subclass of works in the (super)genre. TomT0m (talk) 19:08, 14 June 2015 (UTC)

Specifying academic rank

I am wondering what is the best way to specify academic ranks. In Soviet Union, Russia and some other countries there is (or was) a parallel to academic degree system of ranking. For Soviet/Russian scientists this is important information included in ru.wikipedia in infoboxes. Namely, I am interesting in distinction between corresponding and full members of the Russian Academy of Sciences. As for now, membership itself is specifying by means of member of (P463) which seems to be reasonable. I also can use some qualifier (say, P794 (P794)) to specify type of the membership. But this way requires creation of two entries differ only by qualifier for people who first were corresponding members and then were elected to be full members (this is actually true for most of the full members). So, I am wondering if there is more elegant way to do it. Artem Korzhimanov (talk) 15:52, 15 June 2015 (UTC)

May be I do not get the point, but why can you make two qualifiers, specifying also the periods when they were corresponding and full member? Say a person X was a member, qualifier 1 corresponding qualifier 2 year1 qualifier 3 year2; qualifier 1 full member qualifier 2 year2 qualifier 3 year3? --Ymblanter (talk) 15:57, 15 June 2015 (UTC)
@Ymblanter: Can you demonstrate it, say, on Yury Osipov (Q85404)? The dates may be found here. Artem Korzhimanov (talk) 16:03, 15 June 2015 (UTC)
Pls have a look. (In his case, thighs are even more complicated, since he was both the member of USSR Academy of Sciences and of RAS, and also I was too lazy to add sources, but I guess the idea is clear).--Ymblanter (talk) 16:16, 15 June 2015 (UTC)
@Ymblanter: I see, thanks. As I said, what I don't like here is that there are now two claims of the form Yury Osipov (Q85404) member of (P463) Academy of Sciences of the USSR (Q2370801). It's not a big problem, of course, but I hoped there is a more elegant way. Artem Korzhimanov (talk) 16:25, 15 June 2015 (UTC)
Let us see, may be someone has a better solution.--Ymblanter (talk) 16:33, 15 June 2015 (UTC)
@Artem Korzhimanov: I would say that ranks are informations about a class of person. So I think a solution could be
⟨ theperson ⟩ instance of (P31)   ⟨ university teacher (Q1622272)      ⟩
, with
⟨ university teacher (Q1622272)      ⟩ instance of (P31)   ⟨ academic rank ⟩
 and
⟨ university teacher (Q1622272)      ⟩ subclass of (P279)   ⟨ academic ⟩
,
⟨ academic ⟩ subclass of (P279)   ⟨ person ⟩
(let's not be too much human (Q5) centric, maybe a descendant of common octopus (Q651361)      will go to the university some day. TomT0m (talk) 20:17, 15 June 2015 (UTC)
@TomT0m: I like the idea but isn't instance of (P31) too general for such statements? Say, for President of the United States (Q11696) we use more specific property position held (P39). Maybe we need special property "academic rank"? And is it ok to apply to instance of (P31) qualifiers like start time (P580) and end time (P582)? Artem Korzhimanov (talk) 20:45, 15 June 2015 (UTC)
@Artem Korzhimanov: This is not a popular position on this project, but instance of (P31) is inspired from the type property of RDF or other in OWL. This means classes can be as specific as we need them to be. See Help:Classification for a more detailed explanation and on how instance of (P31) could use other properties to define a class. There is a open RfC to adopt it as a guideline if you like :) For example I'd very well see a query about academics as human who teach or work in a university and so on. In my mind we could position instance of (P31) academic to an item if we know only this as an item, and a bot could infer there miss something, or add (or suggest) the missing statements if he can, beeing aware of a query associated to the class.
@Artem Korzhimanov: You can qualify the membership (member of (P463)), no problem. Check Richard Foerster (Q90899) for example. Jonathan Groß (talk) 20:57, 15 June 2015 (UTC)
@Jonathan Groß: The question is whether it's normal to have two claims with the same object but different qualifiers:
⟨ Yury Osipov (Q85404)      ⟩ member of (P463)   ⟨ Academy of Sciences of the USSR (Q2370801)      ⟩
P794 search ⟨ Q2625729 ⟩
⟨ Yury Osipov (Q85404)      ⟩ member of (P463)   ⟨ Academy of Sciences of the USSR (Q2370801)      ⟩
P794 search ⟨ Q414528 ⟩
I think it's not a big problem, of course, but it seems to be somewhat redundant. Artem Korzhimanov (talk) 06:50, 16 June 2015 (UTC)
It does. Couldn't you just use
⟨ Yury Osipov (Q85404)      ⟩ member of (P463)   ⟨ Academy of Sciences of the USSR (Q2370801)      ⟩
P794 search ⟨ Q2625729 ⟩
? What's the point of academician (Q414528)? Every member of an academy of sciences is an academician (Q414528), no matter if corresponding member, honorary member or what not. Jonathan Groß (talk) 07:01, 16 June 2015 (UTC)
@Jonathan Groß: No, I couldn't. As I said there are two types of membership in Russian Academy of Sciences: "correspondence member" and "full member". The last usually called "academician" in a narrow sense (and a correspondence member of RAS is almost never called "academician" in Russia). What is more important is that they are two qualitatively different ranks in Soviet/Russian academic system and must be distinguishable. Artem Korzhimanov (talk) 15:31, 16 June 2015 (UTC)
@Artem Korzhimanov: nothing prevents you to store the value as qualifier of member of (P463) claim, but change appropriate infobox code to display such claim+qualifier as
⟨ subject ⟩ academic degree (P512)   ⟨ academician (Q414528)      ⟩
or
⟨ subject ⟩ academic degree (P512)   ⟨ corresponding member (Q2625729)      ⟩
. -- Vlsergey (talk) 18:04, 16 June 2015 (UTC)
@Vlsergey: Haven't caught an idea, sorry. Can you demonstrate how to store the information about academic rank (not degree) with dates, say, on Gennady Mesyats (Q4291469). The dates may be found here. Artem Korzhimanov (talk) 08:43, 17 June 2015 (UTC)
@Artem Korzhimanov: something like this? -- Vlsergey (talk) 09:00, 17 June 2015 (UTC)
@Vlsergey: that's what I talked about in the first message and then was proposed by Ymblanter and Jonathan Groß. What I dislike here is the redundancy of two claims with the same value (see above an example with Yury Osipov (Q85404)). If we had property "academic rank", I could make two different claims (values below are not valid as they are instances of list articles, but it's not a problem: if needed, we can create two new items which will be instances of corresponding member (Q2625729) and academician (Q414528), another option — is to use corresponding member (Q2625729) and academician (Q414528) with qualifiers):
⟨ Yury Osipov (Q85404)      ⟩ academic rank search ⟨ 4156897 ⟩
⟨ Yury Osipov (Q85404)      ⟩ academic rank search ⟨ 4516936 ⟩
But we don't have such a property (academic degree (P512) is not suitable in my opinion as this is a different system of ranking) and the question is: Do we need it or we could do the same with already existent properties? Artem Korzhimanov (talk) 09:16, 17 June 2015 (UTC)
@Artem Korzhimanov: well, how about this then:
-- Vlsergey (talk) 09:38, 17 June 2015 (UTC)
@Vlsergey: Well, the thing is that full member of RAS (Q4156897) and full member of RAS (Q4156897) are formally and semantically not of type organization (Q43229). In other respect the idea has sense. Artem Korzhimanov (talk) 09:55, 17 June 2015 (UTC)

Merge problems

Can someone merge en:Category:Meat industry (Q8614887) with German de:Kategorie:Fleischherstellung (Q8954695) 92.76.115.221 18:31, 15 June 2015 (UTC)

These items each have a commons category, this prevents merging. Mbch331 (talk) 18:52, 15 June 2015 (UTC)
Removed one of the commons category and merged. MSGJ (talk) 08:41, 17 June 2015 (UTC)

Can someone merge en:Category:Pork dishes (Q7213601) with German de:Kategorie:Fleischgericht (Schwein) (Q19474558) 92.76.115.221 18:37, 15 June 2015 (UTC)

  Done - Mbch331 (talk) 18:52, 15 June 2015 (UTC)

Can someone merge en:Category:Lamb dishes (Q5885348) with German de:Kategorie:Fleischgericht (Schaf) (Q19474550) 92.76.115.221 18:42, 15 June 2015 (UTC)

  Done - Mbch331 (talk) 18:52, 15 June 2015 (UTC)

Can someone merge en:Category:Beef dishes (Q7213600) with German de:Kategorie:Fleischgericht (Rind) (Q19474543) 92.76.115.221 18:47, 15 June 2015 (UTC)

  Done - Mbch331 (talk) 18:52, 15 June 2015 (UTC)

Can someone merge en:Category:Sausage dishes (Q16811235) with German de:Kategorie:Wurstgericht (Q19616978) 92.76.115.221 18:58, 15 June 2015 (UTC)

  Done Lymantria (talk) 07:40, 16 June 2015 (UTC)

Can someone merge en:Category:Maize dishes (Q8596746) with German de:Kategorie:Maisgericht (Q19481594) 92.76.124.0 19:37, 15 June 2015 (UTC)

  Done Lymantria (talk) 07:40, 16 June 2015 (UTC)

Absent in a source, used as a source?

Would you consider the lack of being mentioned in a source as a source in itself?

An example: Hallinge (Q17048639) is present in the report: Småorter 1990 (Q20087097) describing the situation 1990. But she is missing in the report about 1995: Småorter 1995 (Q20087135). I consider that as a proof that the locality has lost its status as a minor locality in Sweden (Q14839548) 1995, but would you accept that as a reference for "end date 1995-12-30" for P31:Q14839548? Statistics Sweden is the only authority in this case and it is in these reports they publish things like this. Corrections of these reports are sometimes published, but I fail to find any in this particular case. -- Innocent bystander (talk) 18:38, 15 June 2015 (UTC)

I would say have a more vague end date, maybe just "1995", since I don't know how you might specify "somewhere between 1990 and 1995". Popcorndude (talk) 20:48, 15 June 2015 (UTC)

ATC code J01(Q434066)

On the page above click Wikipedia → en. The list of languages is empty. Other languages are OK. Gyimhu (talk) 20:05, 15 June 2015 (UTC)

Revert analysis

Hey. I analyzed all 2073 reverts of IP edits during the last 30 days and identified the country of origin of the IP's. In the graph below all countries with more than ten reverts are listed.

Remarkably, seven out of the ten top countries are Spanish speaking countries. In total, IP's from these seven countries are responsible for 1117 reverted edits or 55% of all reverted edits during the studied period. I hypothesize that many IP's are coming from Spanish Wikipedia articles with a [editar datos en Wikidata] link as in es:Jesé Rodríguez. Such links are one hand very welcome as Wikipedia authors can faster improve the connected Wikidata item, on the other hand they open a new playground for vandals. I checked if the vandalizing IP addresses from Spanish speaking countries are also active on Spanish Wikipedia. Though only 14% of these IP's have also reverted edits on Spanish Wikipedia. This means that vandals on Wikidata and Spanish Wikipedia are different people but most probably many Wikidata vandals are coming from Spanish Wikipedia --Pasleim (talk) 20:26, 15 June 2015 (UTC)

Maybe we could attempt to display the (additional) edit link on eswiki only to users who are logged-in?
If there is a technical way to do this, please suggest it at es:Módulo_discusión:Ficha#.22editar_datos_en_Wikidata.22. --- Jura 20:36, 15 June 2015 (UTC)
Looking at all IP edits from the same region(s), what is the proportion of edits that have been reverted? --- Jura 21:40, 15 June 2015 (UTC)
I created a second plot showing the number of reverts by country relative to the total number of IP edits made in the country. Only countries with more than 100 edits are shown. In average, 7% of all IP edits get reverted. --Pasleim (talk) 20:08, 16 June 2015 (UTC)
25% seems fairly high. Is it? --- Jura 06:42, 17 June 2015 (UTC)
Are they editing Spanish-speaking content? From what I have heard, South America is one of the main hosts for Zombies. -- Innocent bystander (talk) 07:34, 16 June 2015 (UTC)
Personally, I came across vandalism on football players and politicians. --- Jura 13:50, 16 June 2015 (UTC)
@Pasleim: My hypothesis is same. Compare with this discussion in February.
In cswiki, infoboxes using data from Wikidata show similar text but it says "some data may come from Wikidata" instead of "edit the data in Wikidata". Matěj Suchánek (talk) 14:31, 16 June 2015 (UTC)
Interestingly, eswiki still displays the "editar" link on semi-protected pages. So items like Q165125 get edited instead. --- Jura 06:42, 17 June 2015 (UTC)

Great work, Pasleim! I see that 'Editar datos en Wikidata' was added on 8 feb 2014. Can we see a growth of IP edits to wikidata around this time? Or a growth of spanish descriptions after 8 feb 2014? --Atlasowa (talk) 07:52, 17 June 2015 (UTC)

This is awesome: Wikidata:Database_reports/Number_of_edits! (It doesn't show a spanish IP pattern, though). Fantastic graph, explains so much about wikidata (widar et al.) --Atlasowa (talk) 13:39, 17 June 2015 (UTC)
Hm, it might suggest that IP edits haven't increased, just vandalism. --- Jura 13:47, 17 June 2015 (UTC)
Chart at User:Jura1/test3. --- Jura 14:02, 17 June 2015 (UTC)

Looks like similar "edit infobox at wikidata" notes were added in french WP (march 2015, see fr:The_Octagon_House), inspired by hungarian WP infoboxes with « +/- » next to infobox-data from wikidata (see hu:Mohács). --Atlasowa (talk) 13:28, 18 June 2015 (UTC)

"infobox monument" is likely to draw a different public. What type of vandalism/IP edit have you observed? --- Jura 13:42, 18 June 2015 (UTC)
I think if the spanish infobox-wikidata-link is indeed responsible for more vandalism, then we should compare different models of infobox-wikidata-links in different WP and how they affect Wikidata. Especially since it looks like enWP will copy the spanish infobox-wikidata-links despite the revert analysis: en:Wikipedia_talk:Wikidata#Edit_in_Wikidata_links! --Atlasowa (talk) 22:42, 21 June 2015 (UTC)
Protecting the page locally and adding more links to edit Wikidata is just asking for trouble. Other than that, it might be preferable to attempt to convince logged-in users to edit Wikidata rather than IPs. (Not that I think IP should be prevent from contributing in general).
Personally, I like the edit links leading to the actual source for specific infobox fields. --- Jura 04:40, 22 June 2015 (UTC)
  • Towards Vandalism Detection in Knowledge Bases: Corpus Construction and Analysis SIGIR’15, August 09 - 13, 2015, Santiago, Chile. http://dx.doi.org/10.1145/2766462.2767804 : "Our corpus is based on a database dump of the full revision history of Wikidata until November 7, 2014 (...) about 85% of revisions are made automatically by bots approved by Wikidata’s community. (...) As we are interested in detecting ill-intentioned contributions by humans, and not errors in bots, we base our corpus on the 24 million manual revisions."
    • Revisions made on Wikidata 167,802,227 100 %
    • Revisions made on meta pages 1,211,628 1 %
    • Revisions made on special items 11,167 0 %
    • Revisions made automatically 142,574,999 85 %
    • Revisions made manually 24,004,433 14 %
  • "Of the 24 million manual revisions made on Wikidata, a total of 103,205 have been reverted via rollbacks, and 64,820 via undo/restore. Based on our below validity analysis, we label roll-back revisions as vandalism, whereas this cannot be done with confidence for undo/restore revisions."
  • (Figure 2: Manual revisions on Wikidata per month. Revisions affecting textual content (labels, descriptions, and aliases) are distinguished from revisions affecting structural content (statements and sitelinks). Major growth events are labeled.) "The first jump of growth rate was caused by enabling statement creation for first time. In the months around this event, Wikidata was connected to the Wikipedias in various languages, adding millions of statements and sitelinks. (...) The second growth rate increase is due to the emergence of semi-automatic editing tools for Wikidata, most notably the Wikidata Game.
  • Vandalized Item Categories: "Table 1: Top vandalized items Cristiano Ronaldo, Lionel Messi, One Direction, Portal:Featured content, Justin Bieber, Barack Obama, English Wikipedia, Selena Gomez (...) the least vandalized category Places gets almost 4 times as much attention by all editors (31%) (...) The focus of vandals deviates significantly from typical editors (...) while categorizing the revision samples, we noticed that 11% of the vandalized items concerned India, cross-cutting all categories, compared to 0.5% overall."
  • Vandalized Content Types: "About 57% of the vandalism happens in textual content like labels, descriptions, and aliases; and about 40% happens in structural content like statements and sitelinks. The remaining 2% of miscellaneous vandalism includes merging of items and indecisive cases."
  • Vandals: "About 86% (88,592 of 103,205) of vandalism on Wikidata originates from anonymous users. (...) Unregistered users primarily vandalize textual content and sitelinks, whereas registered users primarily vandalize statements and sitelinks."

Interesting analysis. --Atlasowa (talk) 07:55, 30 June 2015 (UTC)

Seems that I had never fixed any vandalism as defined by the study as I hadn't used the roll-back function until recently. I'm curious about the response to Lydia's question to the authors. --- Jura 10:13, 30 June 2015 (UTC)
Jura, you mean "Lydia is in touch with them to figure out how to best use the insights for us." (Wikidata:Status_updates/2015_06_27#Other_Noteworthy_Stuff) or are there more specific questions? Things i noticed:
1) The study doesn't analyse Wikidata vandalism in general, but only specific detected vandalism. Maybe 50% of wikidata vandalism is getting fixed, or 5% or 90%, we don't know? It's problematic to generalise from the analysed detected vandalism to all vandalism.
2) Did the study take into account Special:AbuseFilter? Which puts up some edits for screening if certain conditions are met. Or did the study not know this, and will in circular fashion advise wikidata to built the abuse filters that we already have?
3) The study didn't analyse IP origins like User:Pasleim, it seems. Great to have a corpus of wikidata vandalism edits (2012 until November 2014) then for further analysis! :-) --Atlasowa (talk) 11:44, 30 June 2015 (UTC)
Thanks. I'll add those to the questions I'll ask them. Currently still trying to schedule a call. If you have more questions for them let me know. --Lydia Pintscher (WMDE) (talk) 17:17, 30 June 2015 (UTC)

Mapping between Freebase and Wikidata

Hey everyone,

One of the big challenge of the importation of a part of Freebase into Wikidata project is to do a good mapping of the properties of Freebase to Wikidata ones in order to be able to feed the Primary Source tool with as much good data as possible. As there are a few thousand of properties it is a task too big to be done alone.

I have created a page to work on it and your help is more than welcome: Wikidata:WikiProject Freebase/Mapping. Tpt (talk) 22:11, 15 June 2015 (UTC)

Dissambiguation + Surname

Hi! Should we (majestic plural) split items like this (Madariaga (Q342908)) into two independent Qs sort of ("disambiguation" and "surname")? Or happy together? Strakhov (talk) 14:20, 16 June 2015 (UTC)

I've been splitting them, since family name (P734) has Conflicts with instance of (P31): Wikimedia disambiguation page (Q4167410), but I've really only done it twice. Popcorndude (talk) 14:49, 16 June 2015 (UTC)
I have splitted as usual (Q342908 + Q20119834). --ValterVB (talk) 18:26, 16 June 2015 (UTC)
Thanks to you both, next time I'll do the same way! Strakhov (talk) 18:44, 16 June 2015 (UTC)
But together was happier :(( --Infovarius (talk) 18:57, 18 June 2015 (UTC)

North America

I keep trying to split North America (Q49) into two pages because in some languages "North America" means everything north of Panama, but in other languages it only means Mexico, United States, Canada, and Greenland. These two versions of the continent have different populations, different sizes, and different images, so they need different pages. However, others keep reverting my split. May I please get a consensus to keep them split? --Arctic.gnome (talk) 15:20, 16 June 2015 (UTC)

I think it would be easier if we had a solution related to interwiki so to avoid this, comment here : #Bonny_and_Clyde_.3B_arbitrary_access. Moreother, I have a potential solution : create a meta item, like North america definition, a metaclass in the sense of Help:Classification, with
⟨ North America (Q49)      ⟩ instance of (P31)   ⟨ North America definition ⟩
to serve as a bridge into the two items. We could also say something like
⟨ North America definition ⟩ instance of (P31)   ⟨ metaclass of Conflicting definition ⟩
, and put a warning in the merge item that two instances of the same conficting definition class should NOT be merged. TomT0m (talk) 15:28, 16 June 2015 (UTC)
@Arctic.gnome: added your problem in WikiProject Cross Items Interwikis. TomT0m (talk) 16:19, 16 June 2015 (UTC)
We already have a perfectly good solution for handling this type of issue and it is the solution that Arctic.gnome proposed - Create two separate items - "North America (including Central America and the Caribbean)" and "North America (excluding Central America and the Caribbean)". These are clearly two different concepts. Weird meta-classes to resolve word definitions belong in Wiktionary, not in Wikidata. Wikidata is concerned with the definition of concepts - it is not concerned with the definition of words and phrases. Filceolaire (talk) 11:50, 18 June 2015 (UTC)
@Filceolaire: Yeah, that's what I propose also. But as competing definitions for very close concepts is a pretty common case. Some metaclasses regrouping those definitions could prove useful, if for no other reason to link them together. I don't think wiktionary can really do that as it's different terms in different languages that will have little to do with each other and won't translate from one to the other. TomT0m (talk) 16:12, 18 June 2015 (UTC)
    • Although I think they need to be two items for the purpose of logging population and area, it does create the odd situation where each project's article about North America only has inter-language links to half of the other projects (i.e., the projects which use the same definition). --Arctic.gnome (talk) 16:54, 18 June 2015 (UTC)
  •   Support keeping them split. We're probably going to have a lot of situations like this, since language's distinctions tend to be very different from each other. The current North America (including Panama etc.) should, in my opinion, just have "North America" as the English label, since that's what North America means in English. The other, I'm not sure. We already have a Northern America (Q2017699); maybe Mexico plus it should be labeled "Northern America and Mexico"? --Yair rand (talk) 12:22, 18 June 2015 (UTC)
    Well, the Swedish article sv:Nordamerika does not tell any definite definition of the word. It tells in the beginning of the article that Mexico and Central America is included in various degrees depending on how/when the word is used. One definition only includes USA, Canada, Belize, Guyana and the English-speaking parts of West Indies (Q669037). If Greenland is included in that case is not obvious. But I think Greenland sometimes is regarded as a part of Europe, and not as a part of North America. The definition of Europe also have the same problem in the other directions. Are the cities of Vladivostok, Ankara, Ceuta, Melilla and Las Palmas located in Europe or in Asia or Africa? -- Innocent bystander (talk) 16:40, 18 June 2015 (UTC)
    • There are a few languages that discuss both definitions. Those languages will have to decide which Wikidata item to link to. If they want an infobox with the population and area, then use the Wikidata page with the numbers they want. If they want to discuss both definitions and give both equal weight, maybe they should create a new Wikidata item that is an instance of "term". --Arctic.gnome (talk) 16:54, 18 June 2015 (UTC)
      • In my opinion, if an article discusses both then the article is about the larger concept (including Central America) and includes a discussion of various parts of that concept. When we have arbitrary access (being rolled out now) it should be possible for an article to have multiple infoboxes, each linked to a different wikidata item (at least that is my understanding). Filceolaire (talk) 16:59, 18 June 2015 (UTC)
        • @Filceolaire: That's actually one example of where the multiple definitions item could prove useful. If it's unclear which is the main definition of the article, then the article talks about all the alternatives, then it become a viable option to link it to the meta item. Then when we know all of their instances, we can generate automatically one infobox for each of the definitions (the definitions which are instances of the metaclass). TomT0m (talk) 17:10, 18 June 2015 (UTC)
          • Multiple infoboxes in one article is not popular where I come from. :) I guess this often is a typical subject, when you need to discuss why or why not some nations/dependant areas are included in the term or not. It's not a typical infobox-subject. The only nation that is included in all definitions in the Swedish article is U.S (ex Hawaii). -- Innocent bystander (talk) 19:10, 18 June 2015 (UTC)

Creation of the Cross Item Interwiki Project

\o_ I think the creation of WD:XLINK would deserve a broad announcement as it's a problem global to kind of everybody.

How could we do that ? A global message, a personalized message from Wikidatas (with a checklist of majors Wikipedia) on bistros or village pumps, and so on ?  – The preceding unsigned comment was added by TomT0m (talk • contribs).

  • There are a lot of "global" questions. Almost any Wikidata question is "global". But I doubt every Wikidata question need to be distributed via global message. At ruwiki we have distinguish "village pump" for Wikidata (ex-interwiki) questions. Some of projects may have too. We should collect such pages and create wikidata-specific distribution list. -- Vlsergey (talk) 18:08, 16 June 2015 (UTC)
    • That's not hard, it's available with a link on this page : https://www.wikidata.org/wiki/Q16503#sitelinks-wikipedia But that's roughly what a global message does :) What's interesting with a global message is that we could initiate a global traduction before sending it, like wikitech ambassadors does , and that it would be translated in languages for which we do not have a usual Wikidatan involved. What's interesting with the per chat solution is the social personalized communication for those who don't leave their native projects easily. TomT0m (talk) 18:26, 16 June 2015 (UTC)

Item confusion with regard to United States

Item Q16071431 seems to be redundant of United States of America (Q30). I recommend the two be merged in some way. I also noticed that some of the Interwiki links on the United States of America (Q30) item page also seem to refer to categories while others do not and someone who can speak those languages and are more experienced in Wikidata may want to try and sort that out. I wasn't sure if this was a for deletion type issue so I thought I would drop the note here and see what folks say about it. Let me know if you think I should submit it to the for discussion/deletion board. Additionally, I notice that there are duplicate categories for Category:United States (Q1410960) and Category:United States (Q8879680) which both seem to mostly refer to the same things. Reguyla (talk) 19:37, 16 June 2015 (UTC)

Q16071431 is a zhwiki Wikimedia disambiguation page (Q4167410), and should not be merged. I have merged the categories. Popcorndude (talk) 19:59, 16 June 2015 (UTC)
Should it be merged with United States (Q232865), or does zh.wiki recognize a difference between "United States (disambiguation)" and "The United States (disambiguation)"? --Arctic.gnome (talk) 20:06, 16 June 2015 (UTC)
I would say yes, merge it. (though note that I am working off of a google translate of a snippet, you should probably check it yourself to be sure) Popcorndude (talk) 20:23, 16 June 2015 (UTC)
Thanks for the help, I'm still new to Wikidata and learning how to do stuff. This particular knot was a little more than I wanted to try and unravel without consultation. Reguyla (talk) 20:47, 16 June 2015 (UTC)
I have now merged them. MSGJ (talk) 08:29, 17 June 2015 (UTC)
There are multiple "United States". The USA is just the most well known. GerardM (talk) 15:39, 18 June 2015 (UTC)

Can't merge

As often happens here (a clue that the system is not properly designed), I found two same entries: Q4074925 and Q16186578.

I emptied Q16186578.

Now, I want to redirect Q16186578 to Q4074925.

Special:RedirectEntity tells me that Q16186578 isn't empty. Looks empty to me. What else do I need to delete (and how)?  – The preceding unsigned comment was added by Knife-in-the-drawer (talk • contribs).

@Knife-in-the-drawer: I don't know anything about using Special:RedirectEntity directly, but I would recommend reading Help:Merge#Automatic_merge - there are easier ways to merge things. :) - Nikki (talk) 03:41, 17 June 2015 (UTC)
This edit shows it indeed wasn't empty. There were still some labels present. If you activate the merge gadget in your preferences and use that to merge, you don't have to worry about an item being empty or not. Mbch331 (talk) 07:33, 17 June 2015 (UTC)
Note you can also use Special:MergeItems. -- Bene* talk 07:44, 17 June 2015 (UTC)

Thanks for the responses. @Mbch331: thanks for the link to the differences. Labels for nl, it, and fr. However, when I click on "Revision as of 03:17, 17 June 2015" on the Difference page to see the version that I saved, the "labels" table (for lack of a better name) shows English, Czech, German, Polish (with no labels defined).

Where are nl, it, and fr hiding?

I never configured the "labels" table to show only English, Czech, German, Polish. Interestingly, though, those four languages represent my native language and geographic location. Did WikiData automatically configure those languages as an "aid" for me? If yes, it was clearly a hindrance, not an aid.

Just like all of the statements in a WikiData entry, the "labels" table should show only currently defined labels. If a label isn't defined for English, as an example, then don't show English. If I'm an English speaker and I want to add an English label to an entry, then I should click an [add] link...just like all of the statements. If labels for Dutch, Italian, and French are defined, then they should always appear in the "labels" table...just like all of the statements. Knife-in-the-drawer (talk) 02:48, 19 June 2015 (UTC)

@Knife-in-the-drawer: Although this is not the case for you, since you have no user page, if a user has one and uses the Babel extension, then he will see those languages as suggestions. Alternately, you can edit Special:Preferences. —Justin (koavf)TCM 02:58, 19 June 2015 (UTC)
@Knife-in-the-drawer:. It shouldn't show labels used, it should show labels the user could understand. If they are empty the user is triggered to fill them. If it would show me labels that are filled in eg. Chinese and Russian, I wouldn't understand it and if it didn't show empty labels of languages I know I wouldn't be triggered to fill them. In case of creating a redirect, it can be a bit difficult if you see empty labels and don't see filled label, but the user interface is designed for showing and adding information. Merging and redirecting is something different and not the default actions you'd expect, so the UI isn't designed for that. Mbch331 (talk) 07:09, 19 June 2015 (UTC)

Meridiophila same as Loxostege

Meridiophila appears to be a synonym or alias for Loxostege and most Wikipedia language instances redirect one to the other. However, one does not. As a result, merging the Wikidata items for Meridophila and Loxostege is not immediately straight forward. Can someone confirm they are indeed the same? If so, can someone in the vi language code merge the two articles on that Wikipedia installation to allow the Wikidata entries to be merged?  – The preceding unsigned comment was added by Brantgurga (talk • contribs).

Vi is a common offender for biology articles. I usually take a stab at merging them myself with the help of Google Translate (using translate only to identify the text to-be-merged, not to write new text). You can always try slapping a Template:Merge (Q6919004) on the pages, but who knows how long that will take? --Izno (talk) 00:15, 18 June 2015 (UTC)

Option to hide all statements

I think it would be useful to add an option to collapse all the statements in so you could just focus on the links if you wanted. Some items have a lot of statements and it takes a while to scroll down to find the links at the bottom. FallingGravity (talk) 17:02, 17 June 2015 (UTC)

@FallingGravity: You might find the "KeyShortcuts" gadget in the preferences useful, that adds a keyboard shortcut "i" which jumps to the sitelinks section. - Nikki (talk) 17:44, 17 June 2015 (UTC)
That's a nice gadget, but the change I'm thinking of would make the links more accessible to new users and IPs. FallingGravity (talk) 18:18, 17 June 2015 (UTC)
You should also be aware of that the statements can be usefull to see the differences between two items. That is maybe somtimes the only way to see why there is two items with the same label, who should not be merged. -- Innocent bystander (talk) 18:30, 17 June 2015 (UTC)
Uh, I'm thinking the statements would be open (un-collapsed) by default or when someone opens the page. The option to collapse the statements and easily get to the links would then the choice of the user. FallingGravity (talk) 20:02, 17 June 2015 (UTC)
The interface redesign will contain this, they are working on this. Sjoerd de Bruin (talk) 21:36, 17 June 2015 (UTC)

LCAuth, LCCN and LCOC LCCN

Here is another strange property knot.

LCAuth identifier(P244) shows aliasas that include LCCN. When you add LCCN it pops up as LCAuth identifier and also gives the option for LCOC LCCN (bibliographic)(P1144) which also references several of the same aliases as the LCAuth identifier did. So I think someone might want to take a look at these and clarify them or merge them if needed. I don't know how to do that and I don't know if these are the same field or if they are actually different somehow.

Additionally, if you search for LCCN you get Library of Congress Control Number(Q620946). This one does not include any of the aliases from the other two. Now the first two are obviously properties and the latter is the Wikibase item about the thing, but should it also include the previously mentioned aliases? Reguyla (talk) 21:09, 17 June 2015 (UTC)

Those aren't errors and merger would be very wrong. The Library of Congress simply has multiple identifier systems:
All different systems, with different numbers, used in different ways. - AdamBMorgan (talk) 12:34, 18 June 2015 (UTC)

Unit-based properties - a work-around

I'm about to propose a number of properties for chemicals (and later other things), to facilitate the impending Freebase import.

A number of these might use unit-based measurements, such as a property for molar mass (Q145623). We await data types for unit-based measurements, but a work-around would be to define properties such as "molar mass in g/mol", whose value would simply be a number.

Similarly, we could have "length in metres", "speed in kmh", etc.

We would have the option at a later date to employ a bot to convert statements using such properties, to unit-based equivalents.

What does the community feel, about using this method?

Courtesy @Lydia Pintscher (WMDE): for info. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:59, 18 June 2015 (UTC)

I prefer to avoid this kind of manipulation which just add more work later to purge the database and all information documentation about these temporary properties. We were waiting until now for the numeric datatype with unit and I find stupid to create the temporary properties just for 6 months (I assume to have the new datatype for the end of the year). Snipre (talk) 15:57, 18 June 2015 (UTC)
+1 to Snipre. -- Vlsergey (talk) 12:43, 19 June 2015 (UTC)
see Wikidata:Project_chat/Archive/2015/05#Open_for_SI-unit_properties --Pasleim (talk) 12:49, 19 June 2015 (UTC)
I can see the attraction to just get on with it and do something but for me personally I have lots of other things to be getting on with so I'm not interested in getting involved with stuff that will be abortive. Filceolaire (talk) 23:16, 19 June 2015 (UTC)
I think this is a good idea and I'm a little confused about what @Snipre: means with how these will be "temporary"... Are you proposing that the mass of an electron will change in six months? —Justin (koavf)TCM 23:51, 19 June 2015 (UTC)
@@koavf: The properties would be temporary. They'd propagate, and then we'd have to unpropagate them after we get numbers with units. --Izno (talk) 01:01, 20 June 2015 (UTC)
Yes. I'm unclear as to why this is seen as a problem. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:17, 20 June 2015 (UTC)
I think it's a good idea in principle, but given the massive amount of data that might come from Freebase, doing it for anything but Freebase is preferable. Besides that, even the number datatype needs improvement before being used more widely. --- Jura 14:38, 20 June 2015 (UTC)
It makes it complicated to convert the statements when the right datatype finally comes. For example, if you today add the distance to The Restaurant at the End of the Universe (Q578895) in a property with "length in meters", when the original source tells the "length in miles", it becomes difficult for the bots to know what the original source told. The risk is that that statements stays with "length in meters" even if the source never told that. -- Innocent bystander (talk) 15:11, 20 June 2015 (UTC)
I think that's a hypothetical problem, and no more likely to occur than a bot entering the wrong unit when we do have unit-based data types. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:18, 20 June 2015 (UTC)
@Pigsonthewing: I have a lot of data from Statistics Sweden, with the "area in hectare (Q35852)". If I add those numbers here into a property for "area in km2", can you guarantee that the robots who later converts the data to the new property will automaticly do that correctly? Who else than Statistics Sweden uses ha as unit? -- Innocent bystander (talk) 17:08, 20 June 2015 (UTC)
Still a hypothetical problem, and again no more likely to occur than you entering the wrong unit when we do have unit-based data types. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:29, 20 June 2015 (UTC)
It's precisely because there is a vast amount of data available that I proposed this work-around. Please explain why the existing numerical datatype is a problem. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:21, 20 June 2015 (UTC)
Vast amount of data means vast amount of rework. If it would just help people start building better infoboxes, this would be another question. Anyways, I suggest you read Pasleim's comment. --- Jura 17:48, 20 June 2015 (UTC)
Reworking that is easily done by a bot I have read Pasleim's comment. And this has nothing to do with infoboxes. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:08, 20 June 2015 (UTC)
Reject proposal per Jura1. I think it's trivial to point out that we must reflect our sources. Pigs' assertion that it's a "hypothetical problem" is clearly wrong.

I think the issue of converting properties after the provision of the number with unit property is a problem also, but of lesser importance in this case. --Izno (talk) 00:22, 21 June 2015 (UTC)

Not only is the problem hypothetical. but rejecting it on the basis that "we must reflect our sources" is a red herring. When we have sources that use "molar mass in g/mol", "length in metres", etc, then having matching data type does reflect them. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:34, 21 June 2015 (UTC)
So, your opinion is that we should create one property for every possible unit, "lenght i meter", "length in kilometer/centimeter/Ångström/astronomic units/lightyears/parsec/English miles/Swedish miles/old Swedish miles/Värmland miles/Uppland miles/Småland miles" etc? That would solve the problems you describe as "hypotetical", but it would most likely not solve every issue, since the number of potential usefull units is very large. -- Innocent bystander (talk) 11:04, 21 June 2015 (UTC)
No. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:03, 22 June 2015 (UTC)

Labs outage

Due to a labs outage, some scripts and tools are not working at the moment. There is no ETA for a fix. Sjoerd de Bruin (talk) 14:36, 18 June 2015 (UTC)

Links between a cycling team and its 2015 season

Hi everybody. Today I come to ask for advice on how to connect the element of a cycling team 2015 season with the element of this cycling team, in order to be fixed on the best way forward.

Jérémy-Günther-Heinz Jähnick (talk) 14:42, 18 June 2015 (UTC)

I think property part of (P361) can be used to link the item for the 2015 squad to the item for the team. Filceolaire (talk) 23:12, 19 June 2015 (UTC)

Massive merge suggestion or duplicate

I am trying to merge Gauss-Jordan elimination into Gaussian elimination (though I think the name should indeed be Gauß-Jordan elimination) but I face a handful of sitelinks in Gauss-Jordan elimination that are either to stub or to articles addressing only the matricial version of the algorithm (the other article addressing in this case the system version). I see why these languages did so, but it seem to be an unnecessary split (as e.g. the fr and en version managed to do well in a single single article) and it gives the illusion that there are only a few languages with articles on this item.

As is, I see to solutions:

  1. Send merges suggestions to the offending languages
  2. Add the sitelinks from Gaussian elimination for the languages were there a single article to Gauss-Jordan elimination

But I don't know if 2 could (or should for that matter) be done, and if there is an usual way to do 1. Suggestions?

Evpok (talk) 15:00, 18 June 2015 (UTC)

One of the items you linked was "Darius" ([[Q265867), maybe you are looking for Gaussian elimination (Q2658).
In obvious cases, it's generally worth tagging the articles for merging and using Wikimedia duplicated page (Q17362920) on one item here.
It doesn't look like that to me, but I couldn't give a suggestion on how to interlink them either.--- Jura 15:06, 18 June 2015 (UTC)
  • @Evpok:. These are different methods and they are unlikely will be merged. You can learn about it in ruwiki articles. --Infovarius (talk) 19:15, 18 June 2015 (UTC)
In general, it is not a good idea to try and make decisions for a Wikipedia. There are Wikipedia's that are mad for merging (enwiki has merged many pages for topics that each deserve a page), and Wikipedia's that are trying to get as many pages as possible, no matter if there is substance (like svwiki). Pragmatically, if different concepts are involved, each should have a Wikidata page of its own; if it is the same concept use "duplicate page". - Brya (talk) 04:29, 19 June 2015 (UTC)
Projects that today are hosts to more than more one language, cannot be merged until all languages gets a project of their own. -- Innocent bystander (talk) 09:06, 20 June 2015 (UTC)

Q20155475 dates of establishment of congregation vs. building first erected vs. current structure erected

Hi, Emanuel African Methodist Episcopal Church (Q20155475) has more than one relevant date. Should we split the building and congregation to separate items? edits welcome. --Jeremyb (talk) 15:17, 18 June 2015 (UTC)

Building and organization have to be separated. The question is to know if the organization is really a independent organization or just a subdivision of the Methodist church. I don't think that all administrative levels of a church are relevant to have an item. I am not a specialist of the protestant churches but if I compare to the catholic structure, I will say this is a parish of the Methodist church. Snipre (talk) 16:06, 18 June 2015 (UTC)
Well I barely know anything at all about any sort of Christianity. cc Danrok because he's also edited the item. --Jeremyb (talk) 20:43, 18 June 2015 (UTC)
A bit problematic! The word church has at least two meanings, one is the building, and the other is the organisation. A similar situation exists with things like libraries and museums. All of these tend to be lumped together in a single Wikipedia article (building and organisation). They should probably be split into multiple items, along the same lines as companies, e.g. 8 Canada Square (Q572887) is the building, HSBC (Q190464) is the occupant and possibly the owner as well. In the case of business organisations and their buildings, these tend to already be split on Wikipedia, so not much of a problem with those. Danrok (talk) 22:08, 18 June 2015 (UTC)
It is maybe so in English, but it does not have to be like that in other languages and understandings of the word. I am Christian, but neiter Lutheran or Catholic, so I maybe see it differently from other persons. I consider "kyrka" normally either to be a building specially designed for some purpose, or the community between Jesus and his followers. For the organisation, I use the word "församling". If the building does not have a tower, icons, altar, painted glass or anything you typically consider as church-like, I do not use the word "kyrka". Instead, I then use "församlingslokal". -- Innocent bystander (talk) 06:34, 20 June 2015 (UTC)
If you look at the article then it is clear that this congregation is a defined organisation with it's own start date etc. This organisation is associated with the African Methodist Episcopalians which is associated with the Methodist Episcopalians. Associated with this organisation are three different buildings, which have succeeded each other on the same site, with new buildings being built as earlier buildings were destroyed. Filceolaire (talk) 23:03, 19 June 2015 (UTC)

Are there rules for the main label

The current Pope Q450675 has a en-label of just Francis. Although it is has aliases, if I want to quickly produce a list of people born on 17 Dec 1936 I would only use the label and get the following result.

Frank Martinus
Tommy Banks
Francis
Al Miller
Kenneth Robinson

The single name Francis looks out of place. To me the main label should be a bit more descriptive but I can not find any policies or guidelines? Periglio (talk) 22:41, 18 June 2015 (UTC)

See Help:Labels. Personally I think his English label should be "Pope Francis I". Filceolaire (talk) 00:48, 19 June 2015 (UTC)
I agree that it should include "Pope", because he is not commonly referred to as just "Francis", although I'm not so sure about the "I". My impression is that "Pope Francis" is more common than "Pope Francis I", see for example the Twitter account and the Vatican news page. - Nikki (talk) 01:27, 19 June 2015 (UTC)
I changed it to Pope Francis, same as the main title in the article. Not sure if it is correct to say Pope Francis I until such time as there is a Pope Francis II? There might never be a Pope Francis II, in which case the numbering is never needed. Danrok (talk) 02:11, 19 June 2015 (UTC)
Pope Francis For what it's worth, Francis has declared that his name in English is "Pope Francis" and not "Pope Francis I"--he has been explicit on the matter. —Justin (koavf)TCM 02:48, 19 June 2015 (UTC)
Thanks for changing. But for future reference, is there any official policy on labelling? Periglio (talk) 13:11, 19 June 2015 (UTC)
No rule or an unique rule: the label should be clear enough to allow a clear identification of the concept defined by the item. Snipre (talk) 14:18, 19 June 2015 (UTC)
Now one risks confusing him with Francis Pope (Q20160365). Maybe the list should give a bit more context than the mere label. --- Jura 14:48, 19 June 2015 (UTC)

I prefer short names, i.e. Francis over Pope Francis, but both are sufficiently short. Also, in a list like this that is meant for human consumption, I would always state "LABEL (DESCRIPTION)" - Wikidata relies on the Label/Description pair to being both unique and identifying, whereas the label itself is not.

But one thing with the pope - it should be consistent with the other popes. So don't just change Francis to Pope Francis if it is decided that this is the right thing to do, but do it for all popes (and adapt the aliases respectively). Such a change might need a bit more consensus. --Denny (talk) 14:59, 19 June 2015 (UTC)

It is really odd now, if one looks at the entire User:Jura1/List of popes. --- Jura 15:02, 19 June 2015 (UTC)
An alternative would be to add something else than the label to describe every item in your list. -- Innocent bystander (talk) 09:11, 20 June 2015 (UTC)
I think Periglio should add something in the line with autodesc, e.g. P106 or P39, P27. --- Jura 09:13, 20 June 2015 (UTC)
Ah, it already exists: Autolist1. --- Jura 09:21, 20 June 2015 (UTC)
It's only odd because someone has taken it upon themselves to change all of the English pope labels by deleting the word Pope. It is normal, in English (US and UK) to refer to Pope Francis as Pope Francis and not Francis, same applies to all popes. See for example [10]. Danrok (talk) 12:38, 20 June 2015 (UTC)
And that is a good reason to not have a universal policy for labels in all languages. In my language is it not natural to have "pope" in the label here, while we often add the word "county" or "municipality" in almost every item for those kinds of entities, while many other languages prefer to not use such words in their label. -- Innocent bystander (talk) 12:48, 20 June 2015 (UTC)
In which case the EN version should say Pope Francis and your language will say Francis? The universal policy can be that a label reads correctly in whichever language it is. I accept now that the description is an integral part of that, but "Pope Francis, 266th Pope of the Catholic Church" still read betters for me than "Francis, 266th Pope of the Catholic Church" Periglio (talk) 22:02, 20 June 2015 (UTC)
It reads "Franciskus", since that is the name in use here. That name has been confirmed by the local catholic churc, I think. The description reads "present pope", but that is less good, better would be only "catholic pope (20XX-)". I try to avoid "266" since such numbers can be disputed. (Was s:t Peter number 1?). I see some projects add such numbers to Swedish kings, but I do not recommended that at all, since we do not know who number one was. -- Innocent bystander (talk) 08:02, 21 June 2015 (UTC)
Why not use names from naming authorities (Library of Congress Name Authority File for writers' names, Getty Union List of Artist Names for artists, etc.)? DavidJHowe (talk) 15:29, 20 June 2015 (UTC)

Something we could do?

I ran into this clip. Maybe some inspiration for tools to build for Wikidata? Multichill (talk) 19:35, 19 June 2015 (UTC)

HTTPS

22:01, 19 June 2015 (UTC)

How do we classify humans who don't have an occupation?

In genral for humans we have

<instance of><human> and
<occupation (P106)><whatever> and
<field of this occupation (P425)><more detail>

What do we do for martyrs and murderers and children? These aren't occupations. Filceolaire (talk) 00:19, 20 June 2015 (UTC)

And this is why having a single P31 is a bad thing. (One of the few times I disagree with User:Emw.) --Izno (talk) 01:06, 20 June 2015 (UTC)
@Izno: there will only be one P31, whatever properties we add :) But I really don't get how it is related to filceolaire problem. With one typing properties we can still create as many class as we need. With class classification we can sort them by domain or whatever we want (see help:classification. TomT0m (talk)
  • "What do we do for martyrs and murderers and children? "
    For the last two there are specific properties: "convicted of" and "date of birth". --- Jura 01:37, 20 June 2015 (UTC)
    • Not everyone should have the occupation (P106) filled in. And if you persist on entering something you can use the "no value" or even "unknown value" options. Edoderoo (talk) 06:37, 20 June 2015 (UTC) PS: right now there are 1.2 million people with no occupation (P106), so pick a few to fill it in.

and martyrs? Filceolaire (talk) 11:11, 20 June 2015 (UTC)

I don't think we should have something like "in general". Wikidata is not an ontology, and we have no rules that every human has to have an occupation. We only have the rule that everything with an occupation is most likely a human. Why can't we just set "instance of" -> "human" and "instance of" -> "martyr"? -- Bene* talk 11:30, 20 June 2015 (UTC)
It just occurred to me that we can state <cause of death (P509):martyr (Q6498826)>. Filceolaire (talk) 12:45, 20 June 2015 (UTC)
That's not really useful though. e.g. Joan of Arc (Q7226) already has cause of death (P509):death by burning (Q468455). Thryduulf (talk: local | en.wp | en.wikt) 19:50, 20 June 2015 (UTC)
@Bene*: All this is not really related to be an ontology or not. Clearly Wikidata is an ontology because it represents in term of properties, items, statements, knowledge, concepts and relations beetween concepts. TomT0m (talk) 19:04, 21 June 2015 (UTC)
Sure, but it doesn't say that all humans have to have some properties and should not contain some other properties. This is the openness which makes Wikidata more than a strict ontology. -- Bene* talk 20:29, 21 June 2015 (UTC)
@bene*: it's not really related to beeing an ontology or not, more to strict typing, inferences or not inferences, and if which, which kind of inferences, inconsistency beetween data and rules or inferences, the nature of the rules and so on, open or closed world assumption in inferences and so on. TomT0m (talk) 11:56, 22 June 2015 (UTC)
How about significant event (P793) to describe a martyrdom or a murder? Periglio (talk) 07:00, 21 June 2015 (UTC)

@Filceolaire: Stating that a person has an occupation is not a classification. It's just that knowing his occupation, we can classify him in the class (or list) of persons who have this occupation. Classification is the art of building classes knowing properties of the classified objects. TomT0m (talk) 08:32, 21 June 2015 (UTC)

I agree. We have a couple of classification properties but occupation certainly isn't one. -- Bene* talk 20:29, 21 June 2015 (UTC)
and yet, most users if asked to complete "is a...." about a person would choose an occupation. Saying that <occupation:author> is different from <is an:author> may be true for you but I can't see the difference. Filceolaire (talk) 22:05, 21 June 2015 (UTC)

Completeness of Wikidata queries

Is there a way to know whether a Wikidata Query is complete or not? Specifically, looking at Query: TREE[3557561[40]] currently lists none the children of Victor Rothschild, 3rd Baron Rothschild (Q3557561). When (and how) will a query like that become complete? Gabbe (talk) 11:53, 20 June 2015 (UTC)

  • @Gabbe: one shall assume that it is not complete and will never be. -- Vlsergey (talk) 22:12, 20 June 2015 (UTC)
In this particular case I guess you mean Query: TREE[3557561[][40]] but generally Vlsergey is right: Wikidata will never be complete. -- Bene* talk 20:33, 21 June 2015 (UTC)
@Bene*: no, that query you provided seems to (recursively) list items that have Victor Rothschild, 3rd Baron Rothschild (Q3557561) as a child. That is, your query provides his ancestors, not his descendents. And let me be clear: I know that Wikidata will never be complete. I was wondering whether a Wikidata Query will ever be. In this specific case, I could use wdq2sparql to turn my query into this one, which is complete. But there are other statements for which even the Sparql queries don't seem to be complete. Gabbe (talk) 06:16, 22 June 2015 (UTC)
This is an example of the latter. It lists none of the children of Aline Caroline de Rothschild (Q4726753). Gabbe (talk) 06:30, 22 June 2015 (UTC)

There is no way to know if a query has all the valid results. What a query does is provide you with all the results Wikidata holds. As a consequence, the question can be answered in several ways. You can validate one query at a time and compare it with sources. Another answer is by importing as much data that is reasonably correct and validate it against other sources. It will increase the likelihood of a query to have all the valid results. There is a third option, not recommended.. it is pissing in the wind. There is nothing but hard work to complement Wikidata with data that we learn about. It will never be complete but it can be as good as we make it. Thanks, GerardM (talk) 17:18, 22 June 2015 (UTC)

@GerardM: to clarify, I don't mean that the query is incomplete in the sense that there is true information that isn't on Wikidata. I'm saying that the query is incomplete in the sense that it doesn't reflect all the data that is on Wikidata. Gabbe (talk) 06:35, 23 June 2015 (UTC)
Query (WDQ) will provide all data that is on Wikidata. One proviso, the data is in an external database and it may take 15 minutes for updates to be included. Thanks, GerardM (talk) 09:06, 23 June 2015 (UTC)
... which is why I find it so strange that Query: TREE[3557561[40]] doesn't list any of the items that have child (P40) from Victor Rothschild, 3rd Baron Rothschild (Q3557561) (such as Jacob Rothschild, 4th Baron Rothschild (Q2026032) for example). Those statements were added to Wikidata weeks ago. Gabbe (talk) 13:40, 23 June 2015 (UTC)
All three were added at the same time: once in a while replication fails and statements get skipped. As this is not an e-commerce system, things like that can happen. You could remove the statements and re-add them later or wait for the database to be refreshed. --- Jura 13:45, 23 June 2015 (UTC)
Great, thanks! How often does the database refresh, approximately? Gabbe (talk) 14:44, 23 June 2015 (UTC)

Federation of individual sports people

Is there an item to attribute individual sports people, like tennis players, athlets, boxers, snooker players, etc. to the confederation they played for? E.g. how would Roger Federer be attributed to the Swiss Tennis Federation? For team players, there is Property:P54, but I couldn't find something similar for indivudual sport people. 85.212.3.151 21:37, 20 June 2015 (UTC)

I think member of (P463) is what you are looking for. Filceolaire (talk) 21:57, 21 June 2015 (UTC)

Wikidata weekly summary #163

There seems to be a bug with our showcase item en:South Pole Telescope – both start and end time of construction are off by one month: South Pole Telescope (Q1513315) says November 2006–February 2007; article infobox says October 2006–January 2007. DSGalaktos (talk) 12:04, 21 June 2015 (UTC)
@DSGalaktos: Thanks for pointing this out! I've raised this at en:Module_talk:Wikidata#Offset_in_months.3F. This showcase/example is still very much a work in progress, as I figure out how to include more of en:Template:Infobox telescope's content directly from wikidata, so bugs are expected - but hopefully everything will be working nicely sometime soon! Thanks. Mike Peel (talk) 20:30, 3 July 2015 (UTC)
Oh hey, someone saw this! :D thanks for raising the issue where it apparently belongs (iiuc it’s a bug with the wiki’s date parser, not with wikidata). —DSGalaktos (talk) 20:59, 3 July 2015 (UTC)

[BUG] Time format does not support BC dates

I am sure this is already known, but I could not find a ticket for this yet.

E.g.: If you query the Great Pyramid of Giza: http://wdq.wmflabs.org/api?q=items[37200]&props=571 You should get 2560 BCE as the inception date. However, you get: "+00000000000-00-01T00:00:00Z"

This is a serious issue, because I was building an app for querying also BC dates - however a query such as ...BETWEEN[569,-3000,-2000]... will not work correctly. Please advice.

There is a whole list of date related bugs. See https://phabricator.wikimedia.org/T87764  – The preceding unsigned comment was added by Jc3s5h (talk • contribs).

Modeling changed attribution

Both me and Jane ran into the same problem again. How to model the fact that a painting was first attributed to painter A and later to painter B (or anonymous)? We set the former attribution to rank "deprecated", but how to model the fact that up until a certain date that was an acceptable attribution? See our examples Still Life with Beer Mug and Fruit (Q152123) & Woman standing at the water's edge (Q17335688). Multichill (talk) 20:32, 21 June 2015 (UTC)

start time (P580) / end time (P582) ? -- Vlsergey (talk) 20:35, 21 June 2015 (UTC)
As a qualifier? That seems to be a bit odd. It's like saying someone was the creator before 2010. Multichill (talk) 20:38, 21 June 2015 (UTC)
(edit conflict) I think the way you did it in Still Life with Beer Mug and Fruit (Q152123) is fine. Just add end time (P582) and possibly start time (P580) to indicate that this was the common opinion in that time. -- Bene* talk 20:39, 21 June 2015 (UTC)
Lots of things look a bit odd when reduced to standardised statements. That is why we have wikipedia articles where our readers can find out more. Anything that encourages people to look at the articles seems like a good thing to me. Filceolaire (talk) 21:54, 21 June 2015 (UTC)
  • How about adding a qualifier "P31" with an item labeled (e.g.) "incorrectly attributed" to the van Gogh statement? --- Jura 22:16, 21 June 2015 (UTC)
  • I think setting it to deprecated is Good Enough at the minimum. The problem I'm feeling is that start/end time makes it look like he actually painted it over that time frame and then handed it off to someone else. So at best, you've got an ambiguous looking qualifier. --Izno (talk) 03:12, 22 June 2015 (UTC)
  • start time (P580) / end time (P582) might be used as qualifier in normal statement for specific purposes. Changing the rank to deprecated shall not change the meaning of the statement, except marking it as invalid as a whole. So the solution for the problem might be another generic qualifier property deprecated date, which could be also a justification why the deprecated statement exists. Paweł Ziemian (talk) 17:40, 22 June 2015 (UTC)

Review of added fact

I added a cause of death with sourcing to George Washington. Is the creation of the entities necessary for that sourcing an appropriate time to create an entity that isn't connected to another Wikimedia site... creating an entity for the publisher and for the edition? --Brantgurga (talk) 20:45, 21 June 2015 (UTC)

Yes it is. For a periodical or newspaper you can create an item for the periodical and use issue (P433) in the reference to identify the particular edition. See Help:Sources#Scientific, newspaper or magazine_article. Filceolaire (talk) 21:50, 21 June 2015 (UTC)

Allow the empty string as value for properties representing a unique identifier?

I am currently having a discussion with another user on whether the empty string should be an admissible value for a property representing a unique identifier at Property talk:P1157#Allow or disallow empty values?. If you have a view on the matter, feel free to look at that discussion, add your opinion and your reasons there! --UV (talk) 23:39, 21 June 2015 (UTC)

Commented there as requested. --Izno (talk) 03:20, 22 June 2015 (UTC)
Thank you. --UV (talk) 21:46, 22 June 2015 (UTC)
Izno is working on fixing his revert. --- Jura 21:49, 22 June 2015 (UTC)

Adding items to complete the results of a query

Hoi, I blogged about the IEEE Cledo Brunetti Award. The current data on Wikipedia is incomplete and most of the people do not have an article. I have added missing items on Wikidata for several of them and I demonstrated that it is easy to produce a list on a Wikipedia.

The point I want to make is that it is trivially easy to add missing items. They then exist to complete queries and are notable because they serve a structural need. However this point of view is not universally shared. So I make the point explicit; items may be added to complete the results of a query particularly when the new items are notable in their own right like these winners for this prestigious award. Thanks, GerardM (talk) 07:19, 22 June 2015 (UTC)

I don't see why you wouldn't create the missing items. Everybody seems to agree on that. --- Jura 07:22, 22 June 2015 (UTC)
Yeah, create items if that allows you to give more data. And, the feeling you get when you have a more complete set of data is fantastic. It's always nice to see a full timeline on Reasonator or a good family tree. Sjoerd de Bruin (talk) 07:26, 22 June 2015 (UTC)

Wikidata:Project_chat/Archive/2015/03#An item for every street? --Atlasowa (talk) 10:56, 22 June 2015 (UTC)

A street is an award? ;) --- Jura 06:46, 23 June 2015 (UTC)

Wikidata is incapable of representing birth and death dates

Wikidata has a data model for dates and times and a user interface. While the user interface is not the only method for entering dates and times, many are entered that way, and users of other methods are likely to be influenced by the user interface behavior. Leaving aside the problems of pre-1924 dates, most modern dates cannot be properly represented.

The data model indicates a TimeValue represents an instant in time; we represent the date and time, not just the date. But the user interface does not allow us to enter a time, and only allows us to enter universal time. Also, the before and after fields cannot be entered. The precision can be entered, but the applicable choices, given the other limitations, are day or month. So if I know a person were born on January 15, 1990, in New York City, I have two choices. I can enter the birth date as January 15, 1990, universal time, precision day, which is probably false. Or I can correctly, but very loosely, enter the same, but change the precision to month.

Our real knowledge about the birth is it occurred, including the ends of the intervals, from 05:00 January 15, 1990, UT, to 05:00 January 15, 1990. But if we enter January 15, 1990, UT, precision day, in the data model, what does it mean? Does it mean from 00:00 January 15, 1990, to 24:00 January 15, 1990 UT? The birth might have occurred at 23:59 AM January 15, 1990, New York time, which falls outside this interval. Or a person might see the Wikidata entry and a birth certificate which says 23:00 January 14, 1990 New York time, and decide they are consistent.

Several other interpretations of the precision are possible, such as one full day either side of the stated date. Or it could be interpreted as the day of the month is uncertain, but the month is certain. In summary, we have hardly any idea what our dates mean. Jc3s5h (talk) 15:12, 22 June 2015 (UTC)

I thought they were already working on this. Besides, you're almost the only one making a issue of it every time... Sjoerd de Bruin (talk) 15:33, 22 June 2015 (UTC)
It can't be fixed until we know what the solution should look like. And if everyone else thinks false dates are just fine, Wikidata should be shut down and tossed on the scrap-heap of history. Jc3s5h (talk) 15:42, 22 June 2015 (UTC)
Jc3s5h, imprecise date is not false or wrong date… thousands of people (on wikidata and pedias, and even more on commons) only have a decade, or a century, as date of birth or death… a few hours does not seem such a problem… we're not in Court here, trying to solve difficult murder or medical errors problems ;) --Hsarrazin (talk) 15:55, 22 June 2015 (UTC)
Wikidata represents instants in time. Of course most instants in time are imprecise; only the ones that are defined by authorities are exact. But the degree of precision can be stated. The data model purports to state the precision. But the user interface does not allow the precision to be stated properly. Purporting to state the precision and then knowingly stating an incorrect precision is a lie, and destroys the credibility of the liar.
Furthermore, we have a great many sources which state dates to a rather clear precision; a certain calendar day in a certain time zone or location. The current state of the Wikidata data model and user interface forces us to state a false precision, or degrade the precision contained in the source. Jc3s5h (talk) 16:06, 22 June 2015 (UTC)
Only my great grandparents know when my grandfather was born, and they had two different opinions about the exact date. That never became any problem for him. Instead he celebrated his birthday twice every june, instead of making a problem out of it. -- Innocent bystander (talk) 16:10, 22 June 2015 (UTC)
The people who created POSIX decided careful consideration of time issues was just too much trouble, and now POSIX is the object of ridicule of those who actually understand time; see this academic paper. If WikiData takes the attitude that there is no need to be able to correctly represent the time of events, either it will be abandoned, or it will be the object of never-ending ridicule, just like POSIX. Jc3s5h (talk) 16:41, 22 June 2015 (UTC)
The academic paper you link mentions POSIX once (page 7, unless pdf.js’ search is broken), and hardly in a way that I would call “ridicule”: “Yet another misconception reflected in early POSIX standards was that more than one leap second could be inserted at a given time.”
And how is the meaning of “date of birth: January 15, 1990” unclear? That person was born on that day. They celebrate their birthday on that day. Do you celebrate your birthday at the exact minute your birth started, and expect all congratulations to arrive on that exact time? (I certainly don’t, I’m sound asleep at that time ☺)
Of course this leaves room for improvement. There are people whose birth starts shortly before midnight and ends shortly after. And it would of course be nice to be able to record the exact start and end time of the birth (presumably with start time (P580) and end time (P582) qualifiers) in those cases where it’s known. But what we have works for the vast majority of cases, and is in no way “broken” or “unusable”. —DSGalaktos (talk) 16:59, 22 June 2015 (UTC)
The typical entry contains the actual calendar date of the birth or death in the jurisdiction where the event occurred, but purports that the event happened at 00:00 UT +/- zero days. A person who believes the data model will at least get the calendar date right for places with a positive time zone offset, but will get the wrong calendar date for anyplace in the Americas. Since most birth and death dates in the Americas will be wrong, this can hardly be described as "the vast majority of cases". Jc3s5h (talk) 18:28, 22 June 2015 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Let me give you a use case. Joe user wants to extract the date and place of birth of a list of people who might run for President of the United States for the term beginning Friday, January 20, 2017, and eliminate those who are not eligible on account of age (must be at least 35 years old). An entry is found for Mary Smith, born January 21, 1982, precision day. Before, after, and time zone are all set to zero. Thus she was born at 00:00 January 20, 1982, universal time, +/- zero days. Also, she was born in New York City, and the date in New York City at that time was January 20, so she is eligible to be the next president (as far as age is concerned). Jc3s5h (talk) 17:31, 22 June 2015 (UTC)

Using Wikidata to determine who is eligable for presidency, I would say: use official sources like birth registers. Mbch331 (talk) 17:38, 22 June 2015 (UTC)
Jc3s5h Your problem is not a problem of WD but a general problem of precision. WD is not more accurate that any other dictionaries or encyclopedias so stop to say that WD is incapable of representing birth dates. If you have a accurate date and time of birth or death WD gives you th possibility to enter this kind of informations. But as few persons take care of details like this one I don't understand why WD should be more catholic than the pope. Really Jc3s5h, when you ask the birth date to one of your friends, do you ask him too the time with the time zone ? So why do you want to have another reaction with WD ? Snipre (talk) 18:27, 22 June 2015 (UTC)
@Jc3s5h: “The typical entry […] purports that the event happened at 00:00 UT +/- zero days” – it does no such thing. The precision of the entry is “day”; anything beyond that (”00:00”, “UT”) is your own interpretation. (I assume that we will eventually have higher time precisions; I suspect that timezone complicatedness might be the reason we don’t have it yet. (But I’m a nobody, don’t quote me on any of that.)) —DSGalaktos (talk) 18:47, 22 June 2015 (UTC)
Snipre's claim "WD is not more accurate that any other dictionaries or encyclopedias" seems badly written. Wouldn't we want to say "WD is no less accurate that any other dictionaries or encyclopedias"? But WikiData is less accurate than most modern dictionaries, books, or encyclopedias, because they usually get the calendar date correct for the place a vital event occurred. If the data is entered through Wikidata's user interface, Wikidata always gets it wrong for the Americas.
If I ask my friend for his/her birthdate, it might be because I want to take her for a drink in a bar. Or maybe I'm hoping he will be able to get a reduced rate on a ski lift ticket due to his advanced age. In any such case, the time of day of the birth does not matter, but the birth date does. Due to the legal treatment of birth dates, small differences in time are magnified to large differences in time. A minute becomes a day for the lift ticket or the alcoholic beverage. A minute becomes 4 years if running for president. A minute becomes a lifetime if a person was born one minute after a country became independent, and the child wishes he could have the citizenship of the pre-independence country. A minute becomes death if the person is in the last batch of people to be inducted into the armed forces just before the end of a war. Jc3s5h (talk) 18:56, 22 June 2015 (UTC)
My passport doesn’t even include my birth time of day. It’s not even possible for a bar, or a ski lift, or whomever to get my age to that precision even if they cared about it (which I, for almost all cases, highly doubt). Likewise, the notion that soldiers in the armed forces were sorted by minute of their birth is ridiculous. And for the US presidency, I don’t think the constitution states at which exact point during the inauguration ceremony the minimum age of 35 must be fulfilled, so even in that case I think the relevance of the minute of birth is highly dubious.
I’m sorry, but I just feel like you’re trying really hard to create a problem that doesn’t exist. —DSGalaktos (talk) 19:10, 22 June 2015 (UTC)
Of course a ski lift counter clerk doesn't care about the minute of a patron's birth. But if if a skier's birth certifcate says the skier was born at 23:59 January 1, 1950 (the time zone being implied by the place of birth), that skier will have an ID (such as driver license or passport) that says January 1, 1950 and when the skier walked up to the Killington Mountain ticket counter on January 1, 2015, that skier paid $78. The next skier in line, who's birth certificate says 00:01 January 2, 1950, had a driver license that said January 2, 1950, and that skier paid $92. Jc3s5h (talk) 19:31, 22 June 2015 (UTC)
I very much agree with Jc3s5h's points here. The date data explicitly says that all dates are in UTC, which is clearly not how we're using them, making the data ambiguous and inaccurate. Fortunately, I think there's an easy remedy for this. Given that the UI doesn't yet allow any sort of easy editing of time zone data, I assume all dates are currently set to timezone:0. I recommend that we ask the dev team to replace all of these with data indicating that the time zone is unknown, and default in the future to unknown time zone unless the user specifically adds it, instead of the current practice of defaulting to UTC. Thus, dates like January 15, 1990 by default have an assumed range of 48 hours. --Yair rand (talk) 19:25, 22 June 2015 (UTC)
It's not as simple as replacing the data. Currently, there is no syntax in the data model to represent an unknown time zone. Also, allowing for an unknown time zone is not the only possible solution. The user interface could be altered to allow entry of time zone, before, and after. Then a correct date, such as June 22, 2015, time zone = -240 minutes, precision day, before = 0, after = 1, could be entered for a birth that occurs today on the east coast of the US and which is mentioned in tomorrow's newspaper with only the date of birth stated.
Then, we would know that nearly all vital events with precision day, before = 0, after = 0, and time zone = 0 are erroneous. Such entries could be flagged for correction. Perhaps barriers to reading such entries could be errected, such as requiring queries to set a flag true indicating the requester is willing to accept dubious dates. Jc3s5h (talk) 19:42, 22 June 2015 (UTC)
I think the syntax needs to be extended to allow an unknown time zone. In many cases we will neither know the time of day nor the time zone, giving a normal range of 48 hours. Presumably, extending the interface to allow inputting time zones and ranges is something that's planned eventually, but I suspect we're not going to see it any time soon. (How do before/after even work, by the way?) Having a range of 48 hours doesn't mean the data is dubious any more than dates that only say "20th century". --Yair rand (talk) 22:45, 22 June 2015 (UTC)
In the user interface, precision, before, and after just don't work. In the data model, precision gives the units of measure for before and after. Before and after give the number of precision units the time could be before or after the stated time. So if the user interface were fixed and all we knew was that someone was born 22 June 2015 with no hint of the location, it could have been as early as 00:00 on Christmas Island in the -14 hour time zone (10:00 UT June 21) or as late as 23:59:59 on Baker Island in the +12 hour time zone (11:59:59 UT June 23). So we could set the precision to 12 (which means hour), set the time zone to 0 minutes, set the date/time to 00:00 January 22, 2015, set before to 14, and set after to 36. Of course, other settings would create the same range of uncertainty, such as date/time of 12:00 January 22, 2015, before = 26, after = 24. Jc3s5h (talk) 23:31, 22 June 2015 (UTC)

Maybe a solution would be that the interface allows to store the information that is found in the sources? UTC before that existed is a derived information. Eldizzino (talk) 16:18, 23 June 2015 (UTC)

The Taiwan Problem

Hi all, I would appreciate it if you can join the discussion between me and Neo-Jay regarding Taiwan (Q865) and Republic of China (1912–1949) (Q13426199) here. —Wylve (talk) 11:20, 23 June 2015 (UTC)

Pick some "official" source if you can, e.g., WIPO uses "Taiwan, Province of China", that's also what you find on ISO 3166-1 alpha-2 if enwiki got it right (ISO 3166 is a non-free standard.) –Be..anyone (talk) 18:36, 23 June 2015 (UTC)
I think you may have misunderstood the dispute. It seems to be about whether the Republic of China as it existed from 1912 to 1949 and the Republic of China since 1949 (commonly called "Taiwan") are the same entity or not. --Yair rand (talk) 19:58, 23 June 2015 (UTC)
My opinion is that in some ways they are the same and in other ways they are different. There are statements you can make about one which are not correct if applied to the other so I believe two separate items are needed. Filceolaire (talk) 20:37, 23 June 2015 (UTC)
Statements about one that would be incorrect of the other, even if qualified by start/end time? Could you give an example of such a statement? --Yair rand (talk) 21:49, 23 June 2015 (UTC)

Donostia-San Sebastián Wikidata Editathon (03.07.2015)

Hi Wikidatans,

We are organizing a Wikidata Editathon in order to encourage Wikidata edits about San Sebastián (wikidata:Q10313) and its culture. The motivation behind the event is that San Sebastián will be European Capital of Culture in 2016. Moreover, we are organizing this event to encourage people in our region to edit and use Wikidata.

We will work in teams and we will edit items of different topics (e.g. music, literature, points of interest related to San Sebastián). We will show them the different use cases one can go through in Wikidata (e.g. create an item, enrich the description of an item, connect items to an external data source, import and review data from Freebase, etc.). We will also discuss application ideas and talk about collaborative systems.

The event will take place in San Sebastián (Spain), on July 3rd, at the Faculty of Informatics (UPV/EHU). You can find more details (in Spanish) here: https://sites.google.com/site/donostiasansebastianenwikidata/

If anyone is interested in either participanting or helping out (resolving doubts or explaining guidelines that one should take into account while editing Wikidata), please let us know. We would be really happy to have you there. You can write an email to: dssenwikidata@gmail.com

Cheers,

Cristina Sarasua ( Criscod)

Hoi, I can imagine that you want to have all the information in the languages of Spain available on all the content.. Consider using Reasonator with Widar enabled. The good news is that as you add more labels in any of the languages, you will find them propagate to all the items where a statement is used with that item.. If you have questions, let us know :) Thanks, GerardM (talk) 18:41, 23 June 2015 (UTC)
Hi, thanks! Really nice. We will consider it.Criscod (talk)

historical country (Q3024240)

@Infovarius: When did England start to be a former country [11]? And when Istanbul [12]? Eldizzino (talk) 16:11, 23 June 2015 (UTC)

@Eldizzino: welcome to Wikidata. You must be confused, Infovarius' talk page is at user talk:Infovarius. Multichill (talk) 16:16, 23 June 2015 (UTC)
@Eldizzino: In ENGLISH, it maybe looks strange to see England as a "former country". In my language, England is not a country. In English, You use the same word to describe the status of England today as you do with Germany, Colombia and Japan. In other languages, you cannot be sure that the same word is used to describe those places. I use the same word to describe England as when I describe Norrland (Q203505), and Norrland has never been a independent nation with a constitution of it's own. Do not look to deep into the glass of your own language, look into what the word means instead of how it is spelled. -- Innocent bystander (talk) 07:35, 24 June 2015 (UTC)
@Innocent bystander, Infovarius: There are two questions:
  1. The claims "instance of" for persons and organizations and are valid for a certain period of time, a start and end point could be attached as qualifier. Would be interesting to see, what language does not allow to make a statement "instance of X" for the past. If there really is a language that cannot make the statement that a former country "was an instance of a country", then maybe that language would need a complete duplication for many more classes? "Carlos I" - "former human", "Santa Marta" - former ship?
  2. The question what term to use, is another one. In Wikidata England and Norrland are instances of different classes, the issue about translation of terms that you brought is solved for these. There is country (Q6256) "country" and Malta is not marked directly as one, but is marked as sovereign state (Q3624078). But, I see "sovereign state" as a subclass of "state" which is marked as "said to be the same as country". I have no position on whether sovereign state is a subclass of state is true. Maybe it is for many Romance language influenced languages that use state/estat/estato/estado/stat etc, but maybe it is not universally true.
Combining both: England is a country of the UK, and was a sovereign state. Now, "former country" is dubious. It is not former in the sense of country as some kind of region. It is only in the sense of country as "sovereign state", but then the label could probably be adjusted to match sovereign state (Q3624078). Eldizzino (talk) 13:15, 24 June 2015 (UTC)

There is also some discussion at User_talk:Neo-Jay#former_country_.28Q3024240.29. It is also related to storing time information and the discussion started above at Wikidata:Project chat#Wikidata is incapable of representing birth and death dates by Jc3s5h. Eldizzino (talk) 16:31, 23 June 2015 (UTC)

former XYZ in instance of (P31) statements

@GerardM, Neo-Jay, Infovarius: One could disallow any "former XYZ" as part of an "instance of" claim, when it means, that something used to be a XYZ for some time. But then, one need to have a qualifier indicating the end. If "end time" has limitations, there might be some reason to have "former XYZ" on an item. Eldizzino (talk) 13:15, 24 June 2015 (UTC)

So? It is ridiculous to have "former XYZ". Thanks, GerardM (talk) 13:24, 24 June 2015 (UTC)
@GerardM: A place might be an agglomeration of ruins, and formerly a city. Is the item then "agglomeration of ruins" or "city"? Or two items, a city with ruins and current population=0 and another one that is an agglomeration of ruins? Eldizzino (talk) 15:09, 24 June 2015 (UTC)
This is why we have start time (P580) and end time (P582) qualifiers. --Yair rand (talk) 15:20, 24 June 2015 (UTC)
@Yair rand: If end time is in the past, but not any further specified? Eldizzino (talk) 02:15, 25 June 2015 (UTC)
Not even specified to the nearest year, century, millenium, or further? Then use end time (P582): unknown value. --Yair rand (talk) 02:32, 25 June 2015 (UTC)
  • I believe instance of (P31) should have only values that can't be changed over time. For example Berlin (Q64) is always human settlement (Q486972), it will never be human (Q5). But would one say Berlin (Q64) IS federal capital (Q257391), he need to make an assumption about time (current time, past, future?).
  • my idea is that class of object is something that never change. Everything else shall be placed in distinct properties. Same with "former XYZ". Since it's not an immutable property of object, it shall not be placed into instance of (P31). -- Vlsergey (talk) 13:45, 24 June 2015 (UTC)
    A city changes size and becomes a town, becomes a settlement (in the case of abandonment). Your idea clearly isn't useful. --Izno (talk) 14:35, 24 June 2015 (UTC)
    Whatever size it has, it's always human settlement (Q486972). The idea is to prevent changes in P31 once object described. Currently P31 is very polluted field. It even has values like big city (Q1549591) that shall never appear in element at all. So, there are two directions. First is to say "everything is okay, go on". Then feel free to add whatever value you want to be, including "former XYZ". Second way is to try cleanup this field, introduce some criteria. And remove 80% of today-only values. -- Vlsergey (talk) 14:47, 24 June 2015 (UTC)
    False dichomoties rarely work as arguments. --Izno (talk) 14:52, 24 June 2015 (UTC)
    neither text untranslatable by google translate ;) -- Vlsergey (talk) 15:16, 24 June 2015 (UTC)
    Try "dichotomies" instead. >_> --Izno (talk) 16:50, 24 June 2015 (UTC)
    @Vlsergey, Izno: If a settlement got the status of a town, it should be said under which system it is a town or city. So, New York (City) is a human settlement. How could one store the information that it became a town/city - as officially decreed by some government? Maybe one can go on and let the field get filled up with anything people want, but the preferred P31 should be something that ~never changes. Eldizzino (talk) 15:09, 24 June 2015 (UTC)
    @Eldizzino: the best thing here will be to create "official status" property (main snak is status itself, when received, given by). -- Vlsergey (talk) 15:15, 24 June 2015 (UTC)
    @Vlsergey: Thank you. Would you allow only one single "instance of" claim? An example of having multiple such claims is Baja California (Q58731) - and all but one are wrong, this entity never was a province of Spain. Eldizzino (talk) 15:55, 24 June 2015 (UTC)
    @Eldizzino: from my point of view it should be territory (Q183366). Well, it's not like "single value rule", I just can't remember a case when single object can't be described by single "read only" class. If some object have two "read only" values in P31 they probably can be merged into single class. -- Vlsergey (talk) 16:30, 24 June 2015 (UTC)
    Merging classes along orthogonal distinctions is bad. It's why Wikipedias have such a hard problem with categories. --Izno (talk) 16:50, 24 June 2015 (UTC)
    @Izno: Not really, it's because it's hard to get the parent categories in Wikipedia and because there is no distinction beetween classes and instances. In Wikidata, by convention, all instances of a subclass is an instance of the superclass and WDQ has a construction to this. So its much less of a problem if Wikidatan uses constructions like CLAIM[31:(TREE[5][][279279])] in queries. TomT0m (talk) 20:51, 24 June 2015 (UTC)
    We already have this in the form of the "preferred" ranking. If there are actually wrong statements, those should be removed (or sourced and qualified). All statements in general should have a source and qualification and this is no less true simply because we're dealing with P31. As for your assertion that it was never a province of Spain, I'm skeptical. The entire country of Mexico (and a good chunk of the US!) was to some extent. --Izno (talk) 16:50, 24 June 2015 (UTC)
    Having classes for "former something" as an alias to "instance of something qualified as an end date" could be useful as well. TomT0m (talk) 20:51, 24 June 2015 (UTC)

@Izno: The United Mexican States is a federation. That federation never was a province of Spain. The State of Baja California was established from the Baja California Norte Territory which was a result of a split of Baja California Territory, which in turn never was a province. But the territory they cover once was covered by one or more provinces of Spain. But there are extra items for theses. If succeeding entities are all provinces of the Spanish Empire, then the list of provinces of the Spanish Empire is growing everyday a new such entity is established. Eldizzino (talk) 02:07, 25 June 2015 (UTC)

Now this is ridiculous : Author (1500–1490) ♂; spanish writer

This is the autodesc of Martí Joan de Galba (Q1338820).

Well, I know, because I've been contributing for quite a time that it means that Martí Joan de Galba was born in the 15th century, and died in 1490, but really... this does not look serious at all.

The display of the century with the last year of the century gives really bad effect.

So, how would it be possible to make it look just a tiny weeny bit more readable for someone who is not accustomed to it ? in BNF AC, the coding is 1400 (for born in 15th century) and 1499 (for died in 15th century). At least, noone is born after one's death :/ --Hsarrazin (talk) 16:36, 23 June 2015 (UTC)

The whole issue of how to handle "century" dating is handled poorly here, and additionally requires localization. StevenJ81 (talk) 16:44, 23 June 2015 (UTC)
I've encountered a lot of issues like this. I mostly change the date to 1401 and precision to century. Sjoerd de Bruin (talk) 16:52, 23 June 2015 (UTC)
What... Why does the system even allow that? I'm very confused. --Yair rand (talk) 17:16, 23 June 2015 (UTC)
Theoretically, Sjoerddebruin is doing it the right way. Unfortunately, 99% of the people do not know how to read that, and seen 1400 (or 1401) ±100 as exactly 1400 or exactly 1401, as they forget about the ±100. Edoderoo (talk) 06:54, 24 June 2015 (UTC)
Well, an uncertainty of 100 years would allow the person to be born in a century different from the sourced one. So I believe in the specific case based on the available information, the birth year could better be specified as 1445±45 (i.e. in the interval from 1400 to 1490), or perhaps something more narrow, e.g. 1440±40 if a 'writer' is deemed to be at least 10 years old... Lklundin (talk) 08:47, 24 June 2015 (UTC)
I think this is more correct. What would be better is 1490 -90 +0, since the only actual datum we have is the death datum. I.e. it's not an arbitrary value. But I haven't looked into whether precision can be reflected in this fashion. --Izno (talk) 16:52, 24 June 2015 (UTC)
Interesting change by Sjoerddebruin. I wonder if this should be filled in that way by default. In any case, changing it manually seems cumbersome. And yes, it's hard to display these values correctly in every tool .. --- Jura 15:29, 24 June 2015 (UTC)
We rather have to fix the datetime datavalue. If it provides an upper/lower bound we should also use that. Introducing some qualifier solution might be easier in the short term but won't work for queries etc. -- Bene* talk 06:22, 25 June 2015 (UTC)
The planned query engine should be flexible enough to combine any piece of property with optional qualifier into single predicate to select interested items. In that case it does not matter if the uncertainty is defined in time property itself or somewhere else. As soon as the final solution is provided all current qualifiers might be converted to new syntax using bot. Paweł Ziemian (talk) 17:22, 25 June 2015 (UTC)
The qualifier was created precisely because "we" wouldn't be fixing the datetime datavalue anytime soon. It already has lived way beyond the short term. I think this is something to bear in mind. --- Jura 06:33, 25 June 2015 (UTC)

User:Sjoerddebruin wrote "I've encountered a lot of issues like this. I mostly change the date to 1401 and precision to century." If Sjoerddebruin is using the user interface, I'm afraid this isn't right, because the user interface does not provide access to the before and after fields. If one uses the user interface to enter 1401 and precision century, what ends up in the database is a date of "+1401-00-00T00:00:00" (whatever that means), precision century, before = 0, after = 0. So this means a malformed date +/- zero centuries.

A valid entry would be "+1401-01-01T00:00:00" for the time, precision century, before = 0, after = 1, but there is no way to enter that with the user interface. Jc3s5h (talk) 20:23, 25 June 2015 (UTC)

I'm just doing the best I can, it's always better than the previous value. Sjoerd de Bruin (talk) 20:25, 25 June 2015 (UTC)

Abbots and their abbeys

Do we have a property for abbots as "abbot of" or did they "manage" their abbies? Same thing for "Priest of xxx parish" and so forth. Do we have a religious modelling center where I can ask q's like this one? Thx Jane023 (talk) 07:47, 24 June 2015 (UTC)

Why not doing it same as for bishops - like in Heinrich Mussinghoff (Q66618) use position held (P39) with abbot (Q103163) and with the qualifier of (P642) link to the monastery. To further specify that position, additional qualifiers like start time (P580) or replaces (P1365) can be used. Ahoerstemeier (talk) 13:50, 24 June 2015 (UTC)
Thanks, I'll do that. Jane023 (talk) 05:11, 25 June 2015 (UTC)

Commons category under Statement or Other sites ?

If there exists a commons category (not to be confused with a commons page) for a given item, then it is technically possible to add this category as a Statement as well as an entry under 'Other sites'. In browsing random items one comes across some that have their commons category listed as a Statement, others that have it under 'Other sites', and some that have the category specified twice, one in each place. What is the preferred place to specify the Commons category? Thanks. Lklundin (talk) 08:58, 24 June 2015 (UTC)

Preferred is both, but sometimes that isn't possible, because the commonscat is also linked to a category item. In that case only a statement is sufficient. A sitelink without a statement is incorrect and should have the statement added as well. Mbch331 (talk) 09:34, 24 June 2015 (UTC)
I still don't think commons categories should be linked to anything other than items about categories. This was also, by the way, the guideline on Commons itself. Otherwise list it as a statement with Commons category (P373). Jared Preston (talk) 09:36, 24 June 2015 (UTC)
Since there sometimes isn't a clean 1:1-relation between a category on Commons and a subject on Wikipedia, I would say that you are free to use both. There is no category on Commons for the former Swedish municipality of Junsele. But there is a category covering the same area. Since a category on Commons cannot be added to more than one item here, you have to use statements in those cases. -- Innocent bystander (talk) 09:38, 24 June 2015 (UTC)
Thanks for all the responses. If there is a single editor with any patience left, then I would appreciate a comment, whether this edit reflects a correct understanding of the policy. Thanks again. Lklundin (talk) 13:32, 24 June 2015 (UTC)
@Lklundin: yes, that's great! Jared Preston (talk) 15:08, 24 June 2015 (UTC)
@Lklundin, Innocent bystander: There is an RfC in the pipe for this: Wikidata:Requests_for_comment/Catégorie_commons_P373_et_"Autres_sites", I'll translate it in a minute. TomT0m (talk) 20:58, 24 June 2015 (UTC)
@Lklundin: why add -and duplicate- it, if it is already there in Other sites? Eldizzino (talk) 02:19, 25 June 2015 (UTC)
It seems that there is yet another model: adding it as qualifier in some statements. Sample: Special:PermanentLink/223491092#P119 --- Jura 05:36, 25 June 2015 (UTC)
A consequence of the lack of Arbitrary Access again? -- Innocent bystander (talk) 05:51, 25 June 2015 (UTC)
Nah, I don't think this is a good solution and afaik we already had a discussion on this. Please don't add information twice to an item. Commons categories only belong into the dedicated statement and other pages only in the sitelink section. We might decide to switch this around (adding commons categories as sitelinks and galleries etc. as statements) if that helps (would be a break of concept though) but duplicating the data is worse. -- Bene* talk 06:20, 25 June 2015 (UTC)
I don't think it does anything if you add it as a statement. Most of the time it doesn't even link to Commons. You can't even check if it works. Really useless. --- Jura 06:38, 25 June 2015 (UTC)

API usage

I just don't understand the basics. Maybe there are explanations but I didn't find them. Simple problem (result as json):

  • I've got the Wikidata ID and want to get the Imdb ID.
  • I've got the Imdb ID and want to get the Wikidata ID.

Example: https://www.wikidata.org/wiki/Q42 . I don't want a complete export of all data of this item but only the part "property":"P345","datatype":"string","datavalue":{"value":"nm0010930","type":"string"}" --88.76.208.17 10:35, 24 June 2015 (UTC)

@88.76 Is this of any help to the first part? -- Innocent bystander (talk) 11:04, 24 June 2015 (UTC)
For the second part, no easy way exists yet. You can query Wikidata for those values using SPARQL ([13]) but there will be better ways to query those, see phabricator:T99899. -- Bene* talk 11:11, 24 June 2015 (UTC)
For the second part WDQ API is much easier: STRING[345:"nm0010930"]. --Lockal (talk) 11:18, 24 June 2015 (UTC)

Thank you. So this would mean using 2 different APIs, the "default" one (if this is the right word) and WDQ. --88.76.208.17 11:40, 24 June 2015 (UTC)

Small note: both WDQ and SPARQL API are based on WikiData dumps, so results can be a few hours old. phabricator:T99899 is about implementing a real time solution. --Lockal (talk) 12:02, 24 June 2015 (UTC)

Links linked to the wrong article

This issue is common in relation to the Arabic Wikipedia (ar). A general Arabic article would get linked to a specific English article (I'll explain), and so a visitor to the general English article wouldn't find the article in Arabic linked in the Languages section.

A specific example would help me both settle my issue, and illustrate it: the article Cardamom is the general article about cardamom, where most people would generally end up while looking for that subject. The article Elettaria cardamomum is what I called a 'specific article'. It's a specialized article under the general 'Cardamom'.

The Arabic Wikipedia has a page for هال (cardamom), but not for Elettaria cardamomum. Yet the Arabic page is linked to Elettaria cardamomum instead of Cardamom.

I wanted to remove the link from Elettaria to be able to link it to Cardamom, but I got the error that I can't remove a link unless the linked page has been deleted or merged. Neither of these is the case, this is just a case of wrong linking.

The two items in question are cardamom-yielding plants and Elettaria cardamomum

Please advise.


Please ping me on my English wiki talk page upon replying --Hexafluoride (talk) 11:56, 24 June 2015 (UTC)

Digging into the history of the specific case, @Brya: help?

In the more general case, you have to ask yourself the question: what is the subject of the article? The general/specific distinction is more or less useless in this case. --Izno (talk) 17:05, 24 June 2015 (UTC)

We are dealing here with three kinds of concepts
  • Cardamon, the spice
  • cardamon-producing plants, as a group
  • the individual species of cardamon-producing plants
The page هال has a taxobox claiming that its topic is Elettaria cardamomum, so it is not surprising that it is linked there. Of course, should this be false, the taxobox could be removed (apparently the introduction does match that of "cardamom-yielding plants"). No idea about the technical aspect. - Brya (talk) 17:24, 24 June 2015 (UTC)
@Brya: A lot of Arabic Wiki articles are un-Wiki, be it in formatting, or provided information. The هال is talking about all three concepts at the same time. It seems like it started as Elettaria cardamomum, but further below (and in the intro) it mentions Elettaria & Amomum (cardamom-producing plants), then it goes on to name various uses, both as a spice & "medical benefits" (again, almost always, in this and any similar article, w/o citation, and often old wives' tales or 'popular factoids')
In this context, the extent of Elettaria mentioning is the intro, and the taxobox (plus the Genera أنواع الهيل وتوزعه section). The article is definitely not going in depth into Elettaria. The rest is a mix of uses as a spice, supposed medical benefits, and then how to preserve the spice (in that section, it continues to mention various folk medicine recipes and uses) Hexafluoride --(talk) 18:52, 24 June 2015 (UTC)
Hexafluoride, simply move the sitelink, if you think this is more appropriate. And you should remove the taxobox. --Succu (talk) 19:07, 24 June 2015 (UTC)
It seems that the ar link has been removed (it's my first time using Wikidata, so I don't know my way around). Nonetheless, most articles in the other languages seem to direct to the general 'cardamom' article. Maybe there was a move in the the original English Cardamom page at some time that caused this?
And should the whole taxobox come down, or just remove the genus & species (similar to the taxobox in Cardamom)? --Hexafluoride (talk) 19:48, 24 June 2015 (UTC)
Oh, I don't see that the 'taxobox' at enwiki is doing any good. Just remove it. - Brya (talk) 10:50, 25 June 2015 (UTC)

Recently deceased at Wikipedia

"A new report lists people categorized as dead in Wikipedia, but lacking Property:P570. It's updated several times per day. There are about 20 to 30 new entries. There is currently some backlog from the (English,) Japanese, Arabic, Polish and Ukrainian Wikipedia. Items updated first in Wikidata wont appear in the report. These are listed directly in Wikidata's recent deaths".

Thanks to Pasleim for their help to set this up. --- Jura 17:16, 24 June 2015 (UTC)

On a related note, I've been using Autolist to generate the list of people classified in the en.wiki category en:Category:Living people but having a date of death (P570) statement on Wikidata. There are roughly 800 remaining. Pichpich (talk) 17:50, 24 June 2015 (UTC)
It might be worth auto-categorizing them in Wikipedia based on property P570. cawiki does that. --- Jura 19:41, 24 June 2015 (UTC)
I think the result would actually be catastrophic because in most cases the error is from Wikidata. In many instances, you see that the date of death was added to Wikidata based on faulty Persondata on en.wiki so what you suggest would complete the circle of errors. Pichpich (talk) 20:23, 24 June 2015 (UTC)
Well, once the initial check is done, maybe some sort of a maintenance category could help. Thanks for working through the backlog for enwiki. It's mostly gone now. --- Jura 04:40, 25 June 2015 (UTC)
BTW, once the information is added to Wikidata, the 2nd last column on the list (sample Special:PermanentLink/223651231) indicates how many days it took between the death for the information to appear in Wikipedia. (0 = same day) --- Jura 04:40, 25 June 2015 (UTC)
Thanks Jura for this very efficient tool to fix dates in wikidata. Would it be possible to indicate the number of items in the table somewhere at the top ? so that we have an idea of the task to do ? :)
I guess the grey background is for non-latin alphabets - Is it correct ? --Hsarrazin (talk) 19:04, 25 June 2015 (UTC)
Currently it outputs quite a lot of statistics to its edit history. If you watchlist it, you will see it regularly. We could add some of it to page text as well.
Yes, the background is based on the alphabets. We can refine it further. --- Jura 20:43, 25 June 2015 (UTC)
@Hsarrazin: I will try to add Template:DR rd numbers to the report's header. --- Jura 14:07, 27 June 2015 (UTC)
Thanks, there are less and less latin-alphabet articles :) - I can do the cyrillic ones, but it would be nice if arabic and asian contributors could do the other languages… I'm never sure if google-translate is ok or not. I recently got an example where GT said "male", when the person was obviously "female", judging by her photograph :D
Also, Jura, do you think you can modify the display format of P570 to match the one for the birth date : it would be easier to sort on death date :D --Hsarrazin (talk) 14:59, 27 June 2015 (UTC)
Sort on the column for P570 should already work. As the conversion used in P569 sometimes breaks, I'd rather display the actual values for P570. {{#property:P570|from=Q123|format=YYYY-MM-DD}} doesn't work yet ;)--- Jura 15:17, 27 June 2015 (UTC)

Mark Help:sources for translation

Please can someone mark the Help:sources page for translation ? Thanks. Snipre (talk) 18:21, 24 June 2015 (UTC)

  Done WD:Translators' noticeboard is better place for such requests though. Matěj Suchánek (talk) 19:26, 24 June 2015 (UTC)

Wikidata for listing zoo inhabitants

So, I was thinking, I wonder how many zoos have fennec (Q131564) exhibits, but I can't really find out with Wikipedia, so what about on items about zoos, we list what animals the zoo has, then later, a query could be ran to find out the answer to my question. We probably need a new property for this, though I don't have any idea how general the name would be (i.e. "houses" or "species housed"?). Thoughts? --AmaryllisGardener talk 20:57, 24 June 2015 (UTC)

Really? Adding names to all inhabitants of a special zoo (Q43501)? Wikipedia loves Zoos? How to source that and maintain this in time? --Succu (talk) 21:11, 24 June 2015 (UTC)
I think without a doubt the data needs to be on Wikidata somewhere, you have a better way? Wikipedia does it, but queries are possible at Wikidata, unlike Wikipedia. About the sources, zoo websites almost always have their species listed. Maintaining it? It shouldn't change that much. Sure this requires ambition but there are plenty of other things we're doing that are more ambitious. --AmaryllisGardener talk 21:15, 24 June 2015 (UTC)
Knut (Q159697), an inhabitant of Berlin Zoological Garden (Q154828), died in 2011. What do you think: how many unnamed exhibited insects died at Berlin Zoological Garden (Q154828) in 2011? --Succu (talk) 21:44, 24 June 2015 (UTC)
While there can still be a maintenance problem, I suppose AmaryllisGardener means listing the species a zoo houses, not all individual animals. Usually a zoo has more than one ice-bear (or none at all), so it will not change that often. Bever (talk) 22:44, 24 June 2015 (UTC)
Correct. --AmaryllisGardener talk 15:30, 25 June 2015 (UTC)
A related issue is how to store animal populations of regions, like how many sheep there are in Australia. Presumably we don't want to have one property for each species, and using qualifiers could be unwieldy. Perhaps once we have unit support we could just say "contains (animal population): 4 polar bears"? Is that outside the scope of how quantity/units are supposed to be used? --Yair rand (talk) 02:04, 25 June 2015 (UTC)
I would think in that case you would just have "Resident Animal": polar bear (Q33609) with qualifier quantity (P1114): 4 Popcorndude (talk) 02:30, 25 June 2015 (UTC)
Haha, I like the idea of a "polar bear" unit. However, I think the qualifier as suggested by Popcorndude should do it as well and might be more intuitive. -- Bene* talk 06:13, 25 June 2015 (UTC)

Proposed at Wikidata:Property_proposal/Place#resident_animal. --AmaryllisGardener talk 22:35, 3 July 2015 (UTC)

Complex arbitrary access

Is it possible to use arbitrary access with x2 or x3 depth + not knowing item IDs? For example, I am on wikipage of Q663846 and i need to get name of the pre-previous asteroid – Q149085. So, there is necessary to use double arbitrary access. I tried something like:
{{#property:P155|from={{#property:P155}}}} and
{{#property:P155|from={{#invoke:Wikidata|pageId|from={{#property:P155}}}}}}, but these doesn't work :)) --92.115.106.250 21:38, 24 June 2015 (UTC)

You may use Lua. But this is probably a bug--GZWDer (talk) 03:27, 25 June 2015 (UTC)
I don't think #property: can output Qid. It could use a few additional options. --- Jura 04:02, 25 June 2015 (UTC)
Is it necessary to use arbitrary access at all in such cases? If you are in the Wikipage of Q663846, you can get the relevant label and sitelink of the item related to P155. This is not more complicated than to get the Sitelink and Label of Berlin, when you are in the article of Germany. It's when you need to know the mayor or the coordinates of Berlin, you need arbitrary access. -- Innocent bystander (talk) 05:57, 25 June 2015 (UTC)
Sorry, did not see "pre-previous"! -- Innocent bystander (talk) 05:59, 25 June 2015 (UTC)

Century dating—localization

Speaking of century dating (as a couple of recent sections have):

How and where can such output be localized? Writing on behalf of Latin Wikipedia, on item Q542518 concerning Property:P569, the output, remarkably enough, is "12. century". That's kind of a hybrid between a German form and an English one. But Latin would really like this to look like "saeculum XII" (or even "saeculum 12"). (Well, Latin would really, really like "saeculo", which is ablative case, but that might be too much to ask, given the many possible uses for such output.) Lua is not implemented on Latin Wikipedia, so invoking a module call is not a good solution here. Is there anywhere to fix this, either here or at translatewiki? StevenJ81 (talk) 14:06, 25 June 2015 (UTC)

+1 - in french too we use a rather different notation, since we use roman figures for centuries "20. century" would be "XXe siècle" for instance…
I remember asking the same exact question something like a year ago, without any answer :/ --Hsarrazin (talk) 19:02, 25 June 2015 (UTC)
It looks like Module:Roman (Q14813761) converts 12 -> XII. -- Innocent bystander (talk) 09:45, 26 June 2015 (UTC)
Thank you. But:
(a) That doesn't help us with century–saeculum–siècle, and
(b) I already said above that Latin Wikipedia is not running Lua, so suggesting a Lua module doesn't help there. StevenJ81 (talk) 13:57, 26 June 2015 (UTC)
Not using Lua is going to make your job hard using Wikidata, regardless of this specific issue. You should consider having your wiki revisit that decision.

As for this specific case, phab: or WD:Contact the development team are better places to provide this request. --Izno (talk) 14:12, 26 June 2015 (UTC)

@Izno:
  1. I don't necessarily disagree with you, but Latin Wikipedia is a small wiki, and I don't know that we have people who can manage it.
  2. I didn't think I was making a request; I thought I was asking a question. Are you telling me that the capability I am asking about doesn't exist? Or are you telling me that probably only the developers know how to get at the answer to the question? StevenJ81 (talk) 15:47, 26 June 2015 (UTC)
Reply to #1! That is a problem for almost every smaller wiki. On svwiki, am I only aware of one single user with the capacity of creating good modules. That is why we on smaller wikis are thankful that also modules are in CC-BY-SA. -- Innocent bystander (talk) 15:58, 26 June 2015 (UTC)
Point taken. But you still have to have users that can manage modules that other people create. And we don't necessarily have those, either. StevenJ81 (talk) 17:24, 26 June 2015 (UTC)
The people who know without some research whether it can be done are the same people who look at those particular places, and they rarely poke into this page unless they're asked to or have some broad piece of info to convey to a lot of people. My presumption, based on the fact that Hsarrazin gave it a +1, is that one cannot presently internationalize this text, and in that case asking a technical person (in effect making a request to do so) is the right direction to go. --Izno (talk) 17:44, 26 June 2015 (UTC)
OK. I'll do that. Thanks. StevenJ81 (talk) 18:12, 26 June 2015 (UTC)

Module on the spanish Wikipedia

Hi, I translate the fr:Module:Infobox/Descriptif course cycliste in es:Módulo:Infobox cycling race report. To get informations on article, I have fr:Modèle:Infobox Descriptif course cycliste on the french Wikipedia and I create es:Plantilla:Infobox cycling race report. But there is a little problem, and it dosn't work in the article es:Tour de Francia 2015 when I write {{Infobox cycling race report}}.

Can somebody solve the problem ? I need to do test to see where are the other problems of translation. My goal is to have a common infobox, to work together. Jérémy-Günther-Heinz Jähnick (talk) 10:12, 26 June 2015 (UTC)

The problem is stated on the page when using the infobox: "Error en la secuencia de órdenes: no existe el módulo «Infobox»". That means that it wants to use the module infobox, which doesn't exist. I checked the French module Infobox and the Spanish interwiki leads me to es:Módulo:Ficha. And that gives the syntax how to invoke an infobox. Mbch331 (talk) 12:37, 26 June 2015 (UTC)
@Mbch331: Thank you, the first problem is solve. Now, I will continue my tests (I see I must create the function build). Jérémy-Günther-Heinz Jähnick (talk) 12:51, 26 June 2015 (UTC)

Translate wiki markup

I can use the tag translate wiki markup in user pages or not? reasons? --นคเรศ (talk) 12:27, 26 June 2015 (UTC)

It depends of what do you mean with "the tag translate wiki markup". --Stryn (talk) 16:52, 26 June 2015 (UTC)
Yes, i thing it. --Nakare✝ (talk) 22:23, 26 June 2015 (UTC)
@นคเรศ: Of course it is possible, but it's nonsense to flood Wikidata with user page translations. Use {{TranslateThis}} instead. Matěj Suchánek (talk) 07:37, 27 June 2015 (UTC)

Null in computing on en.wikipedia.org

Q543287 appears to have dedicated pages on several other languages' wikis, but on the English Wikipedia, "Null in computing" is just a section of the overall "Null" disambiguation page. How can this be handled? The data form won't let me specify a section of a page as the entry for an item. --23.252.53.30 12:49, 26 June 2015 (UTC)

@23.252.53.30: The simplest is to create the article on enwiki.
A more complex and unfinished is to participate to an ongoing project : WD:XLINK.
The third is a hack : create a page on enwiki, a small article or a soft redirect, to link it to the wikidata item, then (if you want) change this new page to a redirect. TomT0m (talk) 13:19, 26 June 2015 (UTC)
I did the hack. --23.252.53.30 14:01, 26 June 2015 (UTC)
The problem with the 3rd approach suggested by TomTom is that it can continue to point to a non-existent section even when an article gets created under a slightly different title. Worse we could end up with two items at Wikidata about the same concept.
It's in the nature of Wikidata that not necessarily every Wikipedia has an article on every item.
A 4th approach could be to develop tools that improve navigation between items. --- Jura

I did the simplest and most useful: create an article.

Any tool that can detect more such cases? Items that exist in de,es,fr,it,nl,pt,ru,sv but not in English? Eldizzino (talk) 23:57, 26 June 2015 (UTC)

Links in detective(Q842782) and private investigator(Q1058617)

Hello, I wanted to add a link between https://fr.wikipedia.org/wiki/D%C3%A9tective and https://ru.wikipedia.org/wiki/%D0%94%D0%B5%D1%82%D0%B5%D0%BA%D1%82%D0%B8%D0%B2_(%D0%BF%D1%80%D0%BE%D1%84%D0%B5%D1%81%D1%81%D0%B8%D1%8F) , and I got a message: "The link ruwiki:Детектив (профессия) is already used by item detective (Q842782). You may remove it from detective (Q842782) if it does not belong there or merge the items if they are about the exact same topic." It looks too difficult for me, but maybe somebody else is more talented than me to solve that problem. Cheers --Rene1596 (talk) 14:51, 26 June 2015 (UTC)

@Rene1596: Usually you just have to merge items, but now it seems to be rather an interwiki conflict (continue on that page). Matěj Suchánek (talk) 07:41, 27 June 2015 (UTC)
Thank you. I post my request overthere. --Rene1596 (talk) 10:57, 27 June 2015 (UTC)

API error?

"VIP's labels" tool just failed, with the error "nameGuzzler.js : Error : {"servedby":"mw1232","error":{"code":"badtoken","info":"Invalid token","*":"See https://www.wikidata.org/w/api.php for API usage"}}". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:14, 26 June 2015 (UTC)

Just tested it on Jasper Publie (Q7000286) and had no problem. Which item did you try to update, which languages did you select and which value did you fill in for the label? Mbch331 (talk) 16:23, 26 June 2015 (UTC)
I tried a couple; but its working now. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:44, 26 June 2015 (UTC)

Change to Help:Label

I just added the following para to the lead of Help:Label

Note that an item will have multiple labels in different languages. Labels in
different languages may be unrelated to each other (one language uses a
person's birth name as a label and another uses their pen name). Including
aliases means an item can have multiple labels in the same language. If
there is something important about a particular name then that name
should be in a statement as well as in the label, even if this looks like the
same information is being given in two different ways - we have a lot of
different properties for names of items.

OK? Filceolaire (talk) 01:47, 27 June 2015 (UTC)

Looks fine! But one major difference is also that you can add a qualifier and a reference/source to a statement, but today, never to a label or alias. A statement can also include more information, like the language of a name. And that a label in a language is written in a certain way, should not be regarded as a stated fact. -- Innocent bystander (talk) 07:44, 27 June 2015 (UTC)
This is the perfect text that proves how broken labels are. Thank you for proving my point. GerardM (talk) 07:45, 27 June 2015 (UTC)
I agree that the current labeling system is quite primitive, but it will take some effort to make it better. As a first step I want to see how external identifiers are handled with their own datatype, because labels should be handled in a similar way, with the added problem that they might be multilingual or monolingual.--Micru (talk) 08:14, 27 June 2015 (UTC)
The automated descriptions cobbled together by Magnus is clearly superior. It is just that discussion about this subject is largely refused.
There is no problem with labels. Except that they are very much all too primitive. This is a conscious decision but wrong. Thanks, GerardM (talk) 10:08, 27 June 2015 (UTC)
To me, this addition just confuses "label", "alias" and "name". --- Jura 11:28, 27 June 2015 (UTC)

Download as PDF

Hello, the option 'Download as PDF' is not what you see, is what you get (wysiwyg). The output is in two columns and it is ignoring tables.
Sample: Wikidata:Mineralogy task force/Native element identifiers
Regards --Chris.urs-o (talk) 03:30, 27 June 2015 (UTC)

Commons linking

Hi, why is there no link to the c: project to link to?06:15, 27 June 2015 (UTC) Lotje (talk) 06:16, 27 June 2015 (UTC)

There are a number of properties by which to do so, as described on c:Commons:Wikidata, if that's what you mean. Gabbe (talk) 06:29, 27 June 2015 (UTC)
As for the sitelinks, it's also possible to link to Commons directly, but under the "Other sites" heading, rather than the "Wikipedia" heading. Gabbe (talk) 06:34, 27 June 2015 (UTC)
Hi, Gabbe, adding a link to c: like I did a second ago, is that okay? Thank you for your time. Lotje (talk) 11:08, 27 June 2015 (UTC)
Regarding this edit, the description I linked to (c:Commons:Wikidata) says "By convention, article-like items should only be sitelinked to gallery pages on Commons, and category-like items should only be sitelinked to category pages." I suspect that means that you shouldn't add links to categories on Commons through the sitelinks like you did, but using the property Commons category (P373) instead. Gabbe (talk) 13:35, 27 June 2015 (UTC)

Lower and upper uncertainty

Hello, for Cosmos Redshift 7 (Q20172932), I wanted to add incertainty on the redshift. From this article, page 6, one sees thatthe redshift is 6.604 +0.001 -0.003. Is it possible to specify this uncertainty or is it only possible, for now, to specify similar lower and upper uncertainty? Pamputt (talk) 11:32, 27 June 2015 (UTC)

Not possible now. Snipre (talk) 15:16, 28 June 2015 (UTC)

Proposal to separate w:en:comfort women from w:ja:慰安婦

  • The term comfort women is a literal translation of the Japanese term 慰安婦. But comfort women and 慰安婦 are used in different meaning. 慰安婦 means military prostitutes in any countries and comfort women means mainly military prostitutes in Japanese Empire. For this reason, the contents of w:en:comfort women is nearer to the contents of w:ja:日本の慰安婦 ( Comfort women in japan) than the contents of w:ja:慰安婦. So I propose to separate comfort women from 慰安婦(Q713540) in WikiData.NiceDay (talk) 07:41, 28 June 2015 (UTC)
I agree and it sounds like there should be no Japanese link at all, because comfort women were not prostitutes. Feel free to delete that sitelink. In English, the word prostitute refers to a woman who chooses to sell herself, whereas the comfort women were forced. It's a philosophical question of course whether or not all prostitutes were forced at some point to sell themselves, but the case of the WWII comfort women is pretty different than the typical prostitute in that they were generally raised in good homes and were not poor at the time that they were taken into custody. Jane023 (talk) 11:32, 28 June 2015 (UTC)
I disagree. In 2014,The Asahi Shimbun which had insisted on forcibly takeing of "comfort women" from the infantility and strongly, withdrew the news. And then, there are no media or scholar who insist on forcibly taking in Japan. I know that there are some media and scholars other than Japan claims the forcibly taking, however they are not all. What is important is that this is a topic including disputes. So we cannot treat this topic definitely.[14] And at least, they obtained money.[15]IP58xv (talk) 18:22, 28 June 2015 (UTC)
I am not sure what you disagree with. The problem here has nothing to do with the controversy, but just making sure that the concept in one language is the same concept in another language. It gets confusing when the names are the same but the meaning behind the names is different. Leave the naming problems to Wiktionary and just keep to the meaning here. If they mean the same, keep the sitelink, and if not delete the sitelink. Jane023 (talk) 07:25, 30 June 2015 (UTC)

Munsö, Munsö and Munsö, help wanted!

I have split the B&C-tem Munsö (Q2149064) into Munsö (Q10590356) and Munsö (Q20477542). This was necessary, since some articles were about the (former) island, some about the village and some about both. (svwiki had two articles) The interwiki-conflict is solved (I hope), but I now need some HELP with the labels and descriptions in es, fr, nl and it. -- Innocent bystander (talk) 14:43, 28 June 2015 (UTC)

Why do you use three items for 2 concepts ? If you split use the former as item for the island only and the new one for the village or the inverse. Now we have a third item which a mix and can't be classified. Snipre (talk) 15:13, 28 June 2015 (UTC)
@Snipre: Thanks for the edits! Ask the Wikipedia-editors, why they write articles about two things in one article! -- Innocent bystander (talk) 15:20, 28 June 2015 (UTC)
It is not my problem to know how WP articles are written, but now I have a problem with your splitting choice: this creates an item which is not a village or an island. WD can't follow the structure of WP articles and the creation of a new item was correct to distinguish between village and island. But I would prefer a solution where the former item would be defined as the island even if some WPs used it for a mix of two different concepts. This will be problem of WPs to correct their article or to use arbitrary access to the item of the village. Snipre (talk) 15:34, 28 June 2015 (UTC)
I agree: two concepts -> two items. Where to put the mixed pages is something of a puzzle, but nothing out of the ordinary. - Brya (talk) 15:44, 28 June 2015 (UTC)
With Arbitrary Access we have no problems to follow the links in Munsö (Q20477542) to reach the data about the village and the island. -- Innocent bystander (talk) 16:20, 28 June 2015 (UTC)
The opposite is more difficult. How are we supposed to know that you should add the infobox-data for the (former) island, in an article about the village or vice versa? And how are we supposed to know that you should do that in the enwiki-article but not in the svwiki-article. If Q219937 can be described as instance of (P31):duo (Q10648343), why cannot the item about the two Munsö? -- Innocent bystander (talk) 17:55, 28 June 2015 (UTC)
Munsö (Q20477542): instance of (P31)geographical object (Q618123)? Paweł Ziemian (talk) 18:19, 28 June 2015 (UTC)
@Innocent bystander: If you can't define which is the correct item, don't do the choice: just let a message in the talk page of the WP article that WP contributors have to deal with two concepts. The problem is the creation of items which mix concepts: at the end people will just continue to use these items and will add statements to this one instead of using the correct items or worst they will add data to both items leading to constraints solving in WD side. To be correct you should now forbid the addition of statement to the mixed items in order to prevent statement addition at the wrong place. Can we do that ? Snipre (talk) 07:26, 29 June 2015 (UTC)
The data-model I am thinking of here is:
If statement PXXX in QYYY then
    use that statement
else
    If there is a P:"has parts" in this item, then
         look for PXXX in those child-items, and if they exists, use them both in the infobox
    end
end
If there for example is a P625 in the first item, you do not have to look into the other two items. You normally do not have two Coord-templates in such case. P625 in all three items maybe look redundant here, but it helps the Lua-code to know that is has found what it has been looking for. We are smart enough to know that Bonnie and Clyde wasn't born twice, and that they are not both female and male. But the infoboxes can most likely not see the difference between that two dates has been added because we are unsure about the exact date or if it's because it's about two different persons. If the statements can be found in two different items, the chance that the infobox understands is much higher.
If the first item do not have a P1082-statement, it will look for such statements in the other two items, and the code is then able to use the P1082-statements for both the island and the village. They are normally not the same for an island and a village, but both are normally present in the infobox. The data for this village is updated every 5 years in this case, but I do not know how often the data for the island is updated.
There are cases when you have a use for two P625 above. If the article is not about a village and an island, but instead about two (or more) neighbour villages, you can have use for more than one P625 and should not add P625 to the B&C-item.
Another problem I see here is the integrity of our database. The bots who will update these items will look for the statement P776:S0457 and add P1082 according to it's own database. It will also change rank from "preferred" to "normal" in the already present statements. The bots may be smart enough to look for such things as applies to part (P518), but I do not trust that they always are that smart. There are so many potential ways to mix different data, that I prefer that we avoid it as often as possible.
On svwiki we have hundreds of articles about two or three villages. nlwiki tends to have articles about them too, but they split them in one article per village.
Yes, Snipre, there is a large risk that random users adds statements that should not be here. But I think the advantages of protecting the integrity of the database is higher. -- Innocent bystander (talk) 08:29, 29 June 2015 (UTC)

Wikidata weekly summary #164