Wikidata

Wikidata:Project chat/Archive/2018/01

< Wikidata:Project chat‎ | Archive‎ | 2018
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

problem with "quantity" datatype for properties....

Hi,

all properties with quantity datatype have the type indicated in greek in fr-interface as can be seen on this screencapture !! how is it possible to correct this, please ? I could not find the page that allows translation from "quantity" :( --Hsarrazin (talk) 14:14, 3 January 2018 (UTC)

exact... thanks Jura :)
what bothers me is translatewiki has been corrected almost 2 weeks ago [1] but I still get wrong input in greek (el) instead of fr. Purging MediaWiki:Datatypes-type-quantity/fr did nothing... - is there a lagging problem, or is it something else that I cannot see ? --Hsarrazin (talk) 16:16, 3 January 2018 (UTC)
@Hsarrazin: Translations have to be deployed like code changes, and there were no deployments over the holiday period. If I’m not mistaken, the next Wikidata deployment should happen in a few hours (20:00 – 22:00 UTC, see the “MediaWiki train” in wikitech:Deployments#Wednesday, January 03). --Lucas Werkmeister (WMDE) (talk) 17:12, 3 January 2018 (UTC)
Seems to be fixed now :) at least I now see «Quantité» on Property talk:P4714?uselang=fr --Lucas Werkmeister (WMDE) (talk) 12:11, 4 January 2018 (UTC)
This section was archived on a request by: Lucas Werkmeister (WMDE) (talk) 12:11, 4 January 2018 (UTC)

Bahamas Leaks

Could a Japanese or Chinese speaker say whether Bahamas Leaks (Q27021541) and Bahamas Leaks (Q27449990) should be merged, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:58, 3 January 2018 (UTC)

  Done Google translate indicates that they are the same subject matter, with document counts, and dates.  — billinghurst sDrewth 03:11, 4 January 2018 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 07:36, 5 January 2018 (UTC)

Deboosting scholarly article (Q13442814)

I've received a number of requests to apply same de-boosting in search rankings as we do for disambiguation pages to items that are instance of (P31) scholarly article (Q13442814). I think it makes sense, since these articles, by their nature, match a lot of terms, but usually they are not the item one looks for when searching for things (i.e., if you look for "dog", you'd probably expect stuff related to dog (Q144) on top and not all scientific articles mentioning dogs, at least not on top). Thus, I plan to apply some negative adjustment to their rankings. Note that this would not remove them from search results, just make them rank lower. I would like to hear if anybody sees any problems or can bring any objections to this? If there's no strong opposition, this will be done somewhere in January, once the holiday deployment freeze is over. Smalyshev (WMF) (talk) 21:47, 21 December 2017 (UTC)

  Support Pamputt (talk) 22:02, 21 December 2017 (UTC)
  Support Mahir256 (talk) 22:29, 21 December 2017 (UTC)
Sorry, searching for "dog" (I switched to enUI) gives no top results marked as scholarly article (Q13442814). --Succu (talk) 22:45, 21 December 2017 (UTC)
"dog" is likely no good example. "heart rhythm" was one example that I suggested that led to this change. In the new search (http://wikidata-wdsearch.wmflabs.org/wiki/Special:Search) it didn't show in the top 500 items at the time. In general it's often hard to find items for technical terms that are similar to "heart rhythm" when the items don't have much use. ChristianKl ❪✉❫ 01:43, 22 December 2017 (UTC)
Then maybe implementing search options like "-scientific article" are a better way to follow. --Succu (talk) 20:07, 22 December 2017 (UTC)
I agree, having such option would be nice. But: a) it's not possible to use it in completion suggesters and b) implementing an option will take time (we'd need to do some refactoring work on underlying parser, probably) and boosts are available right now. Smalyshev (WMF) (talk) 20:14, 22 December 2017 (UTC)
Sorry, Stas but I have no idea what „boosts“ are? --Succu (talk) 22:50, 22 December 2017 (UTC)
@Succu: Changes in search score (and thus ranking) of certain items according to certain rule (in this case, value of instance of (P31)). Smalyshev (WMF) (talk) 22:52, 22 December 2017 (UTC)
And how does that help users searching for scientific articles? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:26, 21 December 2017 (UTC)
Can you give specific queries of how you see users seeking scientific articles? ChristianKl ❪✉❫ 01:21, 22 December 2017 (UTC)
Andy, that was pretty indelicate. :-/ I would think that a statement along the lines, of "good initiative, though we also need to look to how people wishing to search for scientific articles, or other entity based search, are able to perform searches more functionally to get results." Would get a response of "good point, let us see what we can do" rather then "sit and spin Captain Grumpy". I support Smalyshev's suggestion though do share Andy's concerns.  — billinghurst sDrewth 13:17, 22 December 2017 (UTC)
The problem with even your reformulated version is that it contains no concrete ideas of what kind of queries Andy has in mind where this would be a problem. If you do share his concerns, do you have specific queries in mind? Having specific examples would be helpful for discussing the effects. ChristianKl ❪✉❫ 19:43, 22 December 2017 (UTC)
I have no idea of Andy's desires in the area. If you are asking what I consider possible, well freethinking how about biographical article (Q19389637) for George Washington (Q23)? Do I pine for it now? No. Is it a particular search we would like to be able to be run internally, I would think so if we want to have described by source (P1343) filled with our sources. PS. I didn't add examples as I didn't think that it was that great a leap of invention.  — billinghurst sDrewth 01:55, 23 December 2017 (UTC)
Basically your idea is that articles about George Washington should be ranked higher than people called George Washington? Even without any deranking of scientific articles, George Washington Lane (Q5545995) that's currently on place 100 ranks higher than any scientific article. Searching scientific articles on Google Scholar is much more efficient than doing so on Wikidata. On the other hand if you searh the item for a person on Wikidata who shares it's name with 30 other people you are dependent on the Wikidata search engine. ChristianKl ❪✉❫ 15:29, 23 December 2017 (UTC)
No mate, you miss my point. Sometimes I think that you need to stand back a little further and see what is being said. On this forget it, I am not in the mood.  — billinghurst sDrewth 21:25, 24 December 2017 (UTC)
The fact that I take your answer as being an answer to the question I asked, doesn't seem to me as needing to take a step back. I asked for examples because I can't think of examples where I think it's valuable to not downrank the scientific articles. ChristianKl ❪✉❫ 19:55, 26 December 2017 (UTC)
Unsurprisingly, Smalyshev's reply was far more civil than you suggested it would be. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:35, 24 December 2017 (UTC)
It wouldn't help when you search for scientific article, of course, the article would be appearing lower in ranking. The ranking in general is the game of tradeoffs - you frequently find many more results that you can display, and of course only one can be the first, so you need to try and guess what the user means. One approach is to optimize for the common case - which means the users with an uncommon case would have it slightly worse, and the users with common case for have it slightly better. As we add more syntax to it (which will happen eventually, at least for fulltext search) it would be easier to tell the engine what you are looking for, but there will always be tradeoffs. So what I am trying to do is to find a set of tradeoffs with more optimal summary usability. Smalyshev (WMF) (talk) 20:30, 22 December 2017 (UTC)
  Support but @Smalyshev (WMF): I think a better general solution rather than de-boosting rank of particular P31 values is to look at the length of the label: shorter labels that contain the search string would in general be more likely to match what the searcher was looking for, and scientific articles (among other less likely matches) generally have much longer labels. Though maybe that should apply only with the main label and not aliases (which can be just abbreviations). Another option I'd like to see: allow an advanced search option that lets you deliberately filter by including or excluding common P31 values (Q5, Q13442814, ...), similar to how search can now be filtered by namespace. ArthurPSmith (talk) 14:27, 22 December 2017 (UTC)
Interesting idea, I will check if this is feasible. Smalyshev (WMF) (talk) 20:14, 22 December 2017 (UTC)
I would be in favour of such advanced search options that allow to combine the text search and/or completion suggestors with elements of the semantic graph. Not sure deboosting is the best approach for the specific issue at hand — label length and perhaps something closer to tf–idf (Q796584) would seem more promising, especially if they could be tuned by the users as they work on different things at different points in time (e.g. on cardiology-related topics today, on matching papers tomorrow). --Daniel Mietchen (talk) 22:34, 1 January 2018 (UTC)
  •   Support. Or we should change the search algorithm so that a search needs to match a certain proportion of keywords in the title of a scientific article for it to float onto the top of a result... Deryck Chan (talk) 14:54, 22 December 2017 (UTC)

Erroneous request for merging

Hello. I have just been requesting an item for merging, but I found out that this was a mistake. Any suggestions to what I can do? The item is Q12320419, which I tried to merge into Q1857341. In my attempt I ended by pressing a Postpone-button - I am not quite sure, what this does, but I supposed it put the request in some sort of queue, and eventually the merging takes place.

I am sorry for the problems, I've created. --Amjaabc (talk) 10:16, 27 December 2017 (UTC)

Christmas Holiday (Q1857341) describe a film from 1944. Are you sure you want to merge Christmas break (Q12320419) into it? --Pasleim (talk) 13:05, 27 December 2017 (UTC)
"I found out that this was a mistake. Any suggestions to what I can do" Mateusz Konieczny (talk) 16:57, 27 December 2017 (UTC)
As you can see, this button will do nothing with those items. It is supposed to save the item id for later but I'm not sure whether it still works. Matěj Suchánek (talk) 09:31, 28 December 2017 (UTC)
Sure, but @Sjoerddebruin: doesn't seem agree. --Liuxinyu970226 (talk) 04:23, 2 January 2018 (UTC)

Please remove items

Please remove all speculative items at Star Wars Episode IX: The Rise of Skywalker (Q20977110). The principal photography is only set to begin in half a year! Maybe it never will, maybe people will leave the project or even be dead by then. Wikidata should contain definite data and not speculations! --88.71.165.225 16:02, 29 December 2017 (UTC)

Done a little pruning and left a note on the item's talk page. It is just one of those things, planned/contracted/actuality <shrug> all can be argued to be valid.  — billinghurst sDrewth 13:48, 1 January 2018 (UTC)

AGPL vs. AGPL

Happy new year! Now I have a question which asked last year but still haven't any answers, specifically about those items:

Both enwiki and ptwiki are having separated articles, describe different histories of different AGPL, but in many other languages they say both are same, with just a potential purchasing from GNU, to the best of my knowledge merging any two or more of them won't be helpful, and unlikely be possible, so how do I do something for them? --Liuxinyu970226 (talk) 01:44, 1 January 2018 (UTC)

As I mentioned at WD:IC, I temporary marked second as Wikimedia permanent duplicate item (Q21286738). --Liuxinyu970226 (talk) 01:45, 1 January 2018 (UTC)
Affero General Public License, version 1.0 (Q27017230) is the same as Affero General Public License (Q28130012), as there is zero chance of another version of the original non-GNU Affero GPL. IMO they should be merged.
Likewise GNU Affero General Public License, version 3.0 (Q27017232) is the same as GNU Affero General Public License (Q1131681). In this case, there is a chance of another version of the license being established, and in addition there is often legal wording like "version 3.0 or later" vs "version 3.0" (and not later), so there could be a case for Q28130012, Q27017232 and Q27020062 being considered to be separate items covering slightly different concepts. But I find that to be too much extra complexity without any real world use cases to justify it.
All Wikipedia articles will be talking mostly about the (unversioned) GNU AGPL, as that is the name of the 'most commonly used' concept in this area, and the history of GNU AGPL is included in those article. John Vandenberg (talk) 16:44, 2 January 2018 (UTC)

Excel file as a source

If a source is an excel file:

1) How can we use file format (P2701)? With Microsoft Excel (Q11272)?

2) Is there a property to show in which sheet the data is?

Xaris333 (talk) 03:36, 2 January 2018 (UTC)

Excel is a software and not a file format. The item on excel lists the various file formats it can deal with under readable file format (P1072). ChristianKl ❪✉❫ 11:00, 2 January 2018 (UTC)

Calendario Gregoriano

In English: publication date 4 April 1881 Gregorian

In Greek: ημερομηνία δημοσίευσης 4 Απριλίου 1881 Calendario Gregoriano

Calendario Gregoriano is now Greek. How can I change it?

Xaris333 (talk) 04:05, 2 January 2018 (UTC)

→translatewiki:MediaWiki:Wikibase-time-calendar-gregorian/el. Seems that somebody experimented or chose wrong language. Matěj Suchánek (talk) 13:44, 2 January 2018 (UTC)

Wikidata weekly summary #293

 
Here's your first Wikidata Weekly Summary of the year 2018 (Calendar: Gregorian)!
  • Development
    • Fixed a problem with references in history of items (phab:T182767) Thanks to Matěj Suchánek who helped providing a workaround
    • Make statements on forms persistent for lexicographical data (phab:T163724)
    • Fix a bug that removed the collapse button (phab:T175492)
    • Remove cache constraint check results on purge (phab:T182107)
    • Add sitelinks to hif.wiktionary (phab:T180785)
    • Read constraint check results from cache and check freshness (phab:T182106)
    • Re-label the "Save" links to "Publish" (phab:T161367)

You can see all open tickets related to Wikidata here.


Quality control and constraints on subsets

I've been working on improving quality mostly related to paintings. Quality has different aspects. I'm currently looking at completeness. For example every painting (Q3305213) should have a creator (P170), should be in a collection (P195), has at least one creator (P170), should have inception (P571), etc etc. One way to do this is to set constraints on certain properties, but this only works if you happen to have a property that covers your interest. Because the numbers are rather large, I focus on subsets. For example Wikidata:WikiProject sum of all paintings/Wiki monitor/nlwiki only looks at painting items that have a link to the Dutch Wikipedia. This makes it more humans scale. You can actually finish it and keep the quality standard for this subset. The sum of all paintings category contains many more examples of subsets.

This all doesn't scale very well. Wouldn't it be nice to have some sort of tool to assist in this? Some sort of next generation constraints or Recoin? As a user I would like to set a focus area (in this case instance of (P31) painting (Q3305213)). The tool would offer property coverage and intersections. A user could work on a subset and add the missing data. The next step would be to be able to flag subsets with things like "I'm working on this", "cleared it" or maybe even "needs bot import". A user could come back later or get a report of all the cleared subsets which need some attention (like Listeria updating the report page).

Do you think this is feasible? Would this be something to work on in 2018? Multichill (talk) 16:56, 30 December 2017 (UTC)

Why doesn't listeria do the job? ChristianKl ❪✉❫ 18:06, 30 December 2017 (UTC)
It's all static, it doesn't scale. You have to make a report yourself for everything. Multichill (talk) 11:38, 31 December 2017 (UTC)
I have the impression that working through work lists takes a lot more time than creating worklists. Sure it would be possible to make the creation of worklist easier, but I would expect that will result in a lot of work lists with undone tasks. ChristianKl ❪✉❫ 11:53, 31 December 2017 (UTC)

This Recoin idea would be nice to have for Wikidata:WikiProject Women too. Items should have a few basic human (Q5)-related properties, plus some guidelines about occupation (P106), especially for women born before 1850 who could not be professionally trained but whose occupation is still more than "daughter", "wife", "mother", or "noble". After working on various women items I am starting to see some trends, but it is slow-going by hand. It would be nice to be able to generate some specific reports on the fly for certain occupations. Jane023 (talk) 11:57, 31 December 2017 (UTC)

Ping Ls1g about Recoin. @Multichill what you want to have looks a lot like the features provided by SHACL. It is an RDF vocabulary done in order to express constraints on the RDF graph. It allows to express things like "all the nodes that are results of this SPARQL query must have exactly one value for property X, may have values for property Y that are integers between 0 and 10 and must not have values for property Z. It could be nice to have a tool able to evaluate such constraints on Wikidata using the Query Service. Andrawaag have already done some investigation on using such constraints systems with Wikidata. Tpt (talk) 22:00, 31 December 2017 (UTC)
We had some thoughts on building a tool for a similar purpose, so I am very curious to understand what you have in mind. What we were thinking of was something like shown on the right: After the user has specified a class, a set of properties that should be present, and a set of facets, the tool essentially shows the distribution of completeness wrt. the chosen properties along the chosen facets (mock shown on the right, inspired from Listeria examples).
 
Mock Wikidata Profiling Tool
That should work with Dutch paintings or humans of gender female as well, though to do anything beyond instanceOf, I guess a bit of SPARQL might be needed.
Does that have some similarity with what you have in mind? What do you refer to with "intersections"?
Cheers, Ls1g (talk) 19:37, 2 January 2018 (UTC)
@Ls1g: intersection (Q185837) let's say all painting items are a set. The set of paintings that have an article in Dutch is a subset. The set of paintings that don't have a creator (P170) is another subset. These two combined (painting with article in Dutch, but no creator (P170)) would be an intersection. I want to do arbitrary intersections and keep track of them.
With these intersections you get subsets of a size that humans can handle. Multichill (talk) 20:15, 2 January 2018 (UTC)

Search anomaly "nap"

When I type "nap" in the top right search box, neither nap (Q901586) nor nap (Q5242962) appears. This is annoying - I came close to creating a duplicate item because I could not find the one I was looking for. Can the search team explain what's happening here? Thanks! - PKM (talk) 19:41, 1 January 2018 (UTC)

Neapolitan (Q33845) has the language code "nap" and therefore the alias nap. It has 76 sitelinks while the nap (Q901586)/nap (Q5242962) have less and only three statements each. The other hits also have "nap" as aliases.
The new full text search does a more resonable job at finding the right items: http://wikidata-wdsearch.wmflabs.org/w/index.php?search=nap&title=Special:Search&profile=default&fulltext=1&searchToken=7m0erkmodnwn7r4kfya1yik51
https://www.wikidata.org/wiki/Wikidata:Suggester_ranking_input is a page for adding searches where the top right search box does a bad job. I invite you to add this example and any further you encounter to that page.ChristianKl ❪✉❫ 11:00, 2 January 2018 (UTC)
Thanks for that explanation! Will do. - PKM (talk) 20:03, 2 January 2018 (UTC)

Importing data about electric vehicle charging stations

Yesterday I tried out the Query:

https://query.wikidata.org/#%23defaultView%3AMap%0ASELECT%20%3Fitem%20%3Fcoord%20%0AWHERE%20%0A%7B%0A%20%20%3Fitem%20wdt%3AP31%20wd%3AQ10373548.%0A%20%20%3Fitem%20wdt%3AP625%20%3Fcoord.%0A%7D

which gives me a world map of Whisky Distilleries.

I was wondering why the Query https://query.wikidata.org/#%23defaultView%3AMap%0ASELECT%20%3Fitem%20%3Fcoord%20%0AWHERE%20%0A%7B%0A%20%20%3Fitem%20wdt%3AP31%20wd%3AQ2140665.%0A%20%20%3Fitem%20wdt%3AP625%20%3Fcoord.%0A%7D

starting from https://www.wikidata.org/wiki/Q2140665

does not work the same. I found something like: https://commons.wikimedia.org/wiki/Category:Electric_vehicle_charging_stations_in_Germany

But nothing for official list of stations e.g. https://tank.rast.de/emobility/

I would love to work on import for such data if that makes sense. So

  • How much sense does the idea make?
  • How to proceed?

--WolfgangFahl (talk) 16:36, 2 January 2018 (UTC)

Any import of missing information from realiable source is always appreciated. Matěj Suchánek (talk) 19:06, 2 January 2018 (UTC)
This is certainly useful information to import, assuming the source includes coordinates. I don't know how complete OpenStreetMap is, maybe openstreetmap:Tag:amenity=charging_station helps. Nemo 20:09, 2 January 2018 (UTC)
The query does not work because there are no item about electric vehicle charging station (Q2140665) with coordinate location (P625) in Wikidata.
Data about charging station in France (IRVE file): https://www.data.gouv.fr/fr/datasets/fichier-consolide-des-bornes-de-recharge-pour-vehicules-electriques-irve/ (decree).
20:39, 2 January 2018 (UTC)

Changing data type

Property:P627 has a data type of "string", but it should probably be "external-id" instead. I'm not that familiar with Wikidata -- what is the correct venue for requesting this change? --Ahecht (talk) 18:44, 2 January 2018 (UTC)

There is a disagreement that this should happen, see Wikidata:Identifier migration/0#Not going to convert. It's not the first time this question is raised. Matěj Suchánek (talk) 19:04, 2 January 2018 (UTC)
It was also discussed at Wikidata:Project_chat/Archive/2015/10#About_Property:P627. --Jarekt (talk) 20:05, 2 January 2018 (UTC)

How to merge

Diagonalisierung (Q3025885) and Diagonalisierung (Q1208191) can be merged. How do I do this? Naively I would add Q1208191 to Q3025885 and then add ask for deletion for Q1208191. Thanks --Nobelium (talk) 11:19, 9 January 2018 (UTC)

Try Help:Merge.
--- Jura 11:25, 9 January 2018 (UTC)
  Done --Liuxinyu970226 (talk) 11:32, 9 January 2018 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 18:23, 9 January 2018 (UTC)

Population property and census item

Hello. I need a property to use as a qualifier to connect census item with the value of the population of a place. A user suggest statement is subject of (P805), but it is not a valid qualifier for population (P1082). I don't want to use stated in (P248) as a source. Xaris333 (talk) 00:10, 3 January 2018 (UTC)

@Xaris333: Why not? P248 seems like the right way to do this. --Yair rand (talk) 00:16, 3 January 2018 (UTC)
I don't want the census item to be the source. I have add an online url as a source. stated in (P248) make a mess the source in wikipedia articles. Plus now it shows both the sources. Moreover, the census item can not have many properties that need as a property. For example, archive URL (P1065). I know that this is wikipedia problem, not wikidata. But, do we really want to have both sources? reference URL (P854) and stated in (P248) for the same statement? I am confused. Xaris333 (talk) 00:22, 3 January 2018 (UTC)

Can I use stated in (P248) and reference URL (P854) to the same source? [2] Xaris333 (talk) 22:12, 3 January 2018 (UTC)

I do that frequently. Example at League of Cognac (Q47014250). - PKM (talk) 00:31, 4 January 2018 (UTC)
Great. Thanks. I though we had to create two different sources. Xaris333 (talk) 02:26, 4 January 2018 (UTC)

Constraints about villages

I am trying to add statements about a village but some constrains are not clear to me. Ayios Therapon (Q4831306)

1) head of government (P6) "value requires statement constraint. Christodoulos Milonas should have a statement position held (P39)." That means I have to create an item like "Ayios Therapon President of Communal Council" and add it to the person item? A lot of work, maybe useless for small villages.

2) executive body (P208) --> Municipal Councilor in Huy (Q29385989)

"distinct values constraint This property's value must not be present on any other item, but is also present on the following items:..."

That means I have to create an item like "Ayios Therapon Communal Council" and add this item to the property? "Ayios Therapon Communal Council" -->instance of (P31) --> Municipal Councilor in Huy (Q29385989). A lot of work, maybe useless for small villages.

Xaris333 (talk) 12:52, 3 January 2018 (UTC)

Not necessarily - you can use the property of (P642) as a qualifier, like it is done for the bishop Marc-Antoine Berdolet (Q152944). Then only thing needed for your village heads would be a generic item "President of Communal Council". Ahoerstemeier (talk) 00:09, 4 January 2018 (UTC)
Thanks. That solves the first problem. But not the second one. Xaris333 (talk) 02:25, 4 January 2018 (UTC)

Interlanguage links with anchors and redirects

According to Help:Sitelinks#Interlanguage links with anchors I guessed it is not allowed here to add interlanguage links with anchors, but what about redirects? There are already a lot of added here, so one can do this or one should avoid it? See for instance de-wiki links on Q43304118 or Q29924639. Florentyna (talk) 17:15, 3 January 2018 (UTC)

Such links are not allowed yet but if you change an already connected page to a redirect, nothing changes in Wikidata (unless you do cleanup). So the item should be treated as if there was no link (the link can be eventually removed and the item may even get deleted).
See also Wikidata:Requests for comment/Allow the creation of links to redirects in Wikidata. Matěj Suchánek (talk) 17:47, 3 January 2018 (UTC)
As far as I understand the notability policy, it says that redirect don't create automatically notability but not that they "shouldn't" or "mustn't" be created.
The current majority community opinion as shown in the open RfC suggests that the community believes links to redirects are a good idea. If you currently want to create a redirect you have to first create the page as a non-redicrect page. Then you come to Wikidata and add it as sidelinks and then you change the page to be a redirect. On the other hand, you can't simply set a sitelink to a redirect page. ChristianKl ❪✉❫ 02:15, 4 January 2018 (UTC)

Renaming a maintenance category

Posted by Niridya on the French Project chat in English. I copy here his message. Tubezlob (🙋) 21:15, 3 January 2018 (UTC)

Hello everyone,

I think it should be a great idea to rename the category Category:Wikidata:Deletion to Category:Candidates for speedy deletion because it will be the same name than in others english Wikis.

I post this message because I done it today but my modifications was deleted and the wikipedian said me to reach an administrator consensus.

Cordially.

I don't think our Wikidata category equals what counts as Candidates for speedy deletion on EnWiki and see no good reason for renaming it. ChristianKl ❪✉❫ 21:41, 3 January 2018 (UTC)
Hi, the reason I see is to harmonize names between all English-speaking wikis. --Niridya (talk) 22:02, 3 January 2018 (UTC)
BTW Wikidata:Bistro is meant for the French-speaking, not the English-speaking.
--- Jura 22:08, 3 January 2018 (UTC)
Given that the deletion system isn't the same there's no good reason for it having the same name. ChristianKl ❪✉❫ 01:54, 4 January 2018 (UTC)

Change on the editing interface: save becomes publish

Hello all,

Following the change that have been happening since 2016 on all the wikis of the Wikimedia movement, the “save” button will be replaced by “publish” on Wikidata.

The main reason for this change is to avoid confusion for new editors, who need to be clearly aware that what they add on the wiki is immediately published online. (more information here).

On Wikidata, this will mainly have an impact on the editing interface of the items (blue link next to the field one's currently editing). On any other page like the project pages, discussion pages, etc. the button has already been changed to “publish changes”.

 

The change will be deployed on January 3rd. In the meantime, you can try it on Beta.

If you speak another language than English, please consider helping translating in your language. You can find the translations here and here (change the language code at the end of the URL to find or add a language).

Once the change is made, the documentation will need to be updated. Feel free to help updating the screenshots and other mentions of the former “save” link.

Thanks a lot for your understanding, Lea Lacroix (WMDE) (talk) 20:05, 28 December 2017 (UTC)

Edit: it's now done. Please let us know it anything displays wrong on your side. Lea Lacroix (WMDE) (talk) 11:56, 4 January 2018 (UTC)

Is it worth contributing to pywikibot?

I considered submitting a patch for pywikibot, but digging in the source code confirmed by concerns about the state of this project, to the point that I am not even sure it is worth trying to fix anything there. I have summarized my concerns in my userspace, if anybody is interested in that. I think WikidataIntegrator is more promising and I encourage people to use and contribute to that instead. It is still a young library but at least the foundations look saner. Any thoughts? − Pintoch (talk) 14:45, 29 December 2017 (UTC)

@Pintoch: I've used both pywikibot and WikidataIntegrator; the latter was definitely much more in line with what I expected a python library to do, as you outline. I've also attempted to contribute code to pywikibot and ran into trouble getting their tests to run. The data model has problems handling floating point and other special data types correctly relative to what wikidata expects - I was trying to fix that. But I'm sure there are users who do a lot with it, it does generally sort of work. I think it may be trying to do too much to handle both wikidata structure and normal wikipedia pages, which is where the problems stem from. ArthurPSmith (talk) 17:09, 29 December 2017 (UTC)
@Pintoch: I've been using and contributing to Pywikibot for over 10 years now. It's a collection of scripts to do things and a library. It's used and written by people who want to scratch an itch. The old version (the now gone compat version) was still using screen scraping. The current version (core) is API based. The switch between the two and the double effort was a huge burden on the developer community. After that was over, some people thought it was cool to have a pypi version. We ended up with two branches again and nobody maintaining the Pypi branch so we killed it and switched it to release every once in a while releasing when all tests are green. I'm not happy with the pypi situation at all. I don't think anyone is actively maintaining it right now. I've never actually used it myself. I always use the git version and tell other people to do the same.
I've seen many times people coming up with big plans to split up Pywikibot, to rewrite it or just generally complaining. In the end it was usually just a waste of time that could have been spent much better on improving what we have. So yes, contributions are very much appreciated. If you run into things, file tasks in Phabricator and tag pywikibot-wikidata. Please do contribute code. Small easy to digest patches work best. Also code review could use a hand.
This all helps the many people running Pywikibot. Multichill (talk) 14:37, 30 December 2017 (UTC)
@Multichill: I don't dispute the value of pywikibot for many people, but I am questioning its usefulness as a Python library. What I personally need is a Python library to interact with Wikibase. I tried to explain in my note why pywikibot really isn't a satisfactory library yet, and why I think it would be a lot of effort to transform it into a proper library. It's fine to build things that are not libraries - but they should not be advertised as such, because then people invest time into it before they realize they cannot use it for what they need.
I feel bad about complaining without proposing any patch - but that is because after investing a lot of time working with pywikibot (both for Wikipedia and Wikidata), I have just given up: the foundations are broken beyond repair as far as I can tell.
Maybe it's just a cultural issue - the Wikimedia community does not use Python much and does not value the standards from the Python world. In that case, that's obviously fine - take my rant as an explanation of why these two communities differ and why you are losing contributors who have the same sort of expectations as I have. − Pintoch (talk) 09:03, 31 December 2017 (UTC)
Pywikibot is not only Wikidata, unlike WikidataIntegrator. It can work with multiple wikis at once (eg. Wikidata, Wikipedia and Commons). Matěj Suchánek (talk) 14:53, 30 December 2017 (UTC)
Sure - it would be great to have a Python library for the MediaWiki API in general, not just for Wikibase. − Pintoch (talk) 09:03, 31 December 2017 (UTC)
What do you think would be needed to have WikidataIntegrator do what you want? ChristianKl ❪✉❫ 00:56, 2 January 2018 (UTC)
@ChristianKl, Pintoch: As the "inventor" and one of the main developers of Wikidataintegrator, the intention always was to create something which resembles a database driver for Wikibase/Wikidata which integrates the API and the SPARQL endpoint, ignoring most of the other parts of the MediaWiki framework API, so if there are useful features, just let me know about them, we will see if how/these can be implemented, or just submit a pull request. Sebotic (talk) 10:25, 4 January 2018 (UTC)
@Sebotic, ChristianKl: I have only used WikidataIntegrator a few times so I am not very familiar with it yet. From what I have seen, the library tries to provide a convenient high-level interface, but I am not sure all concepts of the Wikibase data model are accurately represented. It looks like users are expected to manipulate dictionary-based representations directly - I have not seen any classes for the various concepts of the data model, for instance (like statements, qualifiers or references). I think a more object-oriented architecture would be nice. Something like WikidataToolkit for Python would be ideal. − Pintoch (talk) 11:45, 4 January 2018 (UTC)
I've added a reply at User talk:Pintoch/Issues with pywikibot. John Vandenberg (talk) 16:31, 2 January 2018 (UTC)
Thanks, I have replied there too. − Pintoch (talk) 11:45, 4 January 2018 (UTC)

How to improve the usefulness of Wikidata descriptions?

The latest discussion about disabling showing Wikidata descriptions alongside enwp content is going on at en:Wikipedia:Village_pump_(proposals)#RfC:_Populating_article_descriptions_magic_word. I'm mentioning this not to encourage people to !vote in it, but because there's an interesting discussion there about a sample of 10k descriptions that @DannyH (WMF): assembled. In particular there are concerns about the usefulness of the descriptions (there's also a ranking going on with a sample of 1k descriptions at en:User talk:Pbsouthwood/Wikidata description Quality Assessment experiment), and concerns about how long vandalism takes to be reverted here. I'd encourage reading through the comments posted there.

Any ideas how we can fix these problems here? I've been looking into using Huggle here, but it doesn't seem to work (phab:T183141) - are there other anti-vandalism tools? (m:2017 Community Wishlist Survey/Wikidata/Better countervandalism tools didn't make it into the top 10 sadly.) Or are there ways we could systematically (usefully) improve the descriptions? Thanks. Mike Peel (talk) 07:19, 30 December 2017 (UTC)

I am still not convinced that we have a problem with Wikidata descriptions at all. They work as desired (disambiguation of items with identical labels) and they are intentionally very short, thus I do not agree with the rating scheme invented on the page of User:Pbsouthwood. Display of Wikidata descriptions in mobile apps and search fields was invented later and one can of course question whether the descriptions as they are used inside Wikidata are useful for that task as well. Personally I think that it is a no-brainer to find that they are, but I understand that others may have a different opinion.
On vandalism: filtering the RC stream to patrol changes is indeed an important problem, and we would be happy if more editors were engaged in counter-vandalism activities. There are some tools which provide filter options, such as Pasleim’s reCh tool (login, then filter for terms and "en" language code) or User:Yair rand/DiffLists.js; the ORES markers for potentially critical edits are useful as well. RC editors are also encouraged apply for the rollback right to work efficiently. —MisterSynergy (talk) 07:55, 30 December 2017 (UTC)
See https://www.wikidata.org/wiki/Wikidata:WikiProject_Counter-Vandalism for more about our counter-vandalism efforts. ChristianKl ❪✉❫ 08:40, 30 December 2017 (UTC)
Looks like another car-crash engineered by the anti-Wikidata cabal™ on en.Wikipedia. That said, one of the issues raised is the discussion is that of Wikidata editor notes in descriptions (the example given being "determined sex of an animal or plant. Use Q6581097 for a male human"). It is in our gift to fix that, and we should do so promptly. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:23, 30 December 2017 (UTC)
There was a separate property for usage notes (Wikidata usage instructions (P2559)), but since they aren't shown when you want to add information on Wikidata, there currently is no other option than to use the descriptions. Mbch331 (talk) 12:05, 30 December 2017 (UTC)
The example in question can be easily fixed by merging male (Q6581097) as I proposed on the talk page with is very benefitial for other concerns as well. ChristianKl ❪✉❫ 14:09, 30 December 2017 (UTC)
There are many other examples of Wikidata-specific instructions in descriptions. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:37, 4 January 2018 (UTC)

Monuments

list of monuments (P1456) link to the list of heritage monuments in the place/area. If a place has only one monument? We can't have a list. Xaris333 (talk) 22:43, 3 January 2018 (UTC)

Why not? I had such list in Latvian WLM lists. --Voll (talk) 08:49, 4 January 2018 (UTC)
We can have lists out of one monument, no problem.--Ymblanter (talk) 15:40, 4 January 2018 (UTC)

Format of Polish cultural heritage register number (P3424): reference in a Polish cultural heritage register


Format of LAU (P782): identifier for a local administrative unit, renamed from NUTS 4 and NUTS 5. Format: 2 letters followed by digits

Geocodes of Cyprus

In Cyprus, the CYSTAT (Q28863790) have statistical codes of municipalities and communities constitute a 4-digit coding system. It called them geographical codes (geocodes). They are national codes. Should I propose a new property for these code or there is another way to add the data to wikidata? Xaris333 (talk) 00:31, 4 January 2018 (UTC)

4-digit system, that is 10,000 possible identifiers? If the identifier is truly an "identifier" (unique etc.), I suggest going ahead. Matěj Suchánek (talk) 07:35, 5 January 2018 (UTC)
Yes, they are unique. 4-digit system (only for quarters are 6-digit numbers). See Wikidata:Property proposal/Statistical Service of Cyprus Geocode. Xaris333 (talk) 07:44, 5 January 2018 (UTC)

Signs displaying the name of a place where they are not situated

Can anyone answer Property talk:P1766#Signs displaying the name of a place where they are not situated? Xaris333 (talk) 02:50, 5 January 2018 (UTC)

Inconsistencies between Q937130, Q3540821, and Q788723

Dear Wikidata users,

Being not used to modify Wikidata, i need your help to correct the inconsistencies observed between Q937130, Q3540821 and Q788723. They all refer to articles related to the Tibesti Mountains.

Q937130 refers in 3 languages to the place called Trou au Natron, but redirect in French to the place called Tarso Toussidé. We have in French an article called Trou au Natron, but it is associated with Q3540821 (where French is the only language).

We have in English an article called Tarso Toussidé, associated to Q788723, which refers in most languages to the highest point of this massif, the Pic Toussidé. By consequence, I suggest:

  1. to associate Q937130 only with the term Trou au Natron, remove the incorrect French link, and replace it with the single link from Q3540821
  2. discard by consequence Q3540821 which becomes unused
  3. have Q788723 only related to the Tarso Toussidé (largest massif)
  4. create a new object only related to the articles referring exactly to the Pic Toussidé (the highest point of the Tarso Toussidé).

What do you think of these suggestions?

I thank you for your help --Ndiver (talk) 11:57, 5 January 2018 (UTC)

Q937130: the statements on this item seem to be about "Trou au Natron". Maybe instance of (P31) needs a better value, but coordinates and elevation seem ok. So I agree with you. From a Wikidata point of view, it's important that the sitelinks are added to the item that has the correct statements.
Q3540821 can be merged into Q937130, once the frwiki sitelink deleted from Q937130.
If you know other features of these massifs, you might want to create items for them as well. It's not required that there are Wikipedia articles about these. Thanks for spotting the problem and suggesting a solution.
--- Jura 22:04, 5 January 2018 (UTC)

Civil Rights Memorial

The article for the Civil Rights Memorial mentions the people named on the memorial and the "forgotten" where the documentation of their death was not fully documented at that time. How do I model this? Thanks, GerardM (talk) 14:31, 6 January 2018 (UTC)

Maybe, as a starting bid
⟨ The Forgotten ⟩ subclass of (P279)   ⟨ group of humans (Q16334295)      ⟩
manner of death (P1196)   ⟨ killed in struggle ⟩
significant event (P793)   ⟨ civil rights movement (Q13406660)      ⟩
and
⟨ killed in struggle ⟩ subclass of (P279)   ⟨ manner of death (Q2438541)      ⟩
? --Tagishsimon (talk) 17:43, 6 January 2018 (UTC)

Worldcat Identities

Would it be a good idea to make a property/identifier "WorldCat Identity", linking to the OCLC/WorldCat Identities like e.g. Charles Henry Hull with identity http://worldcat.org/identities/lccn-n85377312/

The WorldCat Identities contain a lot (but not all writers) that have texts in WorldCat with overviews of publications etc.

This is really an important authority control. See: w:en:Help:Authority_control.

I have no idea how to do this, but perhaps someone can create this new property/identifier.... --Dick Bos (talk) 15:06, 6 January 2018 (UTC)

In this case WorldCat says "lccn-n85377312". That means the person with the Library of Congress authority ID (P244) "n85377312". ChristianKl ❪✉❫ 16:40, 6 January 2018 (UTC)
On the English Wikipedia, they use the VIAF identifier to link to Worldcat .. Thanks, GerardM (talk) 17:53, 6 January 2018 (UTC)

Produced sound

I didn't notice it during the property proposal discussion, but it seem to me that the new produced sound (P4733) is redundant to audio (P51) - if not, many (but not all!) values will have to be moved from the latter to the former. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:22, 12 January 2018 (UTC)

@Thierry Caro, ديفيد عادل وهبة خليل 2, Jsamwrites, ChristianKl, ArthurPSmith: from the proposal discussion. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:23, 12 January 2018 (UTC)
The new property has the item datatype, while the other one has the Commons file one. How can it be redundant then? "produced sound" is to link to the item describing the sound something makes. Sjoerd de Bruin (talk) 19:25, 12 January 2018 (UTC)
@Pigsonthewing: audio (P51) is Wikidata property to link to Commons (Q18610173) and the data type is different.Thank you David (talk) 20:05, 12 January 2018 (UTC)
Ah, sorry, I missed that. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:27, 12 January 2018 (UTC)
This section was archived on a request by: Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:27, 12 January 2018 (UTC)

Free image available at url property?

Hi everyone, for the Sum of all paintings I add and update a lot of items about paintings on Wikidata. Often these paintings don't have an image available on Commons, but do have a free image available on the source website. Uploading these images is currently not part of my workflow and I don't have any plans to add it, but I would like to do something with this data.

What I'm thinking of is a new property (of type URL) "free image available at url" in which we can store the direct link to the out of copyright work that can be uploaded to Commons. As qualifiers some data can be stored that would be needed on Commons to properly upload the image. For example for Red Sail in the Harbor at Venice (Q46924693):
The new property -> [3], qualified with:

(exact properties to use here open for suggestions and improvements)

This data combined with the data already on the item should be enough for a (semi-)automatic tool to upload these images to Commons. The tool should:

  1. grab the data from the Wikidata item
  2. do some checking (to prevent non-free works from slipping through)
  3. upload it to Commons using the artwork template with a linkback to Wikidata.
  4. update the Wikidata item with the new image and remove the "free image available at url" property

Currently we have 234.162 painting items of which (only) 70.696 are illustrated. I think we can get at least double that using this approach. What do you think? Multichill (talk) 15:43, 26 December 2017 (UTC)

Yes that would be great to have. Especially if you have the url at hand but no time for the upload (yet). Jane023 (talk) 21:16, 26 December 2017 (UTC)

  WikiProject sum of all paintings has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. some more input would be nice. Multichill (talk) 21:27, 27 December 2017 (UTC)

This is also useful for works not out of copyright but with a public domain date (P3893) in the near future for semi-automatic uploading to Commons once date has been reached. I would propose we don't call it 'free' as it is not clear what that will mean. I would also propose to have a property with the .jpg file (or even the zoom file (.dzi or ImageProperties.xml etc) as to have the direct download of the image available. In your example: http://collections.cincinnatiartmuseum.org/pictiondmz/ump.di?e=4C36D94713293518451249F6419DF088AFA28295C9C09B27F1430AACF2815D73&s=21&se=149022812&f=%5Cd9335%5Cu17993350%5C1915.135_v01_o4.jpg --Hannolans (talk) 21:42, 27 December 2017 (UTC)
Thanks for your reply Hanno,
As for soon in copyright. That would make it impossible for someone to clean it out. Some statements can't be fixed. And what is soon? 1 year? 1 century?
How would you propose to call it? You just said you didn't like the 'free' part, but you didn't propose an alternative
My link was to the direct download, but it's just a crappy website with multiple options in this case so both links seem to return the same jpeg file. Multichill (talk) 16:59, 28 December 2017 (UTC)
Good idea! --Marsupium (talk) 12:01, 29 December 2017 (UTC)
  Support - PKM (talk) 20:09, 29 December 2017 (UTC)
'  Support' - But it needs clarification of what is meant by 'free'. Maybe 'openly-licenced image available' would be more accurate. Xcia0069 (talk) 09:58, 2 January 2018 (UTC)

It would make sense to make a proper property proposal for this to have further discussion. ChristianKl ❪✉❫ 01:10, 1 January 2018 (UTC)

That was of course the plan, but I first wanted people the time to comment on this. The proposal is now at Wikidata:Property proposal/Commons compatible image available at URL. Let's continue over there. Multichill (talk) 11:44, 7 January 2018 (UTC)

Area precision

Tracked in Phabricator
Task T155910

How serious we are to take the data like this [4]? Shoudn't this be just six? Courtesy pinging @MB-one:--Ymblanter (talk) 16:58, 6 January 2018 (UTC)

Or at least 5,99. --Edgars2007 (talk) 17:01, 6 January 2018 (UTC)
The value, that was entered from the source was in this case 5.99 km². These values are a result of this issue with QuickStatement. While the seemingly random addition or subtraction from the entered value is certainly annoying, it doesn't create any problem here, because the deviation in the result is several orders of magnitude smaller, than the accuracy of the value itself. --MB-one (talk) 17:21, 6 January 2018 (UTC)
Thanks. Hopefully it will be fixed at some point, either from the QuickStatement end or from our end.--Ymblanter (talk) 21:09, 6 January 2018 (UTC)
I haven't been successful with my fix... Matěj Suchánek (talk) 09:25, 7 January 2018 (UTC)

Time to split Wikidata:List_of_properties/Terms

The page displays lua overflow error messages - it's just too long and needs a careful split. Apologies for writing here - the issue was raised on appropriate talk page half a year ago, nothing happened. Retired electrician (talk) 19:08, 6 January 2018 (UTC)

Intercardinal direction items - constraint

If an item Y has a property (for example shares border with (P47)) with value item X with direction relative to location (P654)-->south (Q667), then shouldn't item X has the same property with value item Y with direction relative to location (P654)-->north (Q659)? Is that an appropriate constraint? If yes, I don't know how to do it. :)

However, on intercardinal direction (Q15410629) items there are more than one value to opposite of (P461) with opposite direction (Q21012911), orthogonal direction to the left (Q22672535) and orthogonal direction to the right (Q22672531). Can we have the constraint only if the qualifiers is using with shares border with (P47) or other specific properties?

Cases (for shares border with (P47):

Xaris333 (talk) 07:03, 7 January 2018 (UTC)

We can use {{Complex constraint}} and also generate statements with missing qualifiers for QuickStatements. Matěj Suchánek (talk) 09:21, 7 January 2018 (UTC)
Not necessarily true. For instance, Netherland is on the Northwestern side of Germany, however Germany is on the weastern side instead of southeastern side of Netherland. C933103 (talk) 09:27, 7 January 2018 (UTC)
Ok. I understood. It's now always true. So we can do nothing? Xaris333 (talk) 14:36, 7 January 2018 (UTC)

How does Wikidata handle projected data?

Hi all

Is there any information or any examples of how Wikidata handles projected data? Some projected data forms the basis of large decision e.g project sea level rises by a certain date or projected temperature increases by increased atmospheric CO2. It seems incredibly important to have this data in Wikidata especially in relation to Wikidata fed maps and charts on other Wikimedia projects.

Some thing to consider may be:

  • There may be several projections by different groups or studies eg projected extinctions of species due to overfishing or habitat destruction
  • The same organisation may have changed or updated their predictions over time as the situation changes or modelling improves eg actual and projected ozone layer depletion by a certain date before CFC bans came into place

Thanks

--John Cummings (talk) 10:33, 7 January 2018 (UTC)

My first instinct would be to use sourcing circumstances (P1480) as a qualifier but I don't know whether there's an existing way to do this. ChristianKl ❪✉❫ 12:22, 7 January 2018 (UTC)

Idiot's guide to add interwiki link?

Tracked in Phabricator
Task T163713

I edit Wikipedia for more than decade. Links to other language articles were always part of the article text. Never had any problem with it. But now I am thrown inside wikidata project, into some web form and here I fail. I simply do not know what to do, where to click. Nothing seems to work (in Chrome).

Is there guide for a complete idiot, how to add interwiki (and nothing else)? I tried to look it up but failed too. 85.70.87.168 18:32, 4 January 2018 (UTC)


Specifically I tried to connect this page: 

 https://cs.wikipedia.org/wiki/Kronika_trojánská

with 

 https://en.wikipedia.org/wiki/Historia_destructionis_Troiae

On Czech wiki the interwiki web form is different (much simpler) but it gave me error message: "Chyba: $1. Attempted modification of the item failed."

85.70.87.168 18:48, 4 January 2018 (UTC)

The en. article is linked to Historia destructionis Troiae (Q1241142). The cs article is linked to Kronika trojánská (Q12031574). Those two wikidata records need to be merged. (However, for reasons not clear to me, they're currently refusing to merge.) --Tagishsimon (talk) 18:53, 4 January 2018 (UTC)
Duly reported --Tagishsimon (talk) 18:56, 4 January 2018 (UTC)
Kronika trojánská (Q12031574) has edition or translation of (P629) Historia destructionis Troiae (Q1241142), that's why they cannot be merged.--Ssola (talk) 23:17, 4 January 2018 (UTC)
I've added an English label and description to Kronika trojánská (Q12031574). It looks like the cs article is specifically about the 15th-century Old Czech translation. Wikidata is more granular than Wikipedia. In Wikidata, a Latin work and a translation of it are two separate things. - PKM (talk) 02:49, 5 January 2018 (UTC)
They should be interwikilinked, and that is, what Wikidata is supposed to deliver. An article about Harry Potter in any given language will use the names in that language instead of the original ones, and only give the original ones as another information, but nevertheless all those articles should be interwikilinked. Wikidata has to deliver such stuff, it's the main purpose of wikidata to organise this in a manner wanted by the other projects, where WD is used as a database. So how are those articles, that are about the same item (sometimes in different translations) be interwikilinked? And if Wikidata doesn't deliver such basic needs, the stuff it was build for after all, any longer, where to organize this outside Wikidata? Sänger (talk) 08:24, 5 January 2018 (UTC)
15th century is two centuries after 13th century. If you think it's the same for cswiki, you could use local interwiki links.
--- Jura 08:35, 5 January 2018 (UTC)
1997 is not 1998, so why is de:Harry Potter und der Stein der Weisen interwikilinked with en:Harry Potter and the Philosopher's Stone? Sänger (talk) 11:11, 5 January 2018 (UTC)
The translation of a work two centuries after the original version are inter-related though different concepts. The translation has a translator, which the original does not, the language is different; the editing is different, etc. One could put forward a case that for the Harry Potter works should be different, though there is a lot more commonality and central control to these works. Personally, I would have them as separate items, though I am a separationalist when it comes to works and their translations.  — billinghurst sDrewth 12:09, 5 January 2018 (UTC)
So what is your suggestion for the interwikilinking of this articles? How should this be done in Wikidata, the tool used for interwikilinking in the wikiverse? How could this main purpose of wikidata, besides storing data about relevant items from the wikiverse, be solved in a way usable by the main clients of this database, the different language versions of the wikipedia? Sänger (talk) 12:14, 5 January 2018 (UTC)
Hopefully sooner or later the RfC for redirects will be resolved and we will have that tool for providing links in cases like this. Currently, it seems it isn't resolved because WMDE doesn't give feedback. I added https://www.wikidata.org/wiki/Wikidata:Contact_the_development_team#WMDE_feedback_for_the_RfC_%22Allow_the_creation_of_links_to_redirects_in_Wikidata%22 to ask for feedback. ChristianKl ❪✉❫ 14:07, 5 January 2018 (UTC)
@Sänger: The Wikisources face this issue extensively for interwiki between the Wikisources (edition to edition/translation) and from the Wikisources <-> Wikipedias (edition to creative work/translation). At this time the immediate solution is manual interwikis. The Wikisources are looking at how we utilise the edition or translation of (P629) and has edition or translation (P747) links and using lua scripts to put in place interwikis. The evolution of Wikidata has highlighted that interwikis are a looser relationship than what occurs on an item, so the wikis can continue with manual interwikis for loose relationships or have scripts to work it out; it is not Wikidata's responsibility to do that.

I am not in agreement with ChristianKl that redirects are the means to address that, it seems a mongrel of an approach to dupe the system.  — billinghurst sDrewth 03:20, 6 January 2018 (UTC)

@Sänger, billinghurst: for your information, a @Tpt: is now working on a solution to be able to autolink editions of books to the actual work item from wikisource, and this probably could be usable for wikipedia afterwards. In the meantime, manual interwikilinks, or a specific template can use the edition or translation of (P629) info like the one used by sv wikisource. --Hsarrazin (talk) 07:57, 6 January 2018 (UTC)
How could one do manual interwikilinks? I started the discussion and, so far, it gave no answer to my question. I would happily use the traditional method I know. Btw, there's no realistic chance in Czech Wikipedia to have separate article for the original chronicle. Trojánská kronika is our oldest (or one of the oldest) incunabulum and all it got is a small shallow article. 85.70.87.168 18:31, 7 January 2018 (UTC)
You cannot do 'manual sitelinks', beyond the normal mechanism in wikidata of adding the wikipedia page to the wikipedia section of the item record. The issue in this instance is that wikidata allows a wikipedia page to be added only to a single wikidata item, and your cs article is already linked to a wikidata item. Other language articles about a similar subject are linked to a different wikidata item. So you will not get interwiki links from the cs article until this issue is sorted out. --Tagishsimon (talk) 18:37, 7 January 2018 (UTC)
manual interwikis = traditional interwiki links at the source site, here we are talking csWP so [[en:Historia_destructionis_Troiae]]  — billinghurst sDrewth 06:44, 8 January 2018 (UTC)

Film narration in different languages

Hi. I've become more active in Wikidata again, as a result of shifting work from enwiki to frwiki. I'm very impressed by how frwiki is now using Wikidata to generate Film Infoboxes. I have one question: do you think it'll ever be useful or practical to assign a language to the "narrator" field, sort of like we do with film titles where there's a mandatory language field? Canada is an officially bilingual country (English/French) and films with narration often exist in the two languages, with different people doing the narration. I guess it's true for any non-subtitled film with narration that gets versioned internationally. A good example of this would be the Oscar-nominated short doc Q5123702. As you can see from the respective Eng. and Fr. articles (both of which I worked on) there are two notable voice actors, one for each language. (And different script credits too). But if I add the various names into Wikidata currently, it'll seem like Q606553 and Q3435174 jointly narrated the film -- which they did, but in their respective languages. Hope this makes sense. thanks, Shawn in Montreal (talk) 14:08, 5 January 2018 (UTC)

@Shawn in Montreal: The usual way toi treat that kind of case is to create 3 items: one item for the work, one item for the french version and one item for the canadian french version and to link the 2 versions to the work. The work item will contain all data common to all versions of the movies and the other items will contain only the data rlevant to the version they represent. But this system seems to not be compatible with the rules of the Wikidata: WikiProject Movies, so th ebest is to ask the people from that project to answer. Snipre (talk) 19:49, 5 January 2018 (UTC)
@Jura1: Please explain the good way to model data for movies. Snipre (talk) 19:49, 5 January 2018 (UTC)
@Shawn in Montreal: I've found this in the Wikiproject movies documentation. A separate element for each different adaptations looks like a solution to the problem. -- Peuc (talk) 21:48, 7 January 2018 (UTC)

2017 Chile census

The final population data for the 2017 Chile census (Q21001879) are available since December 22 (censo2017.cl, Cantidad de personas por sexo y edad - XLSX format). I manually uploaded the population (P1082) data (but not female population (P1539) and male population (P1540)) for the region of Chile (Q590080) (first-level subdivision). It is necessary to upload the data of the province of Chile (Q1153408) (second-level subdivision) and commune of Chile (Q1840161) (third-level subdivision). Is there a simpler way to load the information than to do it one by one? If it is not possible, could someone help me finish? Thanks you, Metrónomo (talk) 17:59, 7 January 2018 (UTC)

QuickStatements can help you. Matěj Suchánek (talk) 18:17, 7 January 2018 (UTC)
OK. I tried it but it did not give a good result (or I do not know how to use it correctly). The command Q2118 P1540 315014 P585 +2017-00-00T00:00:00Z/9 P459 Q39825 produced this. The command was correct? How do I add the "preferred level"? It is necessary to distinguish it from the old population data. I can not find it in the documentation. --Metrónomo (talk) 18:39, 7 January 2018 (UTC)
I forgot that QS still cannot work with ranks... (we have a bot that can rearrange them, though).
I prefer having the gender-related population as a qualifier of the statements like this: Q1|P1082|2|P1540|3|P1539|4|P585|+2017-00-00T00:00:00Z/9|P459|Q39825|S248|... (it would be nice to include the source as well). Matěj Suchánek (talk) 18:57, 7 January 2018 (UTC)

Rules for linking to WD from WP

Does anyone know what the rules are for linking to wikidata when wikipedia does not have a corresponding article. I have noticed people removing the links, But I cannot find the rule they are following. --RAN (talk) 20:02, 7 January 2018 (UTC)

Probably depends on the language edition of Wikipedia.--Ymblanter (talk) 21:12, 7 January 2018 (UTC)

Dataset on typhoon warning messages issued by authority

See [5] and [6].

At past, in some articles about typhoon in Chinese Wikipedia, there were a list that includes information about all the trihourly warning issued by Taiwanese meteorology office when a typhoon is affecting Taiwan. These information are now being removed from related articles for various reason. Should these data be included into wikidata? C933103 (talk) 03:02, 7 January 2018 (UTC)

How much data points are we talking about per item? ChristianKl ❪✉❫ 12:42, 7 January 2018 (UTC)
Usually a few dozens data point? The one I linked have 14 entry. Each entry include issued time, serial number, advisory/alert type, cyclone center position, radius for Beaufort scale 7 wind, max wind speed near center, max instantious wind speed, predicted position 24 hours after the issued bullet, and also the area covered by the warning/advisory. C933103 (talk) 13:24, 8 January 2018 (UTC)

The Giant Turnip (Q2068935)

In most languages wikipedia version, the wikidata entry is about the original Russian folklore story. However, in Israel, the linked article is about the Israeli adaption of the story. Many of the descriptor for the wikidata item was also about the Israeli adaption instead of the original work, like the claim that the wikidata item subect was originally written in Hebrew. These claims have been removed, but should the hebrew song get separated into a different wikidata item which would contain those specific details that only applies to the Israeli adaption? While doing so would have the advantage of clearly separating the two entries, it would also mean that it would be impossible to interlink between the Hebrew version article and the English version article. Also, there is a Chinese article regarding the Chinese version adaption of the story but it is not linked in the wikidata item here and thus readers are not able to use interwiki link to navigate between different languages. How to fix that?C933103 (talk) 08:57, 7 January 2018 (UTC)

We need separate items here to be as accurate as possible. For wikis it is still possible to use "old-style" interwiki to connect pages from another item.--Jklamo (talk) 17:18, 7 January 2018 (UTC)
Are those old interwiki bots still functioning in this aspect?C933103 (talk) 13:17, 8 January 2018 (UTC)

Books and articles

What's the current consensus on creating wikidata records for books and articles that are used as sources in wikipedia articles? Say, The Dolby Era: Film Sound in Contemporary Hollywood (Q29384930). These works don't have corresponding articles about them, and most of them never will. Reading Wikidata:Notability and Wikidata:What Wikidata is not together suggests that the answer is no - ?? Retired electrician (talk) 18:43, 6 January 2018 (UTC)

Wikidata weekly summary #294

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

Recoin now available as gadget

Dear community,

I am glad to announce that Recoin is now available as gadget. Recoin adds a status indicator to entities which quantifies the extend of information (completeness) relative to other entities of same profession or type. Furthermore, it shows the top 5 absent properties of an entity at the top of the page.

 
Arno Kompatscher: Politician missing basic information

Looking forward to your comments!

Cheers, Ls1g (talk) 19:56, 2 January 2018 (UTC)

That's excellent, Lsig. But would be more useful if the list produced could be used directly to enter into the addition of a missing property. --Tagishsimon (talk) 20:48, 2 January 2018 (UTC)

Good. Where can we translate it? Xaris333 (talk) 00:13, 3 January 2018 (UTC)

@Xaris333: Thanks for the suggestion! We are working on multilinguality and will push the changes soon. Property labels will be translated automatically, few other strings need to be manually translated here. Preview see picture on the right. Ls1g (talk) 16:18, 9 January 2018 (UTC)
 
Preview of multilingual version of Recoin
excellent Ls1g !! a very useful feature would be a button or link to directly add the wanted property :) --Hsarrazin (talk) 12:59, 7 January 2018 (UTC)
@Tagishsimon, Hsarrazin: Looking into this now! Ls1g (talk) 16:23, 9 January 2018 (UTC)
Good work, @Ls1g:; thank you very much. Let's have Recoin on steroids. You might take a look at https://www.wikidata.org/wiki/Wikidata:Project_chat/Archive/2017/12#New_userscript_quickpresets which is a different thing; but presumably Recoin could drive a presentation a little like this for the missing statements? And thank for for Recoin even as it is; it's a very useful UI addition. --Tagishsimon (talk) 17:53, 9 January 2018 (UTC)

toclimit

Are the CSS classes toclimit-1 through toclimit-6 not enabled on this wiki? I tried adding class="toclimit-3" to a div containing __TOC__ in the header template used on Wikidata:Property proposal/Place but it doesn't seem to be working (even though the class is transcluded properly). Jc86035 (talk) 14:24, 6 January 2018 (UTC)

They are in en:MediaWiki:Common.css but not in d:MediaWiki:Common.css. That's what you need.--Ssola (talk) 06:18, 7 January 2018 (UTC)
@Ssola: Thanks. Jc86035 (talk) 10:32, 9 January 2018 (UTC)

Wikidata data model "deficiency"

  Notified participants of WikiProject Companies

The WD data model is more powerful than RDF because of references and qualifiers. But IMHO it's also weaker in that people are less willing to create items than general RDF nodes.

Eg how to model that Hewlett Packard split into HP Inc and Hewlett Packard Enterprise in 2015? I added two "followed by" values https://www.wikidata.org/wiki/Q80978#P156 but it really is ONE event with 1 preceding entity and 2 succeeding entities.

Guess I could make a prop "significant event" but there's no standard property to link the participants (succeeding companies).

And I don't think many people would agree to make out the Event as an Item... and to standardize a vocabulary of event participant roles. --Vladimir Alexiev (talk) 16:12, 8 January 2018 (UTC)

And I guess we have the same issue in the other direction for mergers? I could argue that if a specific corporate event (like a breakup or merger) had sufficient significance in itself it should have its own item. I'm not sure that all such events deserve an item, but the HP example seems like a good one. ArthurPSmith (talk) 16:55, 8 January 2018 (UTC)
Mergers and splits are complicated, but there is no deficiency.
In fact in most cases of mergers there is no new entity, just second company is merged to the first and first change its name. The correct way to represent this is no new item, but record appropriate changes to existing items. Problem is that sometime new wiki article is created for "new" entity. But that is more likely deficiency of wikpedia, not wikidata.
Same for splits, in most cases there is no one old entity and two new entities, just new entity is created and old one change its name. Again the correct way to represent this is one new item for new entity and record appropriate changes to existing item. Again the problem is that sometimes two new articles are created.
That is the case of HP, "Hewlett Packard Enterprise" is really new entity, but "HP Inc." is nothing but renamed "Hewlett-Packard Company". So correct way to record this is rename Hewlett-Packard (Q80978) to HP Inc. (also add appropriate official name (P1448)), leave followed by (P156) and Hewlett Packard Enterprise (Q19923099) but delete followed by (P156) HP Inc. (Q21404084), and at HP Inc. (Q21404084) left just permanent duplicated item (P2959) and Hewlett-Packard (Q80978). But problem it that articles connected to HP Inc. (Q21404084) already exist.--Jklamo (talk) 17:04, 8 January 2018 (UTC)
There are different scenarios:
  • A + B = A
  • A + B = C
  • A = A + B
  • A = B + C
So we need 4 different event names and 4 different ways to model these events. Snipre (talk) 07:58, 9 January 2018 (UTC)
  • As statements about HP before and after might be considerably different, I'm not entirely convinced by the purely formal approach suggested by Jklamo.
    --- Jura 10:46, 9 January 2018 (UTC)

Wikidata in read-only for 30min on January 9th

Hello all,

On January 9th, Wikidata will be moved to a dedicated server, in order to allow more resources for the project, due to its growth. (ticket)

In order to do this move in good conditions, Wikidata will be in read-only mode on January 9th, from 06:00 UTC to 06:30 UTC. (For some places in the world, it will be January 8th: you can check with your time zone here). During approximately half an hour, humans and bots will be able to read Wikidata, but not to edit it.

Thanks for your understanding, Lea Lacroix (WMDE) (talk) 15:37, 3 January 2018 (UTC)

Bumping; This is 9.5 hours from when I write (early tomorrow morning, in Europe; 11:30 in Delhi, India; 13:00 Jakarta, Indonesia; 14:00 in Perth WA; 17:00 in Sydney NSW). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:36, 8 January 2018 (UTC)

How to query old revisions of items?

Hi,

I am trying to programmatically access old revisions of items. The documentation of the linked data interface mentions it is possible to specify a revision parameter, yet for all cases expect the example of Q42/Revision 112 I get an error (either Not Found (example Q42/Revision 111) or Internal Server Error (example Q42/Revision 1)).

Does anyone know why I am getting these errors? Are there any other ways to get old revisions as RDF/JSON/... other than downloading old dumps? In the MediaWikiAPI I did not find revision parameters.

Would be thankful for any hints! Ls1g (talk) 16:52, 9 January 2018 (UTC)

The revision needs to belong to the specified entity: Q42.json?revision=612406093 works. --Lucas Werkmeister (WMDE) (talk) 16:59, 9 January 2018 (UTC)
Thanks! How do I know which revisions belong to an entity? (To explain better, my intention is to make a binary search through revisions until I find one where a specific change I am interested in happened.) Ls1g (talk) 17:14, 9 January 2018 (UTC)
For that you would probably use the regular MediaWiki Action API, e. g. something like on Special:ApiSandbox#action=query&prop=revisions&titles=Q42&rvprop=ids&rvlimit=500. --Lucas Werkmeister (WMDE) (talk) 17:18, 9 January 2018 (UTC)
However, the Internal Server Error is probably a bug – I’ve opened phabricator:T184537 for it.
Great, that solves my problem. Thanks! Ls1g (talk) 17:23, 9 January 2018 (UTC)

Google Knowledge Graph API

I have the Google Knowledge Graph API and a valid key, but can't find the identifier for Kevin Borland (Q40786519). Can someone look it up for me?  – The preceding unsigned comment was added by Richard Arthur Norton (1958- ) (talk • contribs) at 00:47, 2 December 2017‎ (UTC).

To get this thing archived. --Edgars2007 (talk) 15:38, 10 January 2018 (UTC)

How to model 'infant mortality rate'?

I'd like to import the infant mortality (Q835884) data from UNESCO. It's defined as the 'number of deaths per 1000 live births', and the imported data would be added to items for countries (using a point in time (P585) qualifier to add yearly data).

As the definition is so specific, it seems we may need model it with a new 'infant mortality rate' property. The other option is to create a more general 'mortality rate' property, and then use qualifier applies to part (P518) -> infant (Q998) OR infant mortality (Q835884).

I really like this second more general option as we can then use the same property for any other mortality rates of interest. The problem is that 'mortality rate' in itself does not specify exact parameters for how the rate is defined, so we would technically need to add some sort of other qualifier to say for example "per 1000 people" - this seems like it gets very messy!

So I'm leaning towards the specific property for infant mortality rate. Any feedback or other ideas? Thanks NavinoEvans (talk) 13:38, 9 January 2018 (UTC)

The "per 1000 people" could be treated as the unit, (if there's not an item for that now, create one) - I don't think that's a significant problem. We have a related property proposal for subpopulation that may be helpful for this. ArthurPSmith (talk) 14:53, 9 January 2018 (UTC)
I hadn't thought of that, thanks for the suggestion. I'll wait a bit for any other input but that's swayed me back to going for the more general property. NavinoEvans (talk) 23:31, 9 January 2018 (UTC)

License page?

I just set up Wikidata:Licensing because I could not find a page with Wikidata's license presented. I suspect that I am mistaken - does anyone know of an existing license page?

Where is the page which presents Wikidata's copyright license? I see the license in the footer. Is there anything here like en:Wikipedia:Copyrights or Blue Rasberry (talk) 20:46, 9 January 2018 (UTC)

I thought there was a page, but I only found Wikidata:Copyright as an interwikilink of en:Wikipedia:Copyrights. Not much, but hey… ;-) —MisterSynergy (talk) 20:52, 9 January 2018 (UTC)
Thanks, that resolves this. Blue Rasberry (talk) 14:52, 10 January 2018 (UTC)

Format of Isidore scholar ID (P4491): identifier of a scholar on Isidore, a platform that collects links to scholarly documents and official texts

Box office in visitors

In the film Demain tout commence (Q23899238) box office (P2142) are measured in person (Q215627). Looks like violation of box office (P2142)? - Kareyac (talk) 16:34, 13 January 2018 (UTC)

Yes. checkConstraints does not find it, though. --Tagishsimon (talk) 00:44, 14 January 2018 (UTC)
Thanks. - Kareyac (talk) 12:11, 14 January 2018 (UTC)
This section was archived on a request by: - Kareyac (talk) 16:24, 17 January 2018 (UTC)

Custom javascript breaking other scripts

Hi, I don't know where else to ask this. I added a one-line jQuery thing to the bottom my common.js. It works as it should, but has the unintended consequence of breaking some of the other scripts I'm using, and I don't understand why. Could someone with some more know-how figure out what I've done wrong? Jon Harald Søby (talk) 12:36, 17 January 2018 (UTC)

You’re using $(".wikibase-title-id").html().slice(1, -1) to get the item ID and then replace $(".wikibase-title-id") with something else. If another gadget tries to read the item ID using the same code, it will no longer work. This could be avoided by using mw.config.get( 'wbEntityId' ) instead (both in your common.js and in the hypothetical gadget or user script that also uses $(".wikibase-title-id")).
A few random other notes – if you keep the wikibase-title-id class on the new element, you can remove most of your custom style (color, font size and margin will be inherited from the class). It probably also makes sense to set the readonly attribute on the <input> – I’m guessing this is just intended as a quick way to copy the entity ID? --Lucas Werkmeister (WMDE) (talk) 12:55, 17 January 2018 (UTC)
Thanks Lucas! I tested your suggestions, which worked fine, but the problem persevered – I then tried reverting my common.js back to what it was before, and the problem was still there, so I finally figured it wasn't my common.js that was the problem, but rather the Chrome extension I used to test this. Even though the extension's input was empty, it was still enabled for Wikidata.org, and it loaded jQuery even though jQuery is already here via MediaWiki. So when I turned that off, everything works as it should. Thanks for helping! :-)
Oh, and yes, it's mainly for copying ID without having to use the mouse to select it, but also useful for a couple of other (Python) scripts I'm running. :-) Jon Harald Søby (talk) 13:24, 17 January 2018 (UTC)
This section was archived on a request by: Jon Harald Søby (talk) 14:13, 17 January 2018 (UTC)

Cleanup of unsourced “ethnic group (P172)” claims

ethnic group (P172) is and has always been a controversial property that has a “claim requires source” constraint for good reasons; see Property talk:P172#Constraints and Property talk:P172#Rules for Usage for further requirement regarding applicable sources. However, only 861 claims are sourced, and 48269 are not (i.e. 98.2% of all claims violate the constraint, see Wikidata:Database reports/Complex constraint violations/P172 and this query). In my opinion the constraint itself is not debatable, thus I suggest to remove all constraint-violating claims with a bot.

We have had a similar discussion for sexual orientation (P91) at Wikidata:Project chat/Archive/2016/08#Unsourced and Wikipedia sourced P91 statements and Wikidata:Project chat/Archive/2016/10#Unsourced sexual orientation (P91) statements, which ultimately led to the removal of unsourced P91 claims. Any thoughs on this one? —MisterSynergy (talk) 22:13, 5 January 2018 (UTC)

If you look at #Country of citizenship is Wales you can see that this can be quite a passionate subject matter. Personally I find it a minefield. For some who live in countries where there has been little migration maybe it is easier, however, for those who live in countries with significant migration, in different waves, and resulting inter-marriage it is quite problematic, vaguely or ill-defined, and problematic in its application. So, if we are going to have it then I would agree that it should be sourced by a contemporary source, and the target categorisation should also be notable/encyclopaedic. If we are to bot remove, I believe that any edit summary should have a good link to a page that clearly defines how to add, qualify and cite the link. I would wish to avoid edit wars, complaints about politics, nationalism, etc.  — billinghurst sDrewth 12:43, 6 January 2018 (UTC)
  • As far as I recall, the cleanup on P91 went fairly well. I think we should handle P172 in the same way. It's odd that people ignore the relevant constraint.
    --- Jura 10:09, 7 January 2018 (UTC)
    • Yes it did go fairly well. After the linked WD:PC discussions were finished, I removed ~5000 P91 claims, and there were no serious complaints raised. A couple of removals have been reverted, particularly by one user, who has always added sources to the reverted claims subsequently. The number of unsourced P91 claims is still low at a bit over 100, although I have never repeated another removal run (no idea whether someone else did so). —MisterSynergy (talk) 10:15, 7 January 2018 (UTC)
      • @Heathart: care to comment here? I just ran 
        SELECT ?item WHERE { ?item p:P172 ?ethnicgroup . ?ethnicgroup ps:P172 wd:Q49085 . MINUS { ?ethnicgroup prov:wasDerivedFrom [] } }
        Try it!
        and this seems to affect quite a lot of African American artists. I'm not sure if ethnic group (P172) is as controversial as sexual orientation (P91). I don't have a strong opinion on this, but it would be a waste to loose so much date just to be political correct. More practical problem: Sourcing this is going to be hard. Multichill (talk) 12:25, 7 January 2018 (UTC)
        • @Multichill: and all... I think it is important to “Use this property only together with a reference in that the person itself states her/his ethinicity" but also would propose using notable sources (including Wikipedia or credible news sources) where this information is dependable but secondary, I'm not sure how to state that. Basically if there are no interviews with said person, but they appear in a catalog for an exhibition at the MoMA of paintings by African American Artists, I'd count that as a reference to the persons P172 property. Otherwise I avoid this and agree that something that does not have a reference in this case should be removed. It would be great if all 48269 could be automatically vetted and sourced but I agree the property shouldn't just be assumed. --Heathart (talk) 16:28, 11 January 2018 (UTC)
  • I understand the issue but I would like to discuss another aspect of the problem. For Little Raven (Q535711) I wanted to add the information that he is a member of Arapaho people (Q626136), which is kind of trivial information (it's in his description and the title of the English Wikipedia article). I used ethnic group (P172), and since there is a need for reference, I added "imported from English Wikipedia" . But there are lot of other people we have, for instance, pictures on Commons (with good quality metadata) without an article in Wikipedia (yet). Should we reference them with "imported from Wikimedia Commons" in these cases ? Léna (talk) 15:58, 7 January 2018 (UTC)
    • According to the talk page of this property, the requirement for the reference is: “Use this property only together with a reference in that the person itself states her/his ethinicty”. This means we need external references anyway, and imported from Wikimedia project (P143) references to Wikimedia projects are (in probably all cases) not enough anyway. —MisterSynergy (talk) 15:18, 8 January 2018 (UTC)
  •   Comment It would be interesting to hear from contributors who monitoring recent changes/IP edits with this property.
    --- Jura 08:51, 8 January 2018 (UTC)

PetScan

I can not log-in WiDar in Petscan to edit Wikidata since last few days due to some strange reason. I used to edit a lot but had to stop bulk editing since last few days. Can anybody tell me what should I do? I can edit from other tools using WiDar.--157.32.0.22 06:28, 12 January 2018 (UTC)

Titles for painting (Q3305213)

I would like to ask about how original titles for painting (Q3305213) should be modeled. There's title (P1476) and native label (P1705), and I am not sure I see the difference between the two and which should be used to model the original name of the artwork. I see two options:

  1. title (P1476) is used for all titles in every language and native label (P1705) only for the original one. Then I do not see how title (P1476) is different from the label/alias. Shouldn't the same thing that goes into title (P1476) be a label? It seems so.
  2. title (P1476) is the original title (or titles) in the original language that the creator gave to it or that it was first published/discovered/etc. All the rest of translations go to labels/aliases. But then I am not sure what native label (P1705) is for and whether it should be used for paintings at all?

Right now we have 1094 paintings with title (P1476) and 184 paintings with native label (P1705). Clearly title (P1476) is leading. But, I am not sure all usages are correct - i.e. that only original title and not all the translations which were originally never used are there, e.g. see Mona Lisa (Q12418) - it has a Russian title and I am reasonably sure Leonardo da Vinci (Q762) did not speak Russian.

So, could we have some agreed rules (maybe recorded in Wikidata:WikiProject_sum_of_all_paintings page?) that all paintings would follow? We can fix non-complying items with a bot or script, it's not hard, but we should figure out the rules first.   WikiProject sum of all paintings has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. Laboramus (talk) 08:44, 8 January 2018 (UTC)

  • I think there is some bug in the Russian infobox tool that keeps adding P1705. We had the same problem with films. Unfortunately Putnik isn't following up on bugs in his tools when reported in English at Wikidata (User talk:Putnik).
    --- Jura 08:55, 8 January 2018 (UTC)

I use title (P1476) for all titles I can find in authoritative works. For paintings in Danish GLAMs it often means the Danish and the English titles and sometimes a German title. Multiple Danish titles if there are authoritative title changes (I cannot find an example). As an example: A nude woman doing her hair before a mirror (Q6644913). I have never used native label (P1705). — Finn Årup Nielsen (fnielsen) (talk) 16:36, 8 January 2018 (UTC)

Most paintings have names made up by later generations. I doubt Leonardo da Vinci ever called it Mona Lisa in any language. So does title (P1476) even apply here? Multichill (talk) 17:09, 8 January 2018 (UTC)
True so I am not even sure about title (P1476) - but it looks like native label (P1705) is not the way to go and probably first thing native label (P1705) should be moved to title (P1476) and then we should figure out what to do with title (P1476). I agree that many paintings don't have original titles, but some do, and then if we just use it the same way as labels, why use both? Laboramus (talk) 01:09, 9 January 2018 (UTC)
Sounds reasonable! I once wrote Wikidata:WikiProject Visual arts/Item structure#Titles, I don't know of a discussion about artwork titles up to know. Please add advice you consider useful or agreements from here there! The CONA guidelines are also worth seeing. --Marsupium (talk) 14:40, 12 January 2018 (UTC)

pywikibot / iterating over SPARQL causes error

Tracked in Phabricator
Task T138364

Hi project chat,

I am running into an error with pywikibot when iterating over the results of a SPARQL query following the instructions from https://www.wikidata.org/wiki/Wikidata:Pywikibot_-_Python_3_Tutorial/Iterate_over_a_SPARQL_query

My query

SELECT ?item WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  OPTIONAL { ?item wdt:P856 ?official_homepage. }
}
LIMIT 50000

The python script (official_homepage.qr contains the query from above) 

import pywikibot
from pywikibot import pagegenerators as pg

with open('official_homepage.qr', 'r') as query_file:
    QUERY = query_file.read()
site = pywikibot.Site("wikidata", "wikidata")

def get_items_with_offical_webpage():
	generator = pg.WikidataSPARQLPageGenerator(QUERY, site=site)
	items_with_official_webpage =[]
	for result in generator:
		try:
			itemID = result.getID()
			print(itemID)
			items_with_official_webpage.append(itemID)
		except Exception as e:
			print(e)
	return items_with_official_webpage

print(get_items_with_offical_webpage())

The output looks something like this

Q534258
Q642056
Q2157563
Q2157563
raw write() returned invalid length 20 (should have been between 0 and 10) #this one is strange, but the try-expect handles it
Q2638147
Q2056103
Traceback (most recent call last):
  File ".\official_webpage_query.py", line 20, in <module>
    print(get_items_with_offical_webpage())
  File ".\official_webpage_query.py", line 11, in get_items_with_offical_webpage
    for result in generator:
  File "C:\Users\tozi\AppData\Local\Programs\Python\Python35-32\lib\site-packages\pywikibot\pagegenerators.py", line 2787, in <genexpr>
    items_pages = (pywikibot.ItemPage(repo, item) for item in data)
  File "C:\Users\tozi\AppData\Local\Programs\Python\Python35-32\lib\site-packages\pywikibot\page.py", line 4101, in __init__
    super(ItemPage, self).__init__(site, title, ns=ns)
  File "C:\Users\tozi\AppData\Local\Programs\Python\Python35-32\lib\site-packages\pywikibot\page.py", line 3592, in __init__
    % (self.id, self.entity_type))
pywikibot.exceptions.InvalidTitle: 'P349' is not a valid item page title
<class 'pywikibot.exceptions.InvalidTitle'>

Can someone tell me what I am doing wrong and/or how to fix this?

Thanks for your input!

Regards, --Tozibb (talk) 12:28, 9 January 2018 (UTC)

List of properties directly related to Wikimedia projects

Is there a list somewhere of Wikimedia-related properties? I just found this one: Wikidata:Property proposal/Wikidata SPARQL query equivalent. Thx. Jane023 (talk) 14:19, 11 January 2018 (UTC) Oops should link Wikidata SPARQL query equivalent (P3921). Also found this list of works (P1455). Jane023 (talk) 14:21, 11 January 2018 (UTC)

I don’t fully understand what you look for. “directly related” and “Wikimedia project” leave much space for interpretation… We do in fact have much more properties that help us to organize content, you can find many of them by using Special:Search with keywords “Wikimedia”, “Wikidata”, “Wikipedia”, etc. and the namespace-prefix “Property:”. I don’t know of any systematic list. —MisterSynergy (talk) 14:32, 11 January 2018 (UTC)
I just tried that unsuccessfully. It would be useful to have a list of these somewhere. Jane023 (talk) 14:43, 11 January 2018 (UTC)
Like that: Property: Wikidata, Property: Wikimedia, Property: Wikipedia, and so on. Mind the space after the namespace filter. —MisterSynergy (talk) 14:55, 11 January 2018 (UTC)
We have many properties related to WikiProject Categories (Q4391019): [7]. We also have properties more directly related to Wikidata, e.g. Wikidata time precision (P2803), P2837 (P2837), kinship equivalent in SPARQL at Wikidata (P4316), P4570 (P4570) or related to WikiVoyage (page banner (P948)) --Pasleim (talk) 14:57, 11 January 2018 (UTC)
I've just created {{Wikimedia project properties}} - please feel free to populate it as you see fit. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:25, 12 January 2018 (UTC)
Nice - that seems to work fine! Jane023 (talk) 17:28, 12 January 2018 (UTC)

Translate "publish"

To save something in wiki data item there is "publish". This word was correct translated in Greek (el). Now we can see the English word. What is going on? Xaris333 (talk) 10:14, 12 January 2018 (UTC)

I can't see it translated... Matěj Suchánek (talk) 10:20, 12 January 2018 (UTC)
But it was a Greek word before. Did something changes on Wikidata? Maybe the English word was "save"? Xaris333 (talk) 16:41, 12 January 2018 (UTC)
Right, see Wikidata:Project chat/Archive/2018/01#Change on the editing interface: save becomes publish. Matěj Suchánek (talk) 16:45, 12 January 2018 (UTC)

Castles as occurrences ???

For a project finding Commons categories for map images, I am trying to find the items with a CommonsCat nearest to a known geographical coordinate-pair. I also prefer to avoid items in the subclass tree of occurrence (Q1190554), i.e. to prefer Stonehenge (Q39671) over Stonehenge Free Festival (Q1407488).

However, I've just run a set of lookups, and I'm finding all instances of castle (Q23413) (and other classes) now in the subclass tree of occurrence (Q1190554).

Using this query tinyurl.com/y7g8t9ek to extract the chain, it looks like the problem is down to the following sequence of P279 links:

architectural structure (Q811979) -> real property (Q684740) -> property (Q6422240) -> exclusive right (Q1146011) -> rights (Q780687) -> agreement (Q321839) -> contract (Q93288) -> legal transaction (Q327197) -> legal act (Q27095657) -> intentional human activity (Q451967)

Where do people think would be the right place to break this? It seems to me a castle is not a legal transaction (though ownership of a castle might be the result of a legal transaction).

The immediate cause of the chain creation was this edit (diff) by User:Osteologia on 27 December, making exclusive right (Q1146011)subclass of (P279)rights (Q780687); but that in itself doesn't seem so inappropriate.

How would anyone suggest this could be fixed? Jheald (talk) 12:47, 12 January 2018 (UTC)   WikiProject Ontology has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. Jheald (talk) 12:55, 12 January 2018 (UTC)

Ink Master Spin Off Show Data

There isn't any information about the Ink Master spinoff show.

Ink Master Angels Ryan Ashley Kelly Doty Gia Rose Nikki Simpson


Season 1 Face-Off Winners

S1 Ep1 Viva Las Angels "Tone Chingon" vs (w) Angels "Ryan Ashley"

S1 Ep2 Bigger'n Dallas Angels "Gia Rose" vs (w) "Deanna Smith"

S1 Ep3 Moons Over Miami "Jose Rosado" vs (w) Angels "Kelly Doty"

S1 Ep4 Smells Like Seattle Spirit "Graydon Payne" vs (w) Angels "Nikki Simpson"

S1 Ep5 Angels in Hotlanta Angels "Gia Rose" vs (w) "Tim Furlow"

S1 Ep6 Golden Gate Angels Angels "Ryan Ashley" vs (w) "Daniel Silva"

S1 Ep7 Music City Ink "Tai Orten" vs (w) Angels "Kelly Doty"

S1 Ep8 Steel City Showdown Angels "Gia Rose" vs (w) "Rachel Helmich"

S1 Ep9 Keep Austin Inked Angels "Kelly Doty" vs (w) "Sparks"

S1 Ep10 Angels in the Big Easy "Tai Viper" vs (w) Angels "Nikki Simpson"

S1 Ep11 Angels of the Inner Harbor "Tony Scientific" vs (w) Angels "Ryan Ashley"

The artists that are going on to compete on "Ink Master" Season 10 “Return of the Masters” are: Deanna Smith Tim Furlow Daniel Silva Rachel Helmich Sparks Geralyn69 (talk) 06:16, 13 January 2018 (UTC)

You should probably get your head around wikidata and sort that omission out, @Geralyn69: :) . --Tagishsimon (talk) 00:47, 14 January 2018 (UTC)

Query Service

Could someone perhaps tell me a little more about the implementation of this functionality? Is it intended to replace in the future the standard categorization systems on the Wikimedia projects? Is it just an additional functionality? De Wikischim (talk) 15:47, 13 January 2018 (UTC)

To the extent I understnd your question @De Wikischim:, wikidata is additional to Wikipedias, and is not designed to replace standard categorization systems on Wikimedia projects. Wikidata contains records of 'items' to which structured 'statements' are attached. The query service enables sets of items and their statements to be selected and displayed. A number of tools and techniques exist to utilise wikidata within Wikipedias, such as as lists - e.g. Listeria (Q24045615); wikipedia templates that extract data from wikidata; and interwiki language links. If you can clarify your query, more can be said. hth --Tagishsimon (talk) 20:34, 13 January 2018 (UTC)

Python problem

I'm trying to add (300, so no - I don't want to do that manually) single value exceptions to Sports-Reference.com Olympic athlete ID (archived) (P1447) via bot. Basically I have such code. The problem is, that in line 32 the apires structure gets back to state, that it had in line 24, although I see, that in the for loop (line 31) exceptions gets added to apires. Some ideas? Also on making the code less hacky (formatting isn't important). --Edgars2007 (talk) 08:58, 20 January 2018 (UTC)

Somehow fixed the issue by adapting DeltaBot code. Thanks, Pasleim! --Edgars2007 (talk) 17:32, 20 January 2018 (UTC)
This section was archived on a request by: --Edgars2007 (talk) 17:32, 20 January 2018 (UTC)

Search without 'scientific articles'

Tracked in Phabricator
Task T183510
Resolved

Is it possible to use search and get rid of hundreds of 'scientific article' items? Sometimes I get a few hundreds results and 99% of them are scientific articles... Also: is it possible to search within properties (and without using SPARQL for this)? E.g. I can't find an item using labels in several languages but I know CAS Registry Number (P231). Wostr (talk) 02:48, 10 January 2018 (UTC)

Yes it's already   Done. If something doesn't work, please note on the phab task. Smalyshev (WMF) (talk) 22:04, 10 January 2018 (UTC)
    • I have already noticed a lot of improvements, especially when searching for English terms using another language. Maybe Wostr can give us some example query? Sjoerd de Bruin (talk) 13:45, 10 January 2018 (UTC)
      • I had a problem with searching for e.g. aliphatic hydrocarbon. Now there is an item about it, but I've just created it. Before, I got some 'scientific articles' in the first place, then some chemical reactions and biological processes, and again a bunch of 'scientific articles', then one book and heptene (Q151375) (which was something similar to the searched phrase). In the first 200 results there wasn't Category:Aliphatic hydrocarbons (Q8900110) (I found it just right now and I could link this item to the newly created). But still, I'm not sure whether there was some item describing aliphatic hydrocarbons before and I just missed it in the results. But that was not an isolated problem: I've created some items that I had to merge with older items when I had accidentaly found such items. I think deprioritization of 'scientific articles' won't be of much help here, because (1) items are still poorly described, (2) sometimes I need to find not only specific item, but also categories, disambigs etc. and check connections to existing chemical classification. Maybe there is some hack with search to exclude unwanted phrase with regex or something? Wostr (talk) 15:48, 10 January 2018 (UTC)
@Wostr:
well, for simple search on the labels/alias of items, I'm pretty sure @Magnus_Manske: made a simple tool, that can return all items with the same string you are looking for (I use it for Names). I can't find it right now (not on the right machine)... maybe someone can give it... or I'll do when I'm home ^^ --Hsarrazin (talk) 16:01, 10 January 2018 (UTC)
got it : https://tools.wmflabs.org/wikidata-todo/relabel.php? - simple but efficient when you are looking for a name or a scientific designation of a substance, taxon, etc. ;) --Hsarrazin (talk) 20:26, 10 January 2018 (UTC)
Ok, that is something. Not a solution for the whole problem, but may be useful for me, thanks :) Wostr (talk) 22:53, 10 January 2018 (UTC)
  • This whole deboosting thing with 'scientific articles' is still far from ideal. I lost over 20 minutes today while searching for "alginate" (I stopped at 2k result, as from about 1,5k there were only scientific articles). I found 9 results that I was looking for (incl. 2 duplicated items), but ammonium alginate (Q9147958) was ~900th result? (right now, it's ~400-500th result) Probably because it hadn't English label/description (only French). Fortunately, I knew about this item before searching, but I was determined to find it this way (also, I didn't know if there were more items like this, so I had to check it anyway). Wihout some way to exclude 'scientific articles' from search results, using search at this moment is, sadly, just a waste of time. Wostr (talk) 15:54, 12 January 2018 (UTC)
    If the item doesn't have an English name, then I think it's understandable if it's ranked badly when you search in English. As far as the subject matter goes, should alginate (Q11685373) be linked with conjugate base (P4149) to alginic acid (Q422092)? ChristianKl ❪✉❫ 20:41, 14 January 2018 (UTC)

Cyprus places at the north part

Hello. Many places of the island of Cyprus has a problem. An example:

and

There are many other properties that the item have 2 values, one for Cyprus (Q229) and one for Northern Cyprus (Q23681).

What to do? How to show these informations in Wikidata?

Xaris333 (talk) 21:41, 13 January 2018 (UTC)

Well it is & isn't a problem. Item statements can take multiple values; ideally there'll be a qualifier which can be deployed to provide some context. From the top of my head, perhaps a qualifier of sourcing circumstances (P1480) taking a value of disputed (Q18912752). Ideally P1480 would take values of de jure (Q132555) and de facto (Q712144) - but right now, they're not listed as permissible values. Obvs, from Turkey's point of view, Northern Cyprus is both de jure & de facto. The rest of the world may disagree with the first of those; which is why Q18912752 might be a better option. (And there may well be other useful qualifiers, or other useful properties to be employed - I note in passing territory claimed by (P1336)). --Tagishsimon (talk) 23:29, 13 January 2018 (UTC)
Thanks. But I need a more specific answer what property to use. To be sure about it. Xaris333 (talk) 16:10, 14 January 2018 (UTC)
I'm afraid there's a certain amount of making-it-up-as-you-go. Crimea (Q7835) has the same issue, and solves it with statement disputed by (P1310) and territory claimed by (P1336) qualifiers for an instance of (P31) value (which could equally have been applied to the country (P17) value in that record). That might be the way to go? --Tagishsimon (talk) 23:05, 14 January 2018 (UTC)
Spratly Islands (Q215664) Seems likely, its country (P17) value were removed 2 years ago, and I restored this property as <unknown value> in 1 Jan. --Liuxinyu970226 (talk) 04:37, 15 January 2018 (UTC)

Creating many items

Hello. Is there a tool to create many items by just using the labels?

Xaris333 (talk) 23:07, 13 January 2018 (UTC)

Not sure what you mean by 'just using the labels'. QuickStatements 2 (Q29032512) is an excellent way of creating items and /might/ fit your brief. --Tagishsimon (talk) 23:14, 13 January 2018 (UTC)
Thanks. Xaris333 (talk) 00:11, 14 January 2018 (UTC)
@Xaris333:, just to be clear: You don't intent to create items with no statements, but only labels, right? --MB-one (talk) 17:44, 14 January 2018 (UTC)
Don't worry. I just want to create the item easily, to add the statements later. Already done. Xaris333 (talk) 17:48, 14 January 2018 (UTC)

Next IRC office hour on January 30th

Hello all,

The Wikidata team will organize an IRC office hour, January 30th, at 17:00 UTC (18:00 Berlin time). It will take place in the Wikimedia office IRC channel #wikimedia-officeconnect. As usual, we will present you some news from the development team, the projects to come, and collect your feedback.

This year, we would like to try something different, and have a topic to focus on during the office hour. If you have any topic you'd like to bring for the first meeting, please share it here!

Thanks, Lea Lacroix (WMDE) (talk) 14:31, 4 January 2018 (UTC)

  • Thank you for the information. I would like to learn more about strategies that are being considered for how to address the growth of Wikidata. YULdigitalpreservation (talk) 19:59, 9 January 2018 (UTC)
  • There is that redirect RfC from last year which we still wait WMDE to comment on. I fully understand that the outcome is not easy to implement and maybe somewhat undesired by WMDE (I don’t like many of the details of it, either). However, there is clear demand for improvements regarding sitelink management (per RfC), and several disputes of the past months could have been avoided if a solution akin to the proposal of the RfC was available. I thus think that it would be fair to have a WMDE/dev comment on the outcome of this RfC, and in case that this comment leaves space for an actual implementation at a later time, an IRC office hour could be used to discuss critical aspects with the community. @Lea Lacroix (WMDE), Lydia Pintscher (WMDE), ChristianKl: FYI. —MisterSynergy (talk) 17:30, 12 January 2018 (UTC)

Facto Post – Issue 8

The first 2018 issue of the Facto Post newsletter is out today on Wikipedia's birthday. Editorial "Metadata on the March", and a prompt to get involved in the #1lib1ref campaign. Subscription details are in the newsletter footer. Charles Matthews (talk) 13:01, 15 January 2018 (UTC)

Making P18 values usable

Is there some kind of library that allows to automatically build attribution text for a given image on the commons?

I plan on reusing large amount of images querried from P18 properties and I need some way to automate generating credit. I prefer to avoid writing parser of commons templates from scratch - reusing existing code would be preferable.

I expected that there is some pywikibot script already and I am unable to find it.

I found https://github.com/wikimedia/mediawiki-extensions-MultimediaViewer https://github.com/wmde/Lizenzhinweisgenerator https://commons.wikimedia.org/wiki/MediaWiki:Gadget-Stockphoto.js but all three are usable only as an inspiration

I am pretty sure that somebody already made such code, I would not expect to be the first person who is using WIkidata to find multiple images and intends to not violate their attribution requirements

Mateusz Konieczny (talk) 13:48, 15 January 2018 (UTC)

Opening days and prices

I just discovered this project and did some experiments with SPARQL and it's awesome.

Since I'm currently planning some tourism trips, I am currently digging into castles data.

Some pieces of information which are quite difficult to specify are the opening days and prices of admission.

I saw three properties realted to opening days, but these do not seem to allow specifying things such as "open from 9h00 to 18h00 on Tuesdays, but closed on public holiday, except a particular day for which the opening times are different"

Do any of you know if there's currently some ongoing work on these subject ?

--Gzorg (talk) 12:26, 12 January 2018 (UTC)

I'd like to propose to create a new property using the scheme of osmwiki:Key:opening hours/specification with a new property once again! --Marsupium (talk) 13:30, 12 January 2018 (UTC)
That seems quite interesting indeed. Let's see. The Louvre is open from 9h to 18h on Mondays, Thursdays, Saturdays and Sundays, and from 9h to 21h45 on Wednesdays and Fridays, but is open from 9h to 18h on 04/16 04/17 05/25 06/04 06/05 07/14 11/1 11/11 and closed on 05/01 08/15 and 12/25. Can the hours specification handle this case ?
Another example which is a little different is the Amboise castle : 9h - 16h45 from 01/01 to 07/01, 9h-12h30 and 14h to 16h45 from 01/08 to 31/01, 9h-12h30 and 13h30-17h in February, 9h-17h30 in March, 9h-18h30 in April and May and June, 9h-19 in July and August, 9h-18h in September and October, 9h-18h on 11/01, 9h-17h30 from 11/02 to 11/15, 9h-12h30 and 14h-16h45 from 11/16 to 11/30, 9h-12h30 and 14h-16h45 from 12/01 to 12/24 and 9h-16h45 from 12/26 to 12/31, and closed on 01/01 and 12/25. --Gzorg (talk) 14:44, 12 January 2018 (UTC)
There was a proposal to add open days and opening hours: Wikidata:Property_proposal/opening_hours. Most of the proposal was accepted, except opening hours because it was judged that not enough people were interested in them. Now that more and more people (including me) are interested in them, and with Wikivoyage having integrated Wikidata in their listings editor, I think it is time to open a new discussion about this. I know someone will always be able to find a tricky case such as "everyday except when it snows", but I am confident we can find a way to make this work for 99.9% of the cases. Cheers! Syced (talk) 09:02, 15 January 2018 (UTC)
From having used Foursquare, google places and facebook places editing a lot, I can tell you that opening hours is really hard to keep accurate and up to date. You have seasonal times to model, exceptions like holidays, 24 hours open, does not apply, open to general public, open to tours, open for kids, open vs. consultation hours, open for as long as people drink beer. Most often, people can't be bothered to accurately specify all of those details and then next year, they change it up again... There is a reason why facebook and google back this with automatic "popular times" features, it simply proved more accurate then what people were entering into the entry fields quite often. Another problem was that it was VERY hard for those parties to find reviewers of this information. Most of it is now auto approved if you are an experienced places editor, simply because no one will approve or disapprove small things like this, esp when it gets complicated. When disputed, or when contributed by non-experienced users, they will ask app users when they visit the specific location ("Please confirm if the following opening hours are correct for this venue"). So i'd put some significant consideration into it. TheDJ (talk) 15:44, 15 January 2018 (UTC)
Each Wikivoyage has been maintaining opening hours for a decade, and is doing a reasonably good job at it. Sharing this data via Wikidata would reduce the maintenance cost by factorizing the efforts of all Wikivoyages. I believe this is exactly what Wikidata was built for in the first place, so I support the creation of the property. I agree that now is the time to decide what is the best format, and I see no reason to not use the OpenStreetMap format. Cheers! Syced (talk) 07:31, 16 January 2018 (UTC)
In any event, here's the expression for the Louvre : Mo,Th,Sa 09:00-18:00, We,Fr 09:00-21:45 ; May 05, Aug 15 closed ; Apr 16, Apr 17, May 25, June 04, Jul 14, Nov 1, Nov 11 09:00-18:00, and there is an evaluation tool which is quite fun to play with. I plan on doing Amboise soon for the sake of completion --Gzorg (talk) 14:47, 15 January 2018 (UTC)

Wikidata weekly summary #295

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Discussions
    • Open request for adminship: Mahir256
    • Closed request for adminship: Jarekt (as successful)

Gadget for Scholia?

Scholia (Q45340488)

We have a gadget to add a sidebar link to an items Reasonator equivalent. Could someone do the same for Scholia, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:52, 16 January 2018 (UTC)

English Wikipedia banned links to Wikidata

English Wikipedia banned links to Wikidata and they are being deleted. The discussion was at w:Wikipedia_talk:Manual_of_Style#Linking_to_wikidata  [ < outdated link, was probably archived to w:Wikipedia_talk:Manual_of_Style/Archive_202#RfC:_Linking_to_wikidata ] , too bad no one at Wikidata participated when I brought it up earlier. --RAN (talk) 18:10, 15 January 2018 (UTC)

'too bad' only from the persective of wanting to preserve inappropriate links to wikidata on en.wikipedia. From the quick skim, one/some en.wiki users were linking to wikidata by way of referencing notability ... that seems altogether against en.wiki's reliable source & external links policies; and so it's unsurprising that en.wikipedians are taking a dim view of it. --Tagishsimon (talk) 18:16, 15 January 2018 (UTC)
The example is w:Mayors of Teaneck, New Jersey and you argue it is "inappropriate", can you explain what makes it inappropriate. The mayors appear on the list because they were mayors of Teaneck, New Jersey, not because they appear in Wikidata. Removing the link at the table of mayors leaves us with a names like "John X" that can be any John X. The whole point of Wikidata is to be able to distinguish between people of the same name and point to more data on that person, like their entry in Findagrave or in LCCN and VIAF. --RAN (talk) 18:31, 15 January 2018 (UTC)
I don't really want to relitigate an en.wikipedia matter here, if it's all the same to you. The discussion you linked to discusses the point you are trying to reheat, and I think speaks for itself. --Tagishsimon (talk) 19:19, 15 January 2018 (UTC)

I fail to see the problem. In deWP links to outside deWP are as well forbidden in the article text (with just five exceptions). They are allowed in references, as web-links and in infoboxes, that's it. We have this rule for quite some rime, and it's fine that we do. Grüße vom Sänger ♫ (talk) 19:17, 15 January 2018 (UTC)

  • I also don't see a problem. Such inline links would be forbidden in pl.wiki as well and I really don't know why it should be discussed here – it's en.wiki problem not WD. Wostr (talk) 19:25, 15 January 2018 (UTC)
  • I'm not really convinced by the inline links either, but there should be a way to link rows in lists to Wikidata items they relate to similar to Wikivoyage. This when Wikipedia has no article about the person.
    --- Jura 19:39, 15 January 2018 (UTC)
    It can be done similarly to how it links to other language Wikipedias - essentially, this is an external link. The difference is that Wikivoyage links to entities which can not be at Wikivoyage because they are out of scope, whereas in Wikipedia everything is in scope - if the subject is notable and the article is not there, it should be a redlink.--Ymblanter (talk) 19:47, 15 January 2018 (UTC)
    There are fields where English Wikipedia accepts content in list form, but not as separate items (e.g. TV episodes). Each episode would have an item at Wikidata (and might supply basic information), but there wont be a full Wikipedia article about it. The same may be the case for some mayors. (To be complete, the list needs to include every one, but an individual one might not be noteworthy enough for an article).
    --- Jura 19:51, 15 January 2018 (UTC)
Now the full RfC started, which will likely result in the full prohibition of having anything from Wikidata transcluded to the English Wikipedia.--Ymblanter (talk) 09:35, 16 January 2018 (UTC)
Ok, my two attempts to explain what Wikidata actually is ended catastrophically. People somehow assume that we are a service project which must deliver data to the English Wikipedia exactly in the format they want, and anything else results in personal attacks. I do not think I will try further. We probably need to ignore this and future similar discussions. If they do not need data, let them not have data.--Ymblanter (talk) 14:40, 17 January 2018 (UTC)
@Ymblanter: A useful fact in such circumstances can be to remind them that over the last month WDQS has been answering 3 to 4 million queries per day [8], a number that in recent months has sometimes gone as high as 10 million queries a day. (A number which is also quite distinct from serving up interwiki links or template values, neither of which use WDQS). So yes, Wikipedias are certainly an important group of customers for Wikidata, but Wikidata serves a lot more than just Wikipedias. Jheald (talk) 15:52, 17 January 2018 (UTC)
I actually mentioned that they are likely not the biggest user, but the most vocal opponents obviously only care about themselves. Anyway, I had my portion of shit, I am not going to do it anymore. I am personally not using Wikidata in what I am doing on Wikipedia, and I am not going to do such things out of principle. If people are short-sighted and refuse to order glasses, I can not help.--Ymblanter (talk) 15:58, 17 January 2018 (UTC)
Just to make sure: which RfC do you talk about? I see one being drafted, and a rather odd one regarding linking to Wikidata from main namespace. —MisterSynergy (talk) 14:59, 17 January 2018 (UTC)
The second one (which is also linked from the first message of the thread). Whereas it was indeed supposed to be a technical issue (and IMO a pretty useless one) most of the voters commented no on the technical issue, but on Wikidata in general.--Ymblanter (talk) 15:31, 17 January 2018 (UTC)

Two elements on the same concept have different links to Wikipedia in Arabic

Evidently Category:Botanical taxa by author (Q24900376) and Q16743146 deal with the same concept, but there is a conflict with the element in Arabic. I do not understand anything Arabic, can someone help me solve the conflict? --Metrónomo (talk) 03:37, 16 January 2018 (UTC)

These two do not deal with the same concept: one deals with "botanical taxa", the other with "species". There is some overlap, but not all that much. - Brya (talk) 11:50, 16 January 2018 (UTC)
That is not the difference (at least it is not in the Arabic Wikipedia). ar:أوفتية (genus Oftia) is categorized in ar:تصنيف:نباتات وصفها ميشال أدانسون (Plants described by Michel Adanson), which is categorized in ar:تصنيف:نباتات حسب الواصف (Q16743146). Besides there they also categorize botanists (why?). ar:تصنيف:أصنوفات علم النبات حسب الواضع (Category:Botanical taxa by author (Q24900376)) is categorized in ar:تصنيف:أصنوفات حسب الواضع ("animals by author"), which it is linked in Category:Taxa by author (Q4155659)! ar.wikipedia have a very exotic classification system, or very disorganized. I see that there is no difference between what they categorize in both categories (species + other taxa + botanist), but I do not know Arabic and I can not affirm it with complete certainty. I'm confused. --Metrónomo (talk) 13:53, 16 January 2018 (UTC)
There are two separate issues:
  • Should these two items be merged? No, they deal with different topics.
  • Are the arwiki sitelinks correctly placed? I have no idea, since I don't understand what arwiki is doing.
Brya (talk) 11:34, 17 January 2018 (UTC)

Structure of compound=structure of molecule?

Is this statement correct: "structure of compound is the same as structure of molecule"? Argumentation: "in the most popular definition of compound, it is substance composed of one kind of molecules, and the terms 'structure of compound' and 'structure of molecule' are used interchangeably" --Fractaler (talk) 10:35, 17 January 2018 (UTC)

I don't think they are generally used interchangeably. Wikipedia suggests that H2 for example isn't a compound. ChristianKl ❪✉❫ 11:11, 17 January 2018 (UTC)
Yes, I agree: compound (substance) consists of (identical) molecules. Molecule consists of atomes/groups of atoms. Thus, we have two different objects. And they can not be united by any fractal structure. --Fractaler (talk) 11:41, 17 January 2018 (UTC)
@ChristianKl: As I tried to explain it in WikiProject Chemistry, these two statemantes means the same in most cases. And the problem is not based on title statement, as Fractaler is trying to suggest, but it is strictly connected to chemical classes: we have chemical class pyridine (Q47317020)
  1. pyridines are a class of compounds containing pyridine ring in their structure (here 'chemical compound' means 'chemical substance = portion of matter consisting of identical molecules')
  2. pyridines are a class of compounds containing pyridine ring in their structure (here 'chemical compound' means 'molecule', i.e. 'molecular entity')
Both statements are correct, as both are describing molecular structure (on the same level microscopically), but with two different (both correct) definitions of chemical compound. That's all from me, because from the previous discussions I came to the conclusion that further discussion is pointless. Wostr (talk) 12:50, 17 January 2018 (UTC)

Newly created items are not available.

Hoi, I created "Governor of the Territory of Alaska". I cannot add it to the office holders.. What is going on? Thanks, GerardM (talk) 13:51, 17 January 2018 (UTC)

It took some time, now it is available.. Thanks, GerardM (talk) 13:56, 17 January 2018 (UTC)
I reported this earlier as T185079, but then it solved itself… Then I had the same problem again a bit later. So it seems there is just an unusually long delay in some cases – maybe related to T183053? Jon Harald Søby (talk) 14:12, 17 January 2018 (UTC)

Using data on software versions from Wikidata in Repology project

Just wanted to report that I'm using Wikidata in my Repology project (a service which aggregates data on software packages from a lot of package repositories and other sources, compares software versions etc.) and it was quite pleasant experience as the data is nicely structured and rather fresh (the most useful data for the project is fresh software versions, and about 85% of projects (see statistics) I get from wikidata have actual versions listed). Thanks to all Wikidata editors.

Repology can be used to improve Wikidata too, for example here's the list of software which may need version data updated in Wikidata. I've also shared some other thoughs on how Wikidata could be improved.

While here, maybe someone could review the SPARQL request I use to aggregate all data I need per software project. It turned out to be quite heavy because I need to gather all software versions and their properties. AMDmi3 (talk) 13:53, 17 January 2018 (UTC)

That looks very nice. Thanks for sharing! --Lydia Pintscher (WMDE) (talk) 18:43, 17 January 2018 (UTC)

Retracted papers

We have an issue at retracted paper (Q45182324) where GerardM insists on repeatedly removing subclass of (P279)=academic journal article (Q18918145); apparently in order to justify his repeated removal of instance of (P31)=retracted paper (Q45182324) from RETRACTED: Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children (Q28264479). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:23, 4 January 2018 (UTC)

Thanks for the misrepresentation and for the edit war. As it is, retracted papers are scientific papers. When they are retracted, all kinds of options are open including that papers are reinstated sometimes with changed information but not exclusively so. Consequently the notion that there is such a thing as retracted papers as a subclass is wrong and, what is left is to identify retracted papers as a significant event. Thanks, GerardM (talk) 12:27, 4 January 2018 (UTC)
Using significant event (P793) rather than instance of (P31) to model this sounds fine to me, but we should strive for a reasonable discussion and consensus on this. ArthurPSmith (talk) 13:06, 4 January 2018 (UTC)
I agree that significant event is more appropriate for modeling this relationship then instance of (P31). ChristianKl ❪✉❫ 13:12, 4 January 2018 (UTC)
"retracted papers are scientific papers. " Some retracted papers are humanities papers; which is why subclass of (P279)=academic journal article (Q18918145). "papers are reinstated sometimes" So class them as such. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:39, 4 January 2018 (UTC)
Are we really going to discriminate in the kind of paper involved.. REALLY !! Thanks, GerardM (talk) 14:10, 4 January 2018 (UTC)
If you speak about the identity of the paper than the kind of paper obviously matters a great deal. If you don't want a solution that cares about the kind of paper instance of (P31) isn't your property. ChristianKl ❪✉❫ 13:47, 6 January 2018 (UTC)

Subclass as a variable property

@Pigsonthewing is edit warring on the notion that a retracted paper is in a different subclass as scientific paper. It follows that subclass is no longer stable. As has been indicated, a retraction can be followed by a re-publication. Help me understand, can a subclass be this fickle or is it something that should be solid. If not can you please provide other examples where a change of state changes the subclass for an item? Thanks, GerardM (talk) 06:41, 5 January 2018 (UTC)

GerardM All subclass (P279) means is that if X is a subclass of Y, then when item A is an instance of X (A P31 X), it should also be considered to be an instance of Y, without having to explicitly add that statement in wikidata. If a specific article (A) is retracted (so A P31 X where X is "retracted paper", though as I mentioned above I think the significant event approach is better) then with the current subclass statement, A is also an instance of Y (academic journal article). This seems fine to me. Can you be clearer about what your concern is with this? ArthurPSmith (talk) 14:05, 5 January 2018 (UTC)
In my opinion being part of a subclass is not something fickle that changes on arbitrary events. This is imho quite basic. I understand quite well what a subclass is, my understanding is that the arbitrariness of this approach is wrong. Thanks, GerardM (talk) 14:26, 5 January 2018 (UTC)
Can you clarify? We start with item A, an instance of "academic journal article". It is retracted, and its P31 statement is changed to "retracted paper". Then the retraction is rescinded or whatever and it becomes "academic journal article" again. There is no change in "subclass", the change is to the P31 (instance) statement on A. This seems ok to me. Or did you have a different scenario in mind? ArthurPSmith (talk) 14:42, 5 January 2018 (UTC)
Ok another item .. a human .. he dies ... so we change the instance to "corpse"? Thanks, GerardM (talk) 15:27, 5 January 2018 (UTC)
No, all humans are P31 Q5, but that's a wikidata policy decision. And we don't have a wikidata class for "deceased human" at the moment anyway as far as I can see - if we did it would still be legitimate to make it P279 Q5. We do have woman (Q467) subclass of (P279) human (Q5). We could make a similar policy decision here if the consensus is for that, I certainly agree that would be sensible. But you're the one who created retracted paper (Q45182324), what did you have in mind with it if not as a subcategory of academic journal articles? ArthurPSmith (talk) 16:29, 5 January 2018 (UTC)
We have many things that have start time (P580) and end time (P582) attached to their instance of (P31) statements — e.g. most countries have conditional subclass status for being member state of the European Union (Q185441) or Commonwealth realm (Q202686) or Permanent members of the United Nations Security Council (Q6688521) etc. I think we often use instance of (P31) for things that would probably be better expressed using some other property, but currently this is a fairly well established approach. --Oravrattas (talk) 10:38, 6 January 2018 (UTC)
It shouldn't change based on arbitary events and that's why it's bad to model arbitary events like retraction with instance of (P31). 13:48, 6 January 2018 (UTC)ChristianKl ❪✉❫
Not sure what "arbitrary" means here. "Retraction" for an academic journal article is a major major event, important for anybody considering making some use of the article to be aware of. It would be unwise to favorably cite or otherwise rely on the results of a retracted paper after it was retracted, for instance. So I can see an argument for making that more obvious via the P31 relation. I don't think that's necessary, but I really don't see a significant problem either way. ArthurPSmith (talk) 14:50, 9 January 2018 (UTC)
I also am a little unclear in all this, in my own area I only deal with scientific and academic publications, retracted means the paper is not published and hence is not a citable resource. For the purposes of science, retracted papers effectively do not exist. Why would you try to list them at all? Whatever the reason of retraction there is no implicit garuntee of peer and editorial review, hence they are no better than a self published doc on a website. Cheers Scott Thomson (Faendalimas) talk 21:15, 14 January 2018 (UTC)
@ArthurPSmith: The core question is whether the event of the retraction is an event that happens to the paper or whether it's something that changes the core about what the identity of the paper happens to be. It's arguable whether the word "arbitrary" is good for speaking about this. ChristianKl ❪✉❫ 00:50, 18 January 2018 (UTC)

Lua request for form

Hi all

I can't find a place for a Lua request like on Wikipedia so I'm hoping someone can help me out, I want to create a very easy to use form for the Wikidata Data Import Hub for people to record new datasets. Currently there is a prefilled template which uses Source Editor which many non Wikimedian's find confusing and hard to fill out (one of the key demographics I'm trying to reach). All I want is a when you click a button you get a pop up box with a few fields with a short desciption next to them. I would really love something similar to when you click this link, chose any section and then click add new resource, I understand this is all done in Lua. The fields I would like are (with guidance after the colon):

  • Name: The name of the dataset you want to import
  • Source: The source of the dataset e.g the organisation who created it or published it
  • Link: A URL link to the dataset
  • Description: A short (less than 20 word) description of the dataset outlining the contents e.g A list of protected areas in Pakistan

These fields then then populate a new section on the main Wikidata Import Hub page like so:

==NAME==

Description of dataset Create and import data into spreadsheet Structure of data within Wikidata Format the data to be imported Match the data to existing data Importing data into Wikidata Date import complete and notes
Name:NAME

Source:SOURCE

Link: [LINK here]

Description: DESCRIPTION

Link:

Done:

To do:

Notes:

Structure:

Example item:

Done:

To do:

Done:

To do:

Notes:

Done:

To do:

Notes:

Done:

To do:

Notes:

Date complete:

Notes:

Is this possible/easy for someone to set up?

Thanks

--John Cummings (talk) 21:20, 16 January 2018 (UTC)

(The discussion on enwiki has determined that the page does not actually use Lua to generate the pop-up form, but probably uses JavaScript. Jc86035 (talk) 03:00, 18 January 2018 (UTC))

List of mayors redux

Another mayor list is up for deletion at w:List of mayors of Traverse City, Michigan. The previous discussion for a mayor list involved whether these lists belong on Wikidata, or Wikisource instead of Wikipedia. --RAN (talk) 20:39, 12 January 2018 (UTC)

Wikidata is not a site for hosting lists, per se, it may have the data that enables a list to be built; it may have the data that resembles a list. Wikisources have published works, and if that list is contained within a published work, then yes, if not, then no. Wikipedias have compiled lists, though where they make their notability standards. Wikipedias can have their discussions about their notability data and probably can look to see whether sites have that inclusion data, they should not be theorising on what other sites should have if they choose not to have something.  — billinghurst sDrewth 15:32, 13 January 2018 (UTC)
We do not host lists, we have position=Mayor of X with "replaces" and "replaced by" that would house this type of data as a concatenation. --RAN (talk) 18:16, 18 January 2018 (UTC)

Best practice - list and not-lists

list of figure skating jumps (Q1889337) is an example of an item tagged as a Wikimedia list article where at least one of the linked Wikipedia articles is a list (FR wiki) but many are not. I have created figure skating jump (Q47451002) but if I move the not-list wiki articles and the not-list statements to this item, I'll break the interwiki links. What would you do? - PKM (talk) 20:26, 16 January 2018 (UTC)

If it were up to me I would remove all instance of (P31) Wikimedia list article (Q13406463) statements and replace them with statements relevant to the concept being described on those wikipedia pages - that's what Wikidata should be about, right? In this case I think adding list of figure skating jumps (Q1889337) subclass of (P279) jumping (Q1151752) (or element (Q47450979) or both) would be fine. ArthurPSmith (talk) 20:35, 16 January 2018 (UTC)
It happens frequently that a Wikipedia has both a regular article about a concept and a list article, removing the concept of a list article isn't a good idea. ChristianKl ❪✉❫ 11:18, 17 January 2018 (UTC)
Yes, political information is full of a nasty mix of "Finance Minister of [some country]" vs "List of Finance Ministers of [some place]" type articles, with some Wikipedias having both, and others having only one or the other. I've done a lot of work teasing apart versions that are mis-combined or mis-joined. (It's also worth noting that these should generally also be connected via, e.g., , and is a list of (P360) back again.) --Oravrattas (talk) 17:37, 18 January 2018 (UTC)
Creating a new item to separate it, seems curently the agreed upon way. Once the RfC fro redirect is closed, additionally it would be good to create redirects when doing that. ChristianKl ❪✉❫ 11:18, 17 January 2018 (UTC)

Gary Auerbach (Q5524624) - Gary Auerbach (Q3098512) a probable mess

  WikiProject Movies has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.

Hi,

While trying to add IMDb ID (P345), I found Gary Auerbach (Q5524624) which seems to be a mixup of 2 different persons...

the description read "chiropractor and photographer" but enwp says "American television and film writer, director and producer" :( - should it be linked to Gary Auerbach (Q3098512) instead ?

Could someone please try and clean this up ? it seems very akward. --Hsarrazin (talk) 23:18, 16 January 2018 (UTC)

Hi, I rewrote Tambo (talk) 07:50, 17 January 2018 (UTC)

This comes up a few times a year. Wikipedia does a switch of the main topic for people with the same name. They swap who gets the name "John Smith" because one is considered more notable than the previous one. They never fix the Wikidata entry. --RAN (talk) 18:09, 18 January 2018 (UTC)

Also known as ... symbol

Ukraine Q212 shows a few aliases, including 🇺🇦. A similar symbol is at other items such as Netherlands Q55. What is the idea? Johnuniq (talk) 08:17, 18 January 2018 (UTC)

It seems to be a flag symbol, or rather emoji. See e. g. https://emojipedia.org/flag-for-ukraine/ --Anvilaquarius (talk) 09:10, 18 January 2018 (UTC)

Not only Unicode. ASCII also: +, -, *, /, etc. For example, Here and here I asked about this. Symbol=group of symbols? "+" and "plus sign" are the same (plus sign (Q6265342))? "🇺🇦" and "Ukraine" are the same (Ukraine (Q212))? email and "@" are the same (email (Q9158))? --Fractaler (talk) 12:35, 18 January 2018 (UTC)
On one hand, it seems weird, on the other hand, I can certainly imagine use-cases, for example, automated analysis of Twitter content, where you'd want the system to understand that a mention of "🇺🇦" most likely means a reference to Ukraine. --PeterisP (talk) 13:22, 18 January 2018 (UTC)

Should we re-allow country (P17)-Scotland (Q22)?

As Scottish independence (Q891905) still exists, the Scotland (Q22) is now somewhat like disputed territory (Q15239622), which is also OKay as a P17 value.

Currently we automatically replace this claim to country (P17)United Kingdom (Q145), but now is it still fair for us? If still yes, then why don't we autoreplace country (P17)Northern Ireland (Q26) to that?

Please also note that there's at least El Mamut Traçut (Q47456453)country (P17)Catalonia (Q5705) (even the Catalan pro-independence movement (Q893331) is simply illegal under any of EU laws) Thria (Q13415595)country (P17)Classical Athens (Q844930) (which I'm also confused: How was Athens (Q1524) independented?) and others available, so I think it's the time to unfreeze such claim. --Liuxinyu970226 (talk) 14:56, 18 January 2018 (UTC)

It don't make sense for me country (P17)United Kingdom (Q145) substitute country (P17)Scotland (Q22), once United Kingdom (Q145) is a sovereign state, and in that case, I think the most specific is the better option.
Unfortunately, I can't give a not biased opinion on Catalonia (Q5705), but in the case of Athens (Q1524) and Classical Athens (Q844930), we are talking about the city (Q515) city of Greece (Q41) since 1832 and Classical Athens (Q844930), which were a city-state (Q133442) that was dissolved in 85 BCE. Ederporto (talk) 15:17, 18 January 2018 (UTC)
If consensus on P17 is to treat it as "sovereign state", then the status quo for Scotland seems sensible to me. Scottish independence is a political movement that says Scotland should be independent; it does not say that it currently is independent, so generally speaking it's not really "disputed territory" - if and when it becomes independent, we can switch all the items easily enough, but doing that now wouldn't be appropriate.
In the case of Northern Ireland, as there is an autoreplace for England, Wales & Scotland it may just be an oversight that there isn't one for NI. If we look at the list of items using country (P17)Northern Ireland (Q26), it turns out about 80-90% of them are sporting seasons or sports teams. Only about 20 of the items are physical places like Portstewart (Q2239914). For comparison, there are almost 2000 physical places which are located in Northern Ireland but use country (P17)United Kingdom (Q145). The sporting ones could presumably be moved to country for sport (P1532) or similar (using whatever method we already have for sports in England or Scotland) and the others changed to UK for consistency. Andrew Gray (talk) 16:04, 18 January 2018 (UTC)

See also #Country of citizenship is Wales, above. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:36, 18 January 2018 (UTC)

P4020 format

Please can someone take a look at Property talk:P4020: the property should be formatted Roman (upright), not italics. -DePiep (talk) 15:25, 18 January 2018 (UTC)

Units with quickstatements

Adding unit with quickstatements2 is still unavailable? Xaris333 (talk) 19:01, 18 January 2018 (UTC)

See Help:QuickStatements, the pattern is 1U11573 for 1 metre (Q11573). Matěj Suchánek (talk) 19:38, 18 January 2018 (UTC)

And anyone knows why, if I am adding for example 34,708, the value in Wikidata is 34,708000000009345? Xaris333 (talk) 19:28, 18 January 2018 (UTC)

There is a problem that is diffcult to fix: Wikidata:Project chat/Archive/2018/01#Area precision. Matěj Suchánek (talk) 19:38, 18 January 2018 (UTC)
It shows 34,708000000009345 because the number is internally converted to a floating point numbers. @Magnus_Manske: hasn't yet good around to fixing the bug. https://bitbucket.org/magnusmanske/wikidata-todo/issues/65/quickstatements-should-not-add-decimal is the bug report and it might be worthwhile for other people who have problems with that to leave comments on the bug report or vote on it to show it's importance. ChristianKl ❪✉❫ 19:42, 18 January 2018 (UTC)

hilarri (Q1391249) and discoid stele (Q16678077)

I just stumbled upon this couple of elements related to discoid steles a type of funerary monument, particularly encountered in the Basque culture. Some of the links have local names, whereas others have the Basque name. The two elements should probably be sorted out so that one has the links with the local language terms and the other one has the Basque name. But doing the corresponding changes within Wikidata seems harder than in most cases I've encountered to date, so I wanted to ask the community.

Right now we have this situation:

hilarri (Q1391249) discoid stele (Q16678077)

My proposal would be to have something like this:

hilarri (Q1391249)
(local names) 		discoid stele (Q16678077)
(Basque name "hilarri")

Alternatively, both could be merged and the two articles in French Wikipedia fr:Stèle discoïdale and fr:Hilarri merged as well. Sabbut (talk) 09:38, 19 January 2018 (UTC)

Be bold to change, why not? --Liuxinyu970226 (talk) 15:19, 19 January 2018 (UTC)
@Sabbut: I don't think anyone else should have objections to your proposal; go for it! Mahir256 (talk) 16:45, 19 January 2018 (UTC)

Research in programming Wikidata

FYI, there's an interesting learning project at Wikiversity from w:Petrozavodsk State University about using w:SPARQL and Wikidata called v:Research in programming Wikidata that may be of interest to the community here. --Mu301 (talk) 16:42, 19 January 2018 (UTC)

Patron saint

patron saint (P417). Why value-type constraint (Q21510865) have only human (Q5) and religious and cultural festive day (Q375011)? Some place have angel (Q235113)s, like Michael (Q45581), who never been a human. Xaris333 (talk) 17:50, 19 January 2018 (UTC)

HTML formatting (italics, superscript and subscript) in labels

Hello, I just started a new subject on Help talk:Label#HTML formatting for labels. I report that the impossibility of HTML formatting for labels poses serious problems for Wikidata-based citations in sister projects such as Wikipedia (an exemple is en:Template:Cite Q). I would like to know who cares about it (we have millions of Wikidata items for scientific articles, it could be great to use it rather than filling bibliography templates by hand), and how to ask developpers to consider implementing a solution. Totodu74 (talk) 16:40, 17 January 2018 (UTC)

  • HTML tags aren't essential. What might be essential is formatting. If we introduce formatting we have to choose between the different semantics. In addition to HTML there's also Wikitext and markdown. I personally would prefer markdown. ChristianKl ❪✉❫ 18:33, 17 January 2018 (UTC)
  • Yes, any option of formatting would be good, but I saw some phabricator ticket with statement that this won't be possible. But this has more uses than citations, e.g. chemical nomenclature requires formatting for some prefixes (italics, small script) etc. Wostr (talk) 23:49, 17 January 2018 (UTC) phab:T139573 Wostr (talk) 23:52, 17 January 2018 (UTC)

Maybe a workaround? A qualifier property "italics" that could be added to any statement and that includes the regex necessary to identify the parts that have to be italicized? Same for superscript and subscript? If it's added to the item itself, it would adhere to the label, with qualifiers for language. However, it probably wouldn't help directly for Wikipedia citations (unless Lua is able to process stuff like that, which I don't know). --Anvilaquarius (talk) 09:08, 18 January 2018 (UTC)

Yes, that sounds much easier. --Anvilaquarius (talk) 21:48, 18 January 2018 (UTC)
@ChristianKl: while I like Markdown as well (specifically CommonMark (Q18019138)), I don’t think it’s a good fit for this use case, since it doesn’t support super- or subscript.
I also just realized that we already have one markup format with software support in Wikibase: LaTeX. For example:
\text{Evidence suggesting that }\textit{Homo neanderthalensis}\text{ contributed the H2 }\textit{MAPT}\text{ haplotype to }\textit{Homo sapiens} →  
[\mathrm N_2 \mathrm H_5]^+_2 [\mathrm N_4 \mathrm C - \mathrm N = \mathrm N - \mathrm C \mathrm N_4]^{2-}\text{: A New High-Nitrogen High-Energetic Material} →  
It might even kind of make sense to use LaTeX here because there’s a good chance the original article, including the title, was set in LaTeX as well. (However, we still face some difficulties – we’re in math mode, not in text mode, and we also lack some packages. For instance, the second example above should probably use mhchem’s \bond{} command.)
What do you think? —Galaktos (talk) 18:09, 20 January 2018 (UTC)

Titus Atilius Rufus is not American Treaty Shore

But somehow the article on en.wikipedia for this consul (which should be to Q47452755) is now attached to the Wikidata item "American Treaty Shore" (Q4745275). I'm not seeing an easy way fix this. Any suggestions? -- Llywrch (talk) 03:36, 19 January 2018 (UTC)

Purge the wikipedia page (which I've now done). Seems to be fixed. --Tagishsimon (talk) 03:40, 19 January 2018 (UTC)
Okay, that did it. Guess I should have tried that first. Thanks. -- Llywrch (talk) 01:36, 20 January 2018 (UTC)

Merge help

Can someone help merge Caesalpinia mimosoides (Q192364) into Hultholia mimosoides (Q42673836) - ie the genus should be Hultholia and not in Caesalpinia. Shyamal (talk) 15:49, 19 January 2018 (UTC)   Done I think it's done. Shyamal (talk) 07:02, 20 January 2018 (UTC)

These are two separate concepts, not to be merged. - Brya (talk) 08:30, 20 January 2018 (UTC)
This section was archived on a request by:
--- Jura 12:16, 26 January 2018 (UTC)

Date

Hello. Is there a way to have with start time (P580) the date "at the end of 17th century"? Xaris333 (talk) 17:48, 19 January 2018 (UTC)

As far as I know you can only say 17th century with century precision with earliest date (P1319) to express it in a structured way. You might add object named as (P1932) to provide the unstructured way it was expressed in the source. ChristianKl ❪✉❫ 17:58, 19 January 2018 (UTC)
How do you suggest to do it in Ayios Ioannis (Q4831288) --> patron saint (P417)? Xaris333 (talk) 18:06, 19 January 2018 (UTC)
refine date (P4241) was created for this. Sjoerd de Bruin (talk) 18:09, 19 January 2018 (UTC)
See Help:Dates#Qualifiers. --Kam Solusar (talk) 19:02, 19 January 2018 (UTC)
@Xaris333: "1800 +-10 years" (adjust "10" to suit your case), qualified with latest date (P1326) = "1800". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:25, 20 January 2018 (UTC)
This section was archived on a request by:
--- Jura 12:16, 26 January 2018 (UTC)

Centuries

In time values, why is the English format for "12th century" 12. century? This is not a standard constructon in English, and I was confused for years as to whether "12. century" means "1200s" or "12th century". 12th century (Q7061) says it's "12th century". - PKM (talk) 21:42, 20 January 2018 (UTC)

Tracked in Phabricator
Task T95553
Yes, I've wondered about this for a while - millennia are rendered the same way. Digging around on Phabricator it seems to be a known bug that would require an unexpected amount of effort to fix, which is a bit annoying. We could change it to "12 century" in English (Italian misses out the dot, for example) but not easily turn it into "12th century". Andrew Gray (talk) 21:52, 20 January 2018 (UTC)
it does not concern only English format, but also French format, which should be at least a "e" following the number like "15e siècle"... @Père Igor: --Hsarrazin (talk) 09:27, 22 January 2018 (UTC)
In fact, this is quite harmful, I have found quite a few cases where this caused wrong values. Even worse in my eyes is that a keyboard input "17th century" is processed as {"time":"+1700-00-00T00:00:00Z","precision":7" … } internally which WDQS and Listeria etc. output as "Jan 1, 1700" so that an event that took place in the year 1600 of which only the century is sourced can be displayed as a century later!! This undocumented inconsistency follows nearly all of JakobVoss’ guidelines for bad data standards! --Marsupium (talk) 15:49, 22 January 2018 (UTC), 16:18, 22 January 2018 (UTC)
This section was archived on a request by:
--- Jura 12:14, 26 January 2018 (UTC)

Converting Latin names to statement Q codes

Can some one tell me how I would go about finding the Q codes / statements of Latin names? I have the Latin labels / names only. Is there a tool which would automatically generate their Q codes - for a batch of 450? I can then use QuickStatements to add the Welsh and English names of mosses with references. Thanks! Sian EJ (talk) 20:02, 22 January 2018 (UTC)

Nothing you've written makes me think that the wikidata query service couldn't help, @Sian EJ:. Can you be more specifc: you say "I have the Latin labels / names only" ... what exactly is the set of records for which you want QIds? We can progress from there. --Tagishsimon (talk) 22:40, 22 January 2018 (UTC)
Sorry! The first column of my spreadsheet contains Latin names of different species of mosses eg Blasia pusilla; however, Quickstatements needs the code (Q17299165) so that I can add the label in other languages ("Cenddail-lys" in Lcy). Sian EJ (talk) 23:45, 22 January 2018 (UTC)
SELECT ?scientificname ?item WHERE
{
  VALUES ?scientificname { 'Arbor' 'Leo' 'Alces' 'Blasia pusilla'}  # add your 450 labels here
  ?item wdt:P225 ?scientificname .
}
ORDER BY ?scientificname
Try it!
--Larske (talk) 23:54, 22 January 2018 (UTC)
What Larske said. Come back if you need more, @Sian EJ:. --Tagishsimon (talk) 00:20, 23 January 2018 (UTC)
Fansastic! Many thanks! I'll add a couple and wait for ok. Sian EJ (talk) 15:42, 23 January 2018 (UTC)
This section was archived on a request by:
--- Jura 12:14, 26 January 2018 (UTC)

formatter URL (P1630) - distinct values constraint

I have opened a discussion at Property_talk:P1630#Distinct_values_constraint. --Rotpunkt (talk) 13:05, 23 January 2018 (UTC)

This section was archived on a request by:
--- Jura 12:12, 26 January 2018 (UTC)

Dates when a style or item "flourished"

Often working with costume and textiles, sources will indicate a date range when an item "flourished" or was "popular". Sometimes this is specifically different from the first appearance.

For example, my source states that fashion plate (Q5436787) first appeared in the last quarter of the 17th century, which I have used as the <inception> date, but "flourished from the late eighteenth to the early twentieth centuries". How should that date range be modeled when it doesn't conveniently fit into an era, and what property can be used? - PKM (talk) 20:39, 12 January 2018 (UTC)

I am staring at work period (start) (P2031) and work period (end) (P2032) and thinking that maybe they have potential, though they are worded as being people orientated. I remember them as being floruit start and finish, and wording being less people orientated, and wondering whether that is preferable.  — billinghurst sDrewth 12:57, 13 January 2018 (UTC)
Perhaps we need a new pair of properties parallel to work period (start) (P2031) and work period (end) (P2032)? I'm not sure how to label them - "popular from date" with aliases "common from date", "fashionable from date"? - PKM ([[User talk:PKM|Template:I::nt:Talkpagelinktext]]) 20:26, 13 January 2018 (UTC)
I'm considering whether we could handle this with a <significant event> "peak popularity (peak influence)" described as "period when the subject was most influential, popular, common, or fashionable" and then qualified with start/end times. Is this better than new properties? - PKM (talk) 20:14, 18 January 2018 (UTC)
I think that makes a lot of sense. Also, clients like Reasonator or Histropedia will immediately understand it as an event, and therefore work it into timelines. A controlled vocabulary could be developed to cover slight shades of meaning too -- anything up to total hegemony? Jheald (talk) 21:48, 18 January 2018 (UTC)
Good choice of expressive qualifiers on significant event (P793) too. And useful to have the event on one statement, rather than across two -- eg if one wanted to record that Black Death (Q42005) had peak severity in different places at differernt times. Jheald (talk) 21:57, 18 January 2018 (UTC)
@Jheald: Excellent point re: Reasonator. I hadn't thought of that. I'm going to move forward with this solution. - PKM (talk) 20:13, 19 January 2018 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Created heyday (Q47468023): period when an item was most popular, influential, widespread, or fashionable as <subclass of> key event (Q2245405) and time interval (Q186081). - PKM (talk) 20:21, 20 January 2018 (UTC)

source website for the property (P1896)

Do we accept Wikimedia projects as a value for this property? See this revert. --Succu (talk) 21:01, 19 January 2018 (UTC)

Yes we do since i) that's what was in the property proposal, where there were no objections and ii) that's, er, the source used. BTW, you seem to have accidentally forgotten to notify me that you were discussing my work. Again. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:22, 20 January 2018 (UTC)
This is not about you and your work. And it's not about what you proposed to be a source for MONA ID (P4758). It's a general question. --Succu (talk) 18:44, 20 January 2018 (UTC)
The diff you cited was an edit I made, to a property I proposed, reverting your (unexplained) removal of a value I included in the property proposal I wrote. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:23, 20 January 2018 (UTC)
The world is not spinning around you, Mr. Mabbett. And you did not answer my general question. So, why do we not add this property to all properties where we import values from Wikimedia projects? --Succu (talk) 21:56, 20 January 2018 (UTC)

Petscan question

I am doing a Petscan search that combines a Sparql query with a template search on en-wiki (stored as a PagePile) to find the pages don't currently have the template, but should.

I can get Petscan to run the SPARQL, and subtract the set of pages with the template -- but it insists on giving me the results as a list of Wikidata items.

From m:PetScan it seems I ought to be able to choose for the results to be a list of wiki pages instead, that I can then save and load into AWB. But everything I try still seems to give me Wikidata items. Is there anything I can do, just to get it to output a list of wiki pages? Jheald (talk) 01:57, 20 January 2018 (UTC)

@Jheald: at tab "Other sources" choose "From categories" in "Use wiki" field. Helped? --Edgars2007 (talk) 09:01, 20 January 2018 (UTC)
@Edgars2007: Thanks! That's exactly what I needed. I wasn't using the categories tab, so I never thought of that button to choose. But it works a dream. Result! Jheald (talk) 12:09, 20 January 2018 (UTC)

Second question (@Edgars2007: ?).

On Commons, I'd like to use Petscan to give me a list of Commons categories corresponding to query results, that aren't already in a Category:X. I find it's easy enough to get Petscan to find the categories already in Category:X, but is it possible to get it to find the categories that correspond to the query results? When I select "Use wiki" = "From categories", it returns a list that's a mixture of Commons categories and Commons galleries. What I really want is all Commons categories, as given by Commons category (P373) on the items that match the query.

Is there a way to generate such a list with Petscan, perhaps first creating a PagePile? Thanks, Jheald (talk) 19:41, 20 January 2018 (UTC)

Creating a redirect

I can't seem to redirect Q33323856 --> Q24204315. Thanks! Daylen (talk) 00:01, 23 January 2018 (UTC)

Done, using the (gadgets) merge script. --Tagishsimon (talk) 00:19, 23 January 2018 (UTC)
@Tagishsimon: Can you please move Q5340340 --> Q30271639 as well. I can't seem to get the gadget working. Thanks! Daylen (talk) 05:45, 27 January 2018 (UTC)
  Done by PLbot. You can also use Special:MergeItems if the gadget does not fit. Matěj Suchánek (talk) 08:38, 27 January 2018 (UTC)
This section was archived on a request by: Matěj Suchánek (talk) 08:38, 27 January 2018 (UTC)

Aliases of Pieter Brueghel the Elder (Q43270)

We had 160 English aliases of Pieter Brueghel the Elder (Q43270). Some of the 160 are lower and upper letter combinations of the name, some are last name first versions, some are other languages. Strangely Wikidata does not use references for Aliases, so I have no ides where they come from and it is hard to verify if they are real or made up. Any objections to cleaning that up? I have hard time imagining an item with more than 10 valid English aliases (someone will likely prove me wrong here), 160 is crazy. --Jarekt (talk) 04:53, 16 January 2018 (UTC)

  • Check the edit history. Why should we be missing some names while keeping others?
    --- Jura 05:42, 16 January 2018 (UTC)
  • While we don't have references for aliases we do have references for name (P2561) (and subclasses of it). In cases where it's desired to add references those properties can be used and the also specify more information about the name. ChristianKl ❪✉❫ 09:52, 16 January 2018 (UTC)
  • The aliases are bot added - see Wikidata:Requests for permissions/Bot/MerlIwBot. I'm uncertain what benefits you foresee in pruning the alias list, unsure what harm you see in the status quo. Your inability to imagine more than 10 valid English aliases and your assertion that 160 aliases is crazy don't seem to have a bearing on the matter one way or another. --Tagishsimon (talk) 10:11, 16 January 2018 (UTC)
  • The particular aliases for Pieter Brueghel all seem necessary for any solution that would look for mentions of this entity in English text (full-text search, named entity recognition, entity linking tasks), they are informative about spelling variations and various combinations that seem likely to occur in real published mentions of this person. --PeterisP (talk) 11:20, 16 January 2018 (UTC)
  • The ones in other languages are superfluous no matter how one looks at it. But a lot of the spelling variations are deadwood also, and have no practical effect, beyond confusing the reader. - Brya (talk) 11:55, 16 January 2018 (UTC)
  • It's certainly worth removing those worth only differences of capitalisation. Perhaps a bot could do that, as a regular cleanup job? I recall Jane023 having argued that foreign names of artists are needed, as art institutions vary in which versions they use in catalogues. And of course genuine spelling variants - far from being "dead wood" - are needed for aliases' prime purpose: to aid searchability. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:02, 16 January 2018 (UTC)
The problem with excessive number of aliases in a given item is that they are being displayed in the item and by the infobox templates. For example, I am working on new version of c:template:Creator that displays aliases from Wikidata. An infobox might be able to display some aliases but not 160, so it will have to either disregard all or pick random few. --Jarekt (talk) 14:29, 16 January 2018 (UTC)
Sorting out UI issues by throwing away data is somewhat baby out with the bathwater --Tagishsimon (talk) 14:36, 16 January 2018 (UTC)
Throwing away data is always wrong. And so is mistaking junk for data. Genuine spelling variants should be included, but a lot of the "spelling variations" here are deadwood. - Brya (talk) 18:10, 16 January 2018 (UTC)
Which ones? (I've fixed your broken indenting, again). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:24, 18 January 2018 (UTC)
Answer came there none... Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:52, 20 January 2018 (UTC)
When it comees to the infobox, you could just display the top X aliases. ChristianKl ❪✉❫ 17:08, 16 January 2018 (UTC)
Hi, why not use a collapsible list of aliases? See the infobox in Constituição brasileira de 1946 (ptwiki) for example. There's a lot of information from Wikidata in just one field, so we have collapsed them. Ederporto (talk) 18:07, 16 January 2018 (UTC)
I think “that displays aliases from Wikidata” is the problem. Aliases are not suited for what is needed by c:template:Creator |Alternative names= IMHO. We should rather put that information into statements which can have a datatype, ranks and references. Aliases are a bad fit for that. --Marsupium (talk) 15:04, 22 January 2018 (UTC)

Well I agree that simple changes in capitalization could be removed. We see this on multiple artist pages with "Xxx Yyy Zzz the elder" vs "Xxx Yyy Zzz the Elder". The others are needed. Jane023 (talk) 17:43, 16 January 2018 (UTC)

If things are working even remotely like they should be, it is nonsensical to include a German, French, etc name among the English aliases, since they will be included already among the German, French, etc aliases. - Brya (talk) 18:10, 16 January 2018 (UTC)
German and French aliases are used in English, just as my name, which is English, is used in German and French texts. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:24, 18 January 2018 (UTC)
The search algorithm makes no distinction between English aliases and the German, French, etc aliases. Having the same alias as an English alias and as a German, French, etc alias is pure redundancy. - Brya (talk) 06:55, 19 January 2018 (UTC)
Which algorithm, at what date and which interface?
--- Jura 07:06, 19 January 2018 (UTC)
The search algorithm that runs in Wikidata, any time in the past month, or so. - Brya (talk) 12:13, 19 January 2018 (UTC)
Yes, I object to that. Dutch has changed quite a bit since most of these people lived, variations are very common, translating it is very common and on top of that we have load of English speakers who tend to misspell names (if I would get an Euro for every time my name gets misspelled....). So yes, we end up with a lot of different aliases. Deal with it in the interface. Multichill (talk) 16:56, 17 January 2018 (UTC)
  Comment I boldly removed one en alias "Bauer=Breughel", which looks like a pseudo-equation and, IMHO, how is this person related to mathematics (Q395)? --Liuxinyu970226 (talk) 14:27, 21 January 2018 (UTC)
Was probably intended to be a double hyphen (Q1242843). --Marsupium (talk) 15:11, 22 January 2018 (UTC)

Specify located in time zone (P421) (and similar properties) on every place or just the most expansive place only?

There are many properties for places such as located in time zone (P421) and mains voltage (P2884). Should these be specified for all places? Or just on the largest encompassing place only for which the property is true? For example, the whole of New York (Q1384) state is in the Eastern Time Zone (Q941023). So should we only specify New York (Q1384)located in time zone (P421)Eastern Time Zone (Q941023) or should all places (cities, towns, villages, counties, islands, highways, buildings, national parks, etc.) inside New York also have located in time zone (P421)? —seav (talk) 04:45, 17 January 2018 (UTC)

If you put located in the administrative territorial entity (P131) New York (Q1384) inside those items (cities, towns, villages, counties, islands, highways, buildings, national parks, etc.) wouldn't be enough? It's a suggestion, of course, based on your example. Ederporto (talk) 04:53, 17 January 2018 (UTC)
@Ederporto: that is my preferred method as well. Data consumers would just need to traverse up the located in the administrative territorial entity (P131) tree to determine the timezone (or another property) that applies to a more local place. —seav (talk) 02:57, 18 January 2018 (UTC)
In order to think in a cenario of "all" places with located in time zone (P421), I think only capitals of states, states or countries should have. Ederporto (talk) 04:55, 17 January 2018 (UTC)

OK. Unless anybody has any reasonable objection, I will be modeling data such that if a place's item's located in time zone (P421) or mains voltage (P2884) (or other similar property) is already specified in a parent or ancestor located in the administrative territorial entity (P131), then adding that same property on the place's item is redundant and may be deleted. In this case, data consumers can then traverse up the located in the administrative territorial entity (P131) tree if needed.

For example. If we already have the following statement: New York (Q1384)located in time zone (P421)Eastern Time Zone (Q941023), then the following statement is redundant and may (not should) be deleted New York City (Q60)located in time zone (P421)Eastern Time Zone (Q941023) as long as New York (Q1384) is in the located in the administrative territorial entity (P131) tree of New York City (Q60). —seav (talk) 06:45, 22 January 2018 (UTC)

Formatter URLs for DOIs

It seems to me that several of the formatter URL (P1630) for DOI (P356) should actually be third-party formatter URL (P3303). What do others think? Would moving them break any tools? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:09, 23 January 2018 (UTC)

  Support I think only the doi.org ones should remain as formatter URL (P1630) there, the others should be moved. ArthurPSmith (talk) 12:52, 23 January 2018 (UTC)
  Support, that seems clear to me. − Pintoch (talk) 19:40, 24 January 2018 (UTC)

OK, done. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:26, 25 January 2018 (UTC)

This section was archived on a request by: Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:53, 28 January 2018 (UTC)

Module:Wikidata handle novalue

On the Dutch Wikipedia the special value novalue is seen as a normal value. When getting claims for IMDb ID (P345) this results in IMDb id geen waarde, which creates strange IMDb links (and the space makes it even worse). The text in the infobox now becomes (en) waarde/ IMDb-profiel. I could urlencode it, but that gives a wrong url. Is there a way to let nl:Module:Wikidata handle this as if the value was empty? An example of how it now looks is: nl:Anna Maria Snoek. Mbch331 (talk) 17:38, 22 January 2018 (UTC)

Different language versions of the Wikidata module have implemented different "solutions" to this problem. In frwp fr:Module:Wikidata you can add a parameter excludespecial=true to the call and in the svwp sv:Module:Wikidata2 version you can add the parameter novalue= to suppress the "no value". No such possibility is described on the Module documentation for the nlwp module as far as I can see. If you don't want to (wait for someone to) implement a feature to handle "no value" in the nlwp module, you can modify the Infobox template as a workaround.
If you want to avoid the IMDb-link in the case of P345 returning Geen waarde, you can test for that particular value in the nl:Sjabloon:Infobox acteur.
Try replacing the line

| other2 = {{#If:{{{imdb|}}}|{{en}} [http://www.imdb.com/name/nm{{{imdb}}}/ IMDb-profiel]|{{#if:{{#invoke:wikidata|claim|P345}}|{{en}} [http://www.imdb.com/name/{{#invoke:wikidata|claim|P345}}/ IMDb-profiel]}}}}

with this line

| other2 = {{#If:{{{imdb|}}}|{{en}} [http://www.imdb.com/name/nm{{{imdb}}}/ IMDb-profiel]|{{#if:{{#invoke:wikidata|claim|P345}}|{{#ifeq:{{#invoke:wikidata|claim|P345}}|Geen waarde||{{en}} [http://www.imdb.com/name/{{#invoke:wikidata|claim|P345}}/ IMDb-profiel]}}}}}}

The result is as before when the value returned is anything else than "Geen waarde", but when that particular value is returned the effect is like there was no P345 at all and thus no strange link is presented.
--Larske (talk) 23:16, 22 January 2018 (UTC)
Thanks. I've gone for the infobox solution for now. Changing nl:Module:Wikidata is a lot more work as I'm not a lua guy. Mbch331 (talk) 15:28, 23 January 2018 (UTC)

Wikidata weekly summary #296

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Discussions
    • Closed request for adminship: Mahir256 (as successful)

Quick community survey: deletion of items about “relatively unknown persons” on their own request

Case description: A couple of months ago we had the following situation: there was an item about a human subject, who requested item deleted via another editor whom they trust. The item was formally notable according to our notability criteria, since it was equipped with three serious external identifiers (#2 of WD:N); however, it did not have a sitelink (#1 of WD:N), and it was also not backlinked from any other item (#3 of WD:N, so there was no structural need).

The item was deleted by me after the case has been discussed with three other editors at WD:RfD, per (proposed) Wikidata:Deletion policy#Deletion of items (Phase I), which allows item deletion per “Any other common-sense reason for deletion not listed above.” I made a reference to English Wikipedia’s BLP policy at en:Wikipedia:Biographies of living persons#Deletion of BLPs of relatively unknown subjects due to the fact that we do not have a local equivalent, and rated the subject as “relatively unknown” (which has not changed since then, to the best of my knowledge).

Please note: There is no value in providing a link to the discussion of the case, thus I’d like to avoid doing so. In case of questions feel free to ask for more details or clarifications.

My questions to the community: What do you think about such borderline cases as described above? In which situations and to which degree do you think is a deletion upon the subject’s request appropriate—or not appropriate? Should we adopt our policies, and if so, how? I am happy to see any kind of input at this stage, to get an impression of the community perspective on the matter. Thanks in advance, MisterSynergy (talk) 18:40, 21 January 2018 (UTC)

  • I think it is appropriate. However, it would likely not be appropriate if they had a sitelink - the project article should have been deleted first.--Ymblanter (talk) 21:33, 21 January 2018 (UTC)
  • I agree it is appropriate with no sitelinks or backlinks; if there is a backlink (unless that also can be removed) I would recommend to remove only potentially controversial or private information, rather than the entire item. ArthurPSmith (talk) 05:16, 22 January 2018 (UTC)
  • I think in borderline cases err on the side of discretion and the privacy of the person doing the request. It would need to be judged case by case, but from what you say of the case you described it was a good decision to delete. Cheers Scott Thomson (Faendalimas) talk 05:45, 22 January 2018 (UTC)
  • Hi. I personally don't agree with this proposal; what is exactly a "borderline case"? I find it really subjective. In such cases, I rather suggest, as said by Arthur, to "remove only potentially controversial or private information"—and so (once again, IMO) there will be no more problem. Nomen ad hoc (talk) 22:51, 22 January 2018 (UTC).
  • If the person fulfills a structural need - e.g. being the author of a book in Wikidata or similar - I am having trouble with seeing how to delete the person. But we should have a policy that allows people to ask to have their date of birth or similar be made unspecific (i.e. instead of saying "May 20, 1965" it says "20th century", or instead of "Springfield, Oregon" it says "United States of America", etc.) --Denny (talk) 18:55, 23 January 2018 (UTC)
  • I can agree with Denny above about authors of books/articles and other publications – there is no reason to delete items describing them (also, I don't think that libraries etc. would be willing to delete such record of living author from the catalog based on author's request). What's more, I think that items describing living people should exist (i.e. should not be deleted on request) if there is at least one link to any serious database (e.g. library catalog etc.). However, any personal properties (date/place of birth, occupation, nationality, religion etc.) should be deleted on request if (i) there is no source provided or (ii) notability is not clearly established. Wostr (talk) 20:17, 23 January 2018 (UTC)
  • I don't think we have a policy that allow this at the moment. There's https://www.wikidata.org/wiki/Wikidata:Requests_for_comment/Privacy_and_Living_People which actually proposes a policy and the proposed policy would only allow deletion of individual statements by request. If you want deletion of whole items I would advocate that you add your option of how the policy should read as an additional option to the RfC. ChristianKl ❪✉❫ 11:12, 24 January 2018 (UTC)
  • Totally agree with Wostr. Nomen ad hoc (talk) 11:36, 24 January 2018 (UTC).
  • totally agree with Wostr. As a librarian, the idea of deleting the record of an author, needed for cataloguing a book, is completely absurd. For problematic properties (relative to pricacy), not needed to properly identify the author and differentiate them from homonyms, could be deleted or made vague (i.e. centuries, decades, instead of specific dates), when not already public (to me, data that can be found in a public library catalog IS public), on request from the person, after appropriate identification of the person who requests (OTRS ?). --Hsarrazin (talk) 13:21, 24 January 2018 (UTC)

Production - help?

Can anyone who speaks Japanese or Chinese better disambiguate the EN descriptions for production (Q739302) and production (Q36492844)? Or perhaps production (Q36492844) should be merged with something else? Thanks! - PKM (talk) 20:11, 23 January 2018 (UTC)

Bonnie and Clyde problems

These aren't interwiki conflicts per se, but result in a broken interwiki map. Can we find a better solution? Is there a tech&data discussion about this already?

Example: electrical resistivity and electrical conductivity are common concepts in all major languages. In en:wp they are combined, because one is the inverse of the other, into en:Electrical_resistivity_and_conductivity. We should be able to show interwiki links for the combined concept, somehow. Right now, instead, "Electrical resistivity" shows no english interwiki link, and "Electrical resistivity and conductivity" shows almost no interwiki links at all.

Sj (talk) 20:01, 24 January 2018 (UTC)

This situation is called the “Bonnie and Clyde problem”, and it is as old as the concept of interwikilinks. There has been this RfC last year in which the Wikidata community expressed the desire to have a technical solution for this problem, but we still wait for WMDE/developer input on the matter. —MisterSynergy (talk) 20:05, 24 January 2018 (UTC)
Many thanks, I appreciated that page's discussion. Updating the section title; it's great to have a phrase for it. I know a full solution is an unsolved problem, but perhaps we could at least return to the pre-wd method of capturing one direction of the asymmetric link. Sj (talk) 20:25, 24 January 2018 (UTC)
You can (and should) still use old style local interwikilinks in Wikipedia pages to accomplish that. —MisterSynergy (talk) 20:44, 24 January 2018 (UTC)
Or LUA. Sample at Q21533309.
--- Jura 21:01, 24 January 2018 (UTC)

On this, see a dedicated wikiproject : WD:XLINK. author  TomT0m / talk page 17:48, 25 January 2018 (UTC)

?action=edit

The ?action=edit isn't works when i try edit a page. 83.31.193.132

?action=edit - on the end of the URL - does nothing. There's an edit at the top of the page which allows you to edit the label, description or aliases. Individual statement boxes have their own edit links. New statements are added via the Add Statement link. Does that help? --Tagishsimon (talk) 18:31, 25 January 2018 (UTC)
Yes, but why it doesn't work? It works on wikipedia. 83.31.193.132
because this is not wikipedia. There you get the ability to free-text edit a section of, or the whole of, an article. Here, you're mainly constrained to select a priori properties, and add links to other records ... in that context, an edit page concept does not make much sense. It seems to work :) --Tagishsimon (talk) 18:46, 25 January 2018 (UTC)
Pages at wikidata are of a different content model (type: Wikibase item); whereas those at Wikipedias are wikitext.  – The preceding unsigned comment was added by Billinghurst (talk • contribs) at 26. 1. 2018, 10:32‎ (UTC).

Alexa rank (P1661) for a municipality?

I just saw that the User:Alexabot added a property Alexa rank (P1661) to a municipality (that of course has a web-page as a property) [10]. As a property for the web-page that may be fine, but the item is a municipality, not a web-page. Should I simply revert the bot, or is there a possibility to put the property to the property of the web-page? Grüße vom Sänger ♫ (talk) 22:25, 9 January 2018 (UTC)

Just revert the entry - sorry for it. --Tozibb (talk) 22:44, 9 January 2018 (UTC) (operator of Alexa bot)
Could your bot please revert all this false Alexa rank (P1661) I see in their history? Only items with instance of (P31): website (Q35127) should have it, and I don't see a single correct edit by your bot in this regard in the last 500 edits. It's a bit too much for a real person to clean up after a bot, that ran amok. Grüße vom Sänger ♫ (talk) 22:57, 9 January 2018 (UTC)
@Tozibb: I've reverted the last few of your wrong Alexa-spam, could you please clean up yourself after this mess you created? I think it's very unfriendly to do so much bot-vandalism and leave the clean-up to humans. Grüße vom Sänger ♫ (talk) 20:38, 10 January 2018 (UTC)
@Tozibb: Any reaction from you? Your bot hasn't done anything to clean up this mess. OK, it's as well not vandalising non-webpage properties up to now, but probably hundreds of those edits are still not reverted.
@All: Is there some other bot who can resolve this, or does it have to be done by humans? Grüße vom Sänger ♫ (talk) 05:13, 12 January 2018 (UTC)
One would at least use Petscan to query for items where a P1661 claim is not appropriate, and let this tool remove the claims. Nobody has to do that manually. —MisterSynergy (talk) 06:06, 12 January 2018 (UTC)
I am working on this - thinking about a SPARQL query to search for {P|1661} and delete the wrong ones
@Sänger: reaction  – The preceding unsigned comment was added by Tozibb (talk • contribs) at 09:56, 12. Jan. 2018‎ (UTC).
@Tozibb: What are your plans about this mess? When will you start to do anything about the havoc you created? You and your Bot haven't done anything at all since your vandalism, do you plan to do anything contructive at all? Grüße vom Sänger ♫ (talk) 13:46, 19 January 2018 (UTC)
@Sänger, Tozibb: Maybe add a constraint violation to the property? This would need to be quite broad and include any instance of a subclass of a part of Internet (Q75) to allow it on e.g. Imgur (Q355022) (instance of (P31) image hosting service (Q2721136)). (Tozibb, you should sign your posts with ~~~~.) Jc86035 (talk) 10:47, 14 January 2018 (UTC)
Also, the constraint violation should disallow things like sub-domains, though I'm not sure if that's possible. Jc86035 (talk) 10:50, 14 January 2018 (UTC)
As far as I see this, it's already there on Property_talk:P1661. Or how should I read the fifth clause under the big one? The bot just doesn't act according to it. Grüße vom Sänger ♫ (talk) 10:55, 14 January 2018 (UTC)
That constraint was added on Jan 10, so after this topic was started. —MisterSynergy (talk) 11:14, 14 January 2018 (UTC)
OK, so it was probably a reaction to the bot vandalism. It still hasn't happened anything in regard of reverting this vandalism besides my meagre efforts by hand. How hard is it, to revert a bot, that was wreaking havoc? Grüße vom Sänger ♫ (talk) 12:04, 14 January 2018 (UTC)
We talk about 2867 items that have a Alexa rank (P1661) claim, but not instance of (P31): website (Q35127) (or subclass thereof) (list of items). It is a matter of less than 5 minutes to set up a petscan batch that removes just all of them (the removal itself takes more time, but no human interaction is required; one just has to leave the petscan browser tab opened).
However, on second sight the instance-of relations imply that many of the items listed in the query above may qualify for a P1661 claim, so just removing all of them seems to be inappropriate either. I am not sure whether one could come up with a query that exactly finds the ones which are to be removed—it would probably be easier to remove in smaller sets, i.e. first from human items, then municipalities, then zoos, and so on. —MisterSynergy (talk) 12:15, 14 January 2018 (UTC)
Yes, on first sight and with a big thumb it looks like ~2-5% could be rightfully included in those items with an Alexa rank. That's not much, I think we should remove first all of those completely rubbish claims, like for municipalities, persons, universities and so forth, and leave it to those, who want this rank added to the more complicated cases to add them by hand. Or to invent a method, that fits the description. I fail to see why those bogus properties should be in those items any longer. (And regarding your 5 minutes: I don't have the faintest idea how a petscan-batch could be executed, that's nerdy insider stuff, nothing for me as a normal user. Grüße vom Sänger ♫ (talk) 12:37, 14 January 2018 (UTC)
Running petscan batches is a pretty common job here. First authorize Widar, then open Petscan and go to the “Other sources” tab, run a query such as the this one in the “SPARQL” field. When the results are loaded, you’ll see another input box in the upper right corner of the results list. A command like "-P1661" removes all P1661 claims from all checked items in the results set, as long as the tab remains open.
I have meanwhile started with the more obvious cases (humans, municipalities, etc). I still expect the bot operator to help here. —MisterSynergy (talk) 14:28, 14 January 2018 (UTC)
There are still almost 1 300 non-website items that has the property Alexa rank (P1661). Here is a query showing the distribution on instance of (P31). Note that some objects may have more than one P31:
SELECT ?p31 ?p31Label (COUNT(?item) AS ?count) WHERE 
{ 
  ?item p:P1661 [] . 
  OPTIONAL { ?item wdt:P31 ?p31 . } 
  MINUS { ?item wdt:P31/wdt:P279* wd:Q35127 } 
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en,de,fr". }
} 
GROUP BY ?p31 ?p31Label
ORDER BY DESC(?count) ?p31Label
Try it!
--Larske (talk) 07:30, 26 January 2018 (UTC)
Addition: Basically the same information can be found on this database report page. --Larske (talk) 07:51, 26 January 2018 (UTC)
Is there any objection to shifting the P1661 to be a qualifier on the website statement? It seems reasonable useful information to have, if its readily available -- eg making possible queries like: what are the 100 most accessed city websites in the United States? Jheald (talk) 12:13, 14 January 2018 (UTC)
AFAIK the idea was to add multiple P1661 claims with time, in order to be able to have the Alexa rank evolution of each website as well. This would be impossible with a qualifier approach. —MisterSynergy (talk) 12:17, 14 January 2018 (UTC)
We already have 167 items with more than one Alexa rank (P1661). Unfortunately the point in time (P585) qualifier is missing for many of them.
SELECT ?item ?itemLabel (COUNT(?p1661) AS ?count) (SUM(?with_P585) AS ?with_time) WHERE 
{ 
  ?item p:P1661 ?p1661 . 
  OPTIONAL {?p1661 pq:P585 ?t }
  BIND(IF(BOUND(?t),1,0) AS ?with_P585)
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
} 
GROUP BY ?item ?itemLabel
HAVING (?count>1)
ORDER BY DESC(?count) DESC(?with_time) ?itemLabel
Try it!
--Larske (talk) 07:47, 26 January 2018 (UTC)

Sources in the average Wikipedia infobox

One of the arguments againt Wikidata from Wikipedians seems to be that not all data on Wikidata has source. At the same time no Wikipedia has a rule that requires everything do be sourced. Are there any statistics of how many of the claims in the infoboxes of various Wikipedias are sourced? ChristianKl ❪✉❫ 22:48, 20 January 2018 (UTC)

Not a direct answer, but there are more than 200K articles lacking sources, some of them tagged as such for more than 10 years. See list here - PKM (talk) 23:36, 20 January 2018 (UTC)
That's not entirely true – according to pl.wiki VER policy every information has to be sourced — it is however very hard to accomplish (because of many articles without sources were created years before such version of VER policy came into force; AFAIR it was about 2011), but it is a common practice now to revert any new unsourced edit. Some statistics for pl.wiki are available here, but percentage of sourced articles (odsetek uźródłowionych) indicates only presence or lack of {{reflist}} template (and there are many situations where only some informations in article are sourced, and the rest is not).
As of infoboxes, it is more complicated, because most informations that appears in infoboxes should be sourced in the text (as infoboxes are just some kind of summary of most important informations from the article). But, I can give an example of chemistry infoboxes in pl.wiki: in 2012 there were over 5300 unsourced values (e.g. melting point, density, solubility etc.), now it's 1795. Unfortunately, chemistry infobox is the only one for which such statistics exist in pl.wiki. Wostr (talk) 16:26, 21 January 2018 (UTC)
  • @Wostr: "most informations that appears in infoboxes should be sourced in the text ". And how do you check what is present in the WP article when the infobox in doing the automatic data extraction from WD ? Snipre (talk) 08:04, 23 January 2018 (UTC)
    • @Snipre: at this moment this is not important for me, so I don't check it – (1) chemistry infoboxes are one of the few infoboxes in pl.wiki in which data is not a summary of the text, (2) only Commons category is imported right now (with local value with higher priority, so it's more about checking if local and WD values are identical or importing Commons cat if it's missing in infobox), the rest (CAS, PubChem, DrugBank) are used for data validation (checking if ids are correct, if articles is linked to proper WD items etc.), (3) this whole data extraction to chemistry infoboxes in pl.wiki is now more about checking if Listeria lists are sufficient tool to watch changes in WD items and how often vanadlisms or incorrect changes happen. AFAIK in other pl.wiki infoboxes only Commons categories, labels, images are being imported at this moment (and incorrect labels and images are always the cause of mass criticism, rightly or not and despite the fact, such situations happen quite rarely).
      As for this specific quote of me: I don't know how it will work in the future. Right now is is some kind of ideal article how-to or best practice to write articles in a way, that the introductory paragraph and the infobox are only a summary of the text. However, I don't think that WD will be used to e.g. 'write' article if there is no such article in pl.wiki – that is very unlikely to happen in pl.wiki and simply not possible in the foreseeable future (next few years). Wostr (talk) 10:38, 23 January 2018 (UTC)
  • For dates of birth/death of people, in most languages they are generally sourced in the text itself, not the infobox. plwiki and jawiki seem very thorough. Some wikis use odd places to add sources, e.g. edit summaries on dewiki. itwiki has some special template for the lead sentence of articles.
    --- Jura 17:19, 21 January 2018 (UTC)
  • Sources in the infobox? The infobox should only summarise the info from the article, it's not a substitute for text. And sources should be given to the information in the article text (or, if some menial data is changed manually in the box, like goals of footballers or such, in the summary). Sources in the infobox should be a rare exemption. Grüße vom Sänger ♫ (talk) 18:59, 22 January 2018 (UTC)
    @Sänger: And what happens when a stub is created in one wiki but more data in available in WD ? How can you check what is present in the WP article when you have an automatic extraction of the infobox from WD ? Snipre (talk) 08:04, 23 January 2018 (UTC)
    @Sänger: Your vision is outdated with wikidata: WD will offer much more data in the future than what is available in some articles. So if you built a infobox which retrieve all data available from WD, then there is no more connections with the articles. Just take the example of the creation of a stub in WP:de. If WP.en has already a good article on the subject and someone transferred the data in WD (with sources), then your stub article will never from the beginning contains all the data available in the WP:en and in WD. But the infobox with an automatic extract from WD can propose more data.
    So the question is to know if your wiki allows the use of automatic extraction from WD in the infobox. If yes, then your approach is to restrict the access to available data because WP contributors are not working fast enough to integrate all data available in WD in the article. Perhaps not the best approach when promoting information access. Snipre (talk) 08:04, 23 January 2018 (UTC)
also, we mustn't ignore the fact that, on some wikipedias, there is a habit to put basic biographical data in the infobox, and NOT in the text... thus, sourcing the data in the infobox is a need ^^
It's a policy of the concrete Wikipedia, how to deal with infoboxes. Small wikipedias will handle this in another way as big ones. What you describe looks like a Botpedia, something that is frowned upon in deWP. If anyone would dare to mass-create articles with a bot from WD (or anywhere else) that mainly consist of an infobox, s/he would probably be banned/blocked asap and most articles speedy deleted or at least moved to the user name space. If some WP would only have a few thousand articles yet, and stubs are needed very much, it would probably not this way. The headline here is about sources in the infobox, and I can only write about my WP, not something like the cebWP. Grüße vom Sänger ♫ (talk) 12:08, 23 January 2018 (UTC)
@Sänger: I was not referring to cebwiki or other bot-created articles. While trying to retrieve dates for biographical articles I found thousands of articles where the basic biographical data were in the Infobox of the article, and not in the article that was more "he did this and he did that"... even with long articles... ukwiki, ruwiki, thwiki, etc... have that kind of Infoboxes. I do not say that they use wikidata, I say that they use infoboxes in a way that requires sourcing for Infobox data... --Hsarrazin (talk) 12:28, 23 January 2018 (UTC)
@Hsarrazin:: yes, there are a few wikis (or articles in these wikis) where information is merely given in the infobox and not necessarily repeated in the article itself. Still, I think I had a hard time trying to find what references are used in some of these wikis.
--- Jura 09:59, 24 January 2018 (UTC)
Yes, it is sometimes very difficult to find the used sources... which is why sources in the Infoboxes should be important (I think of lua-fed Infoboxes which can give sources added to wikidata : ) --Hsarrazin (talk) 10:34, 24 January 2018 (UTC)
OK, here is a random sample from enwiki, 10 articles with infobox selected using Random article:
en:Zigzo - no sources in infobox
en:Todd Raleigh - no sources in infobox
en:C. Madhavan Nair - no sources in infobox
en:Basepoint Centres - no sources (and no content) in infobox
en:Tulovo, Serbia - no sources in infobox
en:Phazaca interrupta - one source for one statement
en:WQHA - no sources in infobox
en:William G. Harrison House - one source for one statement
en:Soren Baker - no sources in infobox
en:Myrsine lessertiana - no sources in infobox
Now random sample of relevant 10 WD items (no templates or categories) selected using Random item (omitting imported from xxwiki "sources"):
Caleta kalawara (Q15777739) - app. 50% of statements sourced
Plectris tuberculata (Q14854516) - mostly unsourced
PICO Building (Sanford, Florida) (Q14686335) - unsourced
J-PAS (Q20978033) - unsourced
Catecholamine levels in plasma and CSF in migraine (Q33591925) - well sourced
Beaverdam Ridge (Q31423713) - unsourced
Cytoplasmic free calcium concentration and inhibition in MLC after blood transfusion (Q43509137) - well sourced
Leucine rich adaptor protein 1 (Q43921457) - well sourced
Q13822142 - unsourced
Biomechanical model and evaluation of a linear motion squat type exercise. (Q46813953) - well sourced
Conslusion - average item on wikidata is better sourced than average enwiki infobox.--Jklamo (talk) 10:49, 23 January 2018 (UTC)
There is at least one use case for putting source information in an infobox: infoboxes that contain facts relevant to many articles. An example at the English Wikipedia is the template "Year in various calendars". Ideally, there would be a citation within the infobox for each fact. In the example I gave, there are no such sources, but at least there is a link to the appropriate Wikipedia article, where we can hope there are sources. Jc3s5h (talk) 20:30, 23 January 2018 (UTC)
This 10-article/10-items statistic shows nothing. 3 of 4 well-sourced items are in fact about scientific articles... Wostr (talk) 13:41, 24 January 2018 (UTC)
@Jklamo: J-PAS uses a Wikidata infobox. Although that doesn't change your conclusion as there's no references in that Wikidata entry. ;-) Thanks. Mike Peel (talk) 11:51, 26 January 2018 (UTC)

Governors of US territories who refuse their appointment

Multiple governors of the Territory of Idaho rejected their appointment. The question is how to indicate this using qualifiers. In addition, as all these people were appointed, does it follow that they are civil servants and not politicians ie should all appointed governors be considered civil servants ? Thanks, GerardM (talk) 10:21, 25 January 2018 (UTC)

https://en.wikipedia.org/wiki/Civil_service#United_States covers the distinction between civil servants and political appointees well; by my reading we would not term the Governor a civil servant. The opening line of https://en.wikipedia.org/wiki/Politician suggests we would tag all ITGs as politicians. no answer on how you represent a declined job offer; neither office holder nor position held seem appropriate. --Tagishsimon (talk) 11:02, 25 January 2018 (UTC)
We also have a parallel case (as yet unsolved satisfactorily IMO) in people who stand for election and win, but then decide not to take up their seat. There is also the slightly more distant version, such as with the recent Australian dual-nationality issues, where the courts later rule that someone never actually held an office — even if everyone had been acting all along as if they'd held it (in some cases for decades). Simply removing the position held (P39) statement seems inadequate here, so again I think we need some sort of modifier qualifier to note the exceptional situation. --Oravrattas (talk) 11:44, 25 January 2018 (UTC)
When you consider governors of territories, it should be obvious that at the same time there were regular governors in other states. The governors of territories were employed by the state and in order not to take a position, they often resigned. My problem with many "definitions" is that they are applied after the fact. Thanks, GerardM (talk) 12:51, 25 January 2018 (UTC)
I would have thought that they were not governors, but governors-designate, or similar. If they do not have the powers they didn't have the role. One would think that appointees who refused, died, ... in many roles are quite common.  — billinghurst sDrewth 12:29, 26 January 2018 (UTC)

Preventing to merge disambiguation items with regular items

I've noticed there have been many cases (and regulary there still are) of merging of disambiguation items with "regular" items. This is wrong, since disambiguation items must contain only disambiguation pages (cfr. Wikidata:WikiProject Disambiguation pages and Wikidata:WikiProject Disambiguation pages/guidelines. Is it possible to prevent this kind of merges by a filter which does not allow it, and provides a pop up which explains why this is wrong to the user who is trying to merge the items? Or do you think there are other technical means which are better? (maybe at least a "tag" at the edit, so regularly people can check edits with that "tag" in order to verify if that was correct). --Superchilum(talk to me!) 08:28, 20 January 2018 (UTC)

If an item links to another item in one of the statements (for example with "different from") then they can not be merged. Maybe that helps? --Lydia Pintscher (WMDE) (talk) 15:48, 20 January 2018 (UTC)
Not really. In the cases of disambiguation item, it works only when trying to merge, for example, a disambiguation (Q217465) with a surname (Q1158446) because they have different from (P1889), but it does not work when someone tries to merge the disambiguation item with another page. --Superchilum(talk to me!) 17:33, 20 January 2018 (UTC)
I'd like us to block merging any item with <instance of> Wikimedia list article (Q13406463) or Wikimedia disambiguation page (Q4167410) (or any subclass of Wikimedia internal item (Q17442446)) with any item that has any statements other than different from (P1889). These are very difficult to untangle after the fact. - PKM (talk) 20:30, 20 January 2018 (UTC)
"These are very difficult to untangle after the fact" yes, totally agree. That's why some sort of block is needed IMHO. --Superchilum(talk to me!) 21:14, 20 January 2018 (UTC)
There is a ticket for taking care of everything around it, but apparently WMFDE isn't interested ("need volunteer"): phabricator:T139912.
--- Jura 04:10, 21 January 2018 (UTC)
Even better: phabricator:T141845. Thanks. --Superchilum(talk to me!) 07:44, 22 January 2018 (UTC)
@Lydia Pintscher (WMDE), PKM, Jura1: from phabricator they say there is no need for a ticket since it's a gadget of Wikidata, which should be simply improved from here. --Superchilum(talk to me!) 16:53, 26 January 2018 (UTC)
It can be solved in whatever way possible. I don't think if it's useful for the community, it needs to be developed by volunteers. Isn't WMF providing IT services for the community as it's thought that this benefits from having paid staff?
--- Jura 19:49, 26 January 2018 (UTC)

Tool for easily searching for a specific ID (Viaf, Bnf, IMDb, etc.)

Is there an easy tool that allows to find which item uses an ID : for example, I'd like to find if there is an item with IMDb= nm0311623...

I think of an easy tool, like https://tools.wmflabs.org/wikidata-todo/relabel.php? by Magnus Manske, that would allow to easily find items, without having to make a SPARQL query each time ? So, is there something ? --Hsarrazin (talk) 19:42, 25 January 2018 (UTC)

resolver by … Magnus Manske ;-) —MisterSynergy (talk) 20:00, 25 January 2018 (UTC)
Yes, one can use Resolver, but because of the way it jumps automatically to the new page (and forces you to look up the property number), I quite often find it's easier just have a query in a tab, and cut & paste values into it, eg:
SELECT ?item ?itemLabel WHERE {
  ?item wdt:P345 "nm0000402" .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
Try it!
-- Jheald (talk) 23:42, 25 January 2018 (UTC)
Indeed, I thought Magnus Manske had done something for that :D - Thanks for the adress... and for the tool, Magnus.
too bad you can't keep the tool open, with persistence of last Property requested, to be able to check series of IDs ;) --Hsarrazin (talk) 09:55, 26 January 2018 (UTC)
You can query multiple values at once in SPARQL, like so. Anyway, I have added a "no autoload" checkbox to resolver. --Magnus Manske (talk) 10:19, 26 January 2018 (UTC)
Thanks a lot Magnus Manske :)
However, in this mode, I do not see any result [11]. Where should it be displayed ? --Hsarrazin (talk) 10:24, 26 January 2018 (UTC)
@Magnus Manske, Jheald, MisterSynergy:
Thanks a lot for Resolver : in fact, I found a way to get items in a new tab (by right-clicking on the button "Find wikidata item", thus keeping Resolver in the first tab, ready to change the ID I want to check... Perfect ! --Hsarrazin (talk) 21:00, 26 January 2018 (UTC)
@Maxlath: is it something your hub could do? − Pintoch (talk) 16:09, 26 January 2018 (UTC)
@Pintoch: you could indeed resolve such an id with the Hub, either by being redirected (to Wikipedia https://tools.wmflabs.org/hub/P345:nm0000402, or to Wikidata https://tools.wmflabs.org/hub/P345:nm0000402?site=wikidata), or by getting a JSON version of the request https://tools.wmflabs.org/hub/P345:nm0000402?format=json. If you have to look for a lot of those, I recommend configuring the Hub as a browser search engine. Alternatively, for those who enjoy using the terminal, you could use wikidata-cli wd query command: 
wd query --property P345 --object "nm0000402"
or for short, 
wd q -p P345 -o "nm0000402"
. -- Maxlath (talk) 20:02, 26 January 2018 (UTC)

Unified login

Is it just me, or does the unified login not work on Wikidata? I always have to login here again in addition to the wikipedia login. Agathoclea (talk) 19:41, 26 January 2018 (UTC)

Clearing your cookies mostly works. Sjoerd de Bruin (talk) 20:52, 26 January 2018 (UTC)

Creación de un nuevo actor de doblaje

Hola, quiero crear la ficha de un actor de doblaje, la verdad es que lo estoy leyendo todo, pero no soy capaz de encontrar como hacerlo. Alguien me puede ayudar por favor?

@Javi.abengozar: Empieza aqui / start here [12]. Vuelve si necesitas más ayuda / come back if you need any help. --Tagishsimon (talk) 23:24, 26 January 2018 (UTC)

10k 404 URLs for Cultural Heritage Agency of the Netherlands Art Collection

The Cultural Heritage Agency of the Netherlands Art Collection (Q18600731) has changed its website since we added 10,885 of its URLs as described at URL (P973) statements, resulting in HTTP 404 errors. 

SELECT ?item ?itemLabel ?inventory_no ?url  WHERE {
  ?item wdt:P31 wd:Q3305213 .
  ?item wdt:P973 ?url .
  ?item wdt:P217 ?inventory_no .
  FILTER(CONTAINS(str(?url), "http://data.collectienederland.nl/resource/aggregation/rce-kunstcollectie/") )
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],nl". }
}
Try it!

A URL which works in /some cases/ can be constructed from the inventory number (P217) ... but in some cases not:

I'm guessing we need to remove the URLs, but thought to bring the issue here for discussion before precipitate action. Beyond the search examples, above, I've not found another URL template to use. cc. @Multichill: whose bot, afaik, added the original data. thanks --Tagishsimon (talk) 00:35, 19 January 2018 (UTC)

Why not deprecate?
--- Jura 06:09, 19 January 2018 (UTC)
@Multichill, Tagishsimon:
I tried to understand this site, which is not easy with only Google translate to understant nl ;)
is it possible to build something with this kind of permalink adress ? http://data.collectienederland.nl/page/aggregation/rce-kunstcollectie/E735 (for E735) --Hsarrazin (talk) 18:43, 19 January 2018 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ I think this is the second time they changed their website breaking all urls. I wanted to wait and see what happened 

SELECT ?item ?itemLabel ?inventory_no ?url ?newurl WHERE {
  ?item wdt:P31 wd:Q3305213 .
  ?item wdt:P195 wd:Q18600731 . 
  ?item wdt:P973 ?url .
  ?item p:P217 ?invstatement .
  ?invstatement ps:P217 ?inventory_no .
  ?invstatement pq:P195 wd:Q18600731 .
  BIND(IRI(CONCAT("http://data.collectienederland.nl/page/aggregation/rce-kunstcollectie/", ?inventory_no)) AS ?newurl)  .
  FILTER(CONTAINS(str(?url), "http://data.collectienederland.nl/resource/aggregation/rce-kunstcollectie/") )
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],nl". }
}
Try it!
seems to return the right url's. Needs more checking before someone actually replaces it. And of course the references need to be fixed too. Multichill (talk) 19:00, 19 January 2018 (UTC)
On a tangent, @Multichill: what's with the multiple collection statements on Landscape near Alkmaar (Q28060147)? Where does that data come from? Confused. --Tagishsimon (talk) 12:17, 20 January 2018 (UTC)
@Tagishsimon: that's provenance data. Quite a lot of works have been stolen around WW2 by the German occupier and after the war returned to the Netherlands. Some were returned to the (heirs of) owners, some not and ended up in this collection. Multichill (talk) 11:26, 21 January 2018 (UTC)
@Tagishsimon, Hsarrazin: doing a new run of the collection including some more info and fixing the urls. Seems to leave us with a duplicate described at URL (P973), I fixed that too. Multichill (talk) 15:06, 21 January 2018 (UTC)
Thanks @Multichill: good work much appreciated. I'm sorry collectienederland.nl is keeping you unnecessarily busy :(. Not sure if you're aware, btw, that Magnus has a tool aimed at adding creators to painting items ... that's another thing we need to do. --Tagishsimon (talk) 15:45, 21 January 2018 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ For completeness sake the current query: 

SELECT ?item ?itemLabel ?inventory_no ?wrongurl ?righturl WHERE {
  ?item wdt:P31 wd:Q3305213 .
  ?item wdt:P195 wd:Q18600731 .
  ?item wdt:P973 ?wrongurl .
  ?item wdt:P973 ?righturl .
  ?item p:P217 ?invstatement .
  ?invstatement ps:P217 ?inventory_no .
  ?invstatement pq:P195 wd:Q18600731 .
  BIND(IRI(CONCAT("http://data.collectienederland.nl/resource/aggregation/rce-kunstcollectie/", ?inventory_no)) AS ?newurl)  .
  FILTER(CONTAINS(str(?wrongurl), "http://data.collectienederland.nl/resource/aggregation/rce-kunstcollectie/") ) .
  FILTER(?righturl=?newurl && ?wrongurl!=?newurl)
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],nl". }
}
Try it!
I've been working on missing creators for a while. See Wikidata:WikiProject sum of all paintings/Missing creator for an overview.
My bots add missing creators based on the description, but problem with Cultural Heritage Agency of the Netherlands Art Collection (Q18600731) is that they don't include the full name of the creator making it nearly impossible to find a match for the bot. The tool is probably inspired by https://tools.wmflabs.org/multichill/painters/index.php?collection=Q1043967 . Which you can use to create new painters (and the bot will update the paintings). I haven't advertised it that widely because we have quite a few people who just press the buttons and don't check all the links if it's actually the same person. Multichill (talk) 16:49, 21 January 2018 (UTC)
All the more reason to have an external-ID property rather than using P|973. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:00, 28 January 2018 (UTC)

Another 16k Bavarian State Painting Collections URLs give 404

Our lookup to the Bavarian State Painting Collections (Q812285) is kaput. --Tagishsimon (talk) 02:35, 19 January 2018 (UTC)

Works for me. --Anvilaquarius (talk) 08:46, 19 January 2018 (UTC)
Agreed. --Tagishsimon (talk) 08:50, 19 January 2018 (UTC)

and now DACS ID (former) (P4663)

trying to access artist's page through their ID access a blank record (see for {{Q47496}} - the adress for this artist is now https://www.dacs.org.uk/licensing-works/artist-search/artist-details?ArtistId=a57b7941-8e4e-4cf9-a531-6fb4d8f85ab7) --Hsarrazin (talk) 18:27, 21 January 2018 (UTC)

Documenting and describing PIDs

I'm currently at PIDapalooza 2018 (Q47486859) (website) and have been discussing with a few people how to document PIDs. As an example, I've created KoreaMed Unique Identifier (Q47489994). How could we improve that? For instance, we have no property to hold an example value. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:45, 24 January 2018 (UTC)

Side discussion

Side discussion about "free format descriptions"
At w:KoreaMed Unique Identifier? In general, the best way would be at Wikidata:Property proposal/KoreaMed Unique Identifier. Once created, Wikidata can be queried for countless samples.
--- Jura 18:28, 24 January 2018 (UTC)
@Jura1: I believe he's referring to things like digital object identifier (Q25670) or VIAF ID (Q19832964), which are in fact items about identifiers, and how we can describe them even if the properties themselves don't exist yet (for examples of such identifiers, see Mix'n'match's catalogs without Wikidata properties). Mahir256 (talk) 18:55, 24 January 2018 (UTC)
Here is another identifier at Wikipedia: w:Volcano Number.
--- Jura 19:00, 24 January 2018 (UTC)
Thank you, Mahir256, that is of course what I meant. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:33, 24 January 2018 (UTC)
I think we all understood that. So it doesn't change the answer to the question: where to place free format descriptions.
--- Jura 11:13, 25 January 2018 (UTC)
Who said anything about "free format descriptions"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:55, 25 January 2018 (UTC)

LCCN

  WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.

I added a Library of Congress identifier to this book How do I / we get the link to work? Thanks, GerardM (talk) 12:44, 25 January 2018 (UTC)

this ;) ... something to do with the difference between a work and an edition of a work? --Tagishsimon (talk) 12:55, 25 January 2018 (UTC)
I really wonder to what extend this difference helps or hinders.. What we should achieve is that people can read the books as they are in the Open Library and link them to library systems. So far it is utterly opaque to me.. PS thanks for your help
@GerardM:
this is an version, edition or translation (Q3331189), since there is a date of publication and a publisher (G.E. Eyre and W. Spottiswoode).
Ordinary users will ordinarily need to rely on "editions" to have access to read them : any traditional library catalog record is "edition-level". The "work" level has been introduced (with FRBR) to allow to connect all editions together (and also to translations, movies, etc. done after them), which was not possible before.
For clarifications between Works and Editions, please refer to Wikidata:WikiProject Books. --Hsarrazin (talk) 14:38, 25 January 2018 (UTC)
As a user I want to read a book. I do not want to choose when there is only one edition. When you talk about editions, often there is only one and consequently we do not even add an edition but add "edition" info on the book itself (as I did). To be perfectly honest, from a practical point of view this hinders and does not help. Thanks, GerardM (talk) 15:26, 25 January 2018 (UTC)
well, lua can take perfect care of this so that readers do not have to care about it ;) - this is a database, not a library catalog.
but, just for the possibility of re-editions, or translations, this is needed. If you don't want to get involved in the work of doing it, do not make the work more difficult for those who do it, please : each edition must have it's own item... --Hsarrazin (talk) 15:33, 25 January 2018 (UTC)
what does LUA have to do with this ? Also you want an edition for every book that is added... REALLY ??? Thanks, GerardM (talk) 15:58, 25 January 2018 (UTC)
  • lua ? I don't think readers will access wikidata directly to find books, will they ? lua templates (like w:fr:Modèle:Bibliographie can easily retrieve info from work and edition, and render properly, so that, neither the reader, nor you, will have to know how data is structured on wikidata ;)
  • as for the structure, I do not want anything... this is a whole Project and a collective effort, like all wikidata : please take the time to read it, and you may even discuss with people who take part in it... --Hsarrazin (talk) 16:20, 25 January 2018 (UTC)
@GerardM: Users don't come to WD to read books. It is the job of the coders to utilise the data to point to articles about books (at Wikipedias) and the editions (at Wikisources, or other places). That certain books only have one edition and others have many, or that they are translated into languages, and one of the editions may then spawn translations are in one or many editions is just a demonstration of the complexity of the system that needs to be determined. Hell, I am sure with Shakespeares works the translations have come a variety of different editions.  — billinghurst sDrewth 12:25, 26 January 2018 (UTC)
Indeed, there are many reasons why people do not come to Wikidata to read books. It is a shame. I have added a book by Angela Davis. Is this the right way to provide the information? Thanks, GerardM (talk) 01:11, 28 January 2018 (UTC)

Is there a Findagrave bot

Since English Wikipedia is removing links to Findagrave and the people removing the links refuse to migrate the data to Wikidata before deletion ... I wonder if we have a bot that finds the most probable matches and just needs a human eye to confirm or deny? Both Ancestry.com and Familysearch have this function. We migrated over the Findagrave ID if it used a standard template in external links in Wikipedia but most articles did not use that template. Most articles used it as a reference for the place of burial, and those are the ones being deleted now. --RAN (talk) 18:21, 27 January 2018 (UTC)

That looks like it only ran for flagged "famous" people, mostly actors. I add politicians, and they are almost never famous at Findagrave and I have to search for them manually. Can the bot be adjusted, I would love to have that as a project. --RAN (talk) 05:51, 28 January 2018 (UTC)

Use of P1282 OpenStreetMap tag or key

I am new to this property and have some questions for the use of it. Item dam (Q12323) uses OpenStreetMap tag or key (P1282) Tag:waterway=dam. Should then arch dam (Q890545) and barrage (Q350495) as a subclass of (P279) of dam (Q12323) have OpenStreetMap tag or key (P1282) Tag:waterway=dam?. In other Words, should all three items mentioned have OpenStreetMap tag or key (P1282) Tag:waterway=dam? I have tried to read the property proposal discussion but do not find a clear statement. Best regards Pmt (talk) 17:45, 28 January 2018 (UTC)

No, just the parent (or otherwise, most applicable) item. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:24, 28 January 2018 (UTC)

Batang Gembira

Batang Gembira (Q47483547), labelled in the Malay language as "Batang Gembira", has properties for both a book and an author. There is one interwiki link, ms:Batang Gembira. Can someone who speaks Malay clean it up, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:24, 28 January 2018 (UTC)

@Pigsonthewing: The article is about the book. Removed the author identifiers. Mahir256 (talk) 20:18, 28 January 2018 (UTC)
@Mahir256: Thank you. But instead of discarding that data, why not create an item for the author (or use an existing item, if there is one). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:23, 28 January 2018 (UTC)

Label bots

Do we still have bots that add labels for new sitelinks? I'm constantly adding labels to items that have been created months ago. Sjoerd de Bruin (talk) 09:36, 29 January 2018 (UTC)

I do it, but only in some case, where Upper/Lower case isn't a problem. --ValterVB (talk) 12:37, 29 January 2018 (UTC)

BREAKING CHANGE: wbcheckconstraints status parameter

Hi all!

This is an announcement for a breaking change to the default value of a parameter of the WikibaseQualityConstraints constraint checking API, to go live on . It potentially affects clients that use the wbcheckconstraints API action. (We are not aware of any such clients apart from the checkConstraints gadget, which is not affected.)

Recently, we added a status parameter to the wbcheckconstraints API action, with the intention that API users can declare ahead of time which results they’re actually interested in, so that other results don’t need to be sent to them: specifically, for most items the vast majority of results indicate compliance with a constraint, which we expect most users aren’t interested in.

On , we will change the default value of the status parameter to violation|warning|bad-parameters. We assume that most users of the API will only be interested in results that actually indicate problems, and this should significantly reduce the size of API responses. Users who wish to receive all results, regardless of status, should specify status=* in their API requests.

Our motivation for this change is that we want to enable caching of constraint check results, but don’t want to bloat the cache with tons of compliance and not-in-scope results that we don’t even show in the gadget. With the status parameter, we can store only problematic results in the cache, while still guaranteeing that the response we send is complete, since the request indicated that it only needs these results anyways. This also means that when we enable caching (see phabricator:T184812), only requests with status=violation|warning|bad-parameters will benefit from it.

Relevant tickets:

Please let us know if you have any questions. --Lucas Werkmeister (WMDE) (talk) 12:01, 29 January 2018 (UTC)

WikidataJS grant proposal

I have created a draft proposal to support the development of wikidata-sdk, wikidata-edit, wikidata-cli, wikidata-filter

If you have ever used those tools, or considered using them, or just like the fact that they exist, please review and eventually endorse the proposal :) -- Maxlath (talk) 14:40, 29 January 2018 (UTC)

Wikidata weekly summary #297

 
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Development
    • Constraint violations can now be checked on qualifiers and references (phab:T168532)
    • Implemented usage tracking deduplication to reduce database load (phab:T178079). This should not have any effect on what users see on recent changes and watchlists.
    • Redirects on client wikis that are connected to a Wikidata item can have a tracking category, if set up (phab:T185743). Thanks, Matěj!
    • Improved documentation of the pageterms query module (gerrit:406240). Thanks, Niedzielski!
    • Improved empty "content was:" in deletion logs for entities (phab:T184025)
    • Fixed links to external user pages in recent changes (phab:T183019)
    • Fixed user names beginning with a star sometimes being rendered as a list (phab:T182800)
    • Fixed constraint check results possibly showing up in the wrong language (phab:T185688)
    • Fixed ArticlePlaceholders possibly not showing up in search results (gerrit:406168)
    • Some of the developers attended to the Wikimedia Developer Summit 2018. You can find some notes on the Phabricator board

You can see all open tickets related to Wikidata here.


Automatically copied labels to other languages

I've encountered many items that have wrong/inappropriate LV labels, because they have been automatically copied from EN labels.

For example Andrey Kravchanka (Q27913439) Andrey Kravchanka (should be Andrejs Kravčanka most likely), Wolfgang Pepper (Q104816) Wolfgang Pepper (should be Volfgangs Pepers) and many, many others. History shows that these labels have been created by different users with Widar, copying the EN label to all languages using latin script. This doesn't seem appropriate - there are many other languages as well that (unlike English) transform/transcribe names to their own language instead of using the original transcription. It certainly would be much better if the labels would be missing (so that any users or tools see that they're missing) instead of filled with incorrect data; and at least in LV case copying the EN label is almost certain to be incorrect.

Eyeballing the data it seems that about half of all LV labels for person names are broken in this manner. What would be the proper action in this case?

I'm considering a script that removes the LV label in all cases where it matches the EN label and was created by Widar instead of manually; but it leaves open the question about how to prevent such wrong data in future, and how to identify other languages where such a fix would be necessary. --PeterisP (talk) 11:13, 16 January 2018 (UTC)

Should they not be moved to aliases, rather than removed? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:54, 16 January 2018 (UTC)
*Should* they be moved to aliases? I don't know what exactly is the policy about aliases in wikidata, would that really be appropriate? They are a valid translation in another language, but that's already reflected in the source data; but they are not valid alternative names in written Latvian (for example, they don't even conform to Latvian alphabet that does not use 'y' or 'w', but that's not the only issue). --PeterisP (talk) 15:24, 16 January 2018 (UTC)
Are those names used by Latvian people, and in Latvian texts? The last time we discussed a similar case, that turned out to be so. Also, note that, for example, the Wikipedia page lv:Džons_Lenons includes "John Winston Ono Lennon" as an alias, and gives one of his works as "John Lennon/Plastic Ono Band" Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:26, 18 January 2018 (UTC)
Can conversion of "English" to Latvian labels be automated?--Ymblanter (talk) 12:34, 16 January 2018 (UTC)
Sadly, it's a non-trivial issue, I believe it can't be fully automated in the general case. We could get an approximation that's works for most cases, but getting correctly translated names often requires human expert involvement (I myself am not sure in some cases, there are published guidelines but they're quite large and complex) and extra contextual information such as the origin of the name and how it is in the original language. For example, the EN label for Andrei Volokitin (Q276885) is Andrei (based on the Russian transliteration), but the LV label is Andrijs, according to the original Ukrainian Андрій, so a correct conversion needs to know and take into account that *this* Andrey comes from Ukraine. --PeterisP (talk) 15:24, 16 January 2018 (UTC)
Labdien, Pēteri. Jā, zināma problēma. Ir manā bezgalīgajā to-do listē. Daļa šo nepareizo nosaukumu rodas, jo daži dalībnieki ielikuši latviešu valodu to valodu sarakstā, uz kurām var kopēt nosaukumus no citas (parasti angļu) valodas. Un "aproksimācija" nederēs. Pats noteikti zināt, cik sarežģīti dažkārt veidojas vārdi. For others: yeah, they probably should get moved to aliases. And no - no automation. Each language has own complex rules, how names are written in Latvian. Problem also comes from nameGuzzler (or whatever it was called), where people have "marked" Latvian as good language, to which labels can be copied. --Edgars2007 (talk) 15:57, 16 January 2018 (UTC)
Yep, unfortunately not all users know that different languages use different names. In Finnish we have different rules for cyrillic names. Why would every languages use the English transliteration which "Andrey Kravchanka" seem to be. Stryn (talk) 16:17, 16 January 2018 (UTC)

I agree for actual transcriptions (from Cyrillic alphabets), but certainly not for the example where a German person name without any connection to Latvian should be "translated". That can only done for monarchs, popes and historic figures of that kind, certainly not for a normal mayor of a German city like in the example. If there is any indication that such a "translated" name has been used anywhere outside Wikidata/Wikimedia, go ahead and add an alias, and if it's indeed the mostly used name in the Latvian press, you may change the label, but translating names of virtually unknown people seems a really bad idea. --Anvilaquarius (talk) 16:21, 16 January 2018 (UTC)

That's not how you do things in Latvian. In Latvian, in media and other places you generally won't hear the original version of name (as main one). In media, you probably will hear the wrong version, but that's another story. And note, that we transliterate (maybe it's not the precise verb, but you got the idea), not translate names (very big difference). And this isn't some thing that we Latvian Wikimedians have come up with. It's written in law, that you do transliterate names, but don't leave original version. --Edgars2007 (talk) 16:30, 16 January 2018 (UTC)
In written Latvian language *every* usage of people names will be transformed. That's a core rule of grammar, partly to ensure that reading is possible (i.e. when you pronounce the written text according to Latvian rules, you hear something comparable to the original name; if any text in Latvian would contain "John" then it must be read/pronounced similarly to German word Johann, and if you want to pronounce it similarly as English John, then you must write the appropriate form Džons) and also to ensure that the words can be properly inflected to be used in sentences. The word endings carry significant syntactic and semantic meaning, so if we want to speak about John in Latvian, we need the whole range of systematic inflections Džons-Džonam-Džonu-Džona. The same applies for some other languages, at least Lithuanian. It absolutely applies to every single random German person with no connection to Latvia - if we want to talk about that person in Latvian language, if we use that name in a Latvian sentence, then the name will be, must be transformed. The LV label reflects that usage - the label should contain the spelling that will be used if that person is mentioned in LV texts. In some contexts, the original spelling might be shown as a footnote or once in parenthesis as a clarification. For example, "Volfgangs Pepers (Wolfgang Pepper) bija Augsburgas mērs." is a valid sentence describing that he was the mayor of Augsburg, but simply using Wolfgang Pepper would not be grammatically correct. This is not about press, it's about all written language. --PeterisP (talk) 17:19, 16 January 2018 (UTC)
"In written Latvian language *every* usage of people names will be transformed" Really? So you would change "Wolfgang Pepper" when translating the test "The painting was signed "Wolfgang Pepper"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:29, 18 January 2018 (UTC)
Yes, in common usage (e.g. if a literary work would contain the exact sentence you used) it usually would be expected to be translated. However, if necessary (e.g. when the translator expects that the exact original spelling is important to the reader) it might be preserved in this case as it is a literal quotation and not proper nouns within a sentence - "The painting had the signature of Wolfgang Pepper" must be translated, "The inscription under the painting said "Wolfgang Pepper"" would be permissible though would be rarely used (if there's a real need) and feel a bit weird, it's not the default choice. It's analogous to other literal inclusions/quotes of untranslated material in another language; as a rule of thumb, the original form of "Wolfgang Pepper" could be used in those (and only those) contexts where you could/would use the original form of Лев Толстой or 毛泽东 without attempting to transliterate them.  – The preceding unsigned comment was added by PeterisP (talk • contribs) at 23:32, 19 January 2018‎ (UTC).
So the claim "In written Latvian language *every* usage of people names will be transformed" is false. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:58, 28 January 2018 (UTC)
  • I can only support PeterisP. I think almost every language has its own transliteration and transcription rules. Adding English names automatically or semi-automatically by users that do not know transliteration/transcription rules in specific language is a huge mistake (I had to delete this Andrey Kravchanka Polish label, because in Polish it would be something like Andriej Krawczanka, but even I, as a Polish language speaker, do not know these rules good enough to be able to write this name 100% correctly; BTW this 'Andrey Kravchanka' shouldn't be kept even as a Polish alias, because it's just not correct in this language). Such edits (adding names based on English label) should be stopped and forbidden. Wostr (talk) 18:40, 16 January 2018 (UTC)
We could certainly add a “best practices” note in our help topics that names in English (or any other language using the Roman alphabet) should not be automatically copied to Polish. However, I don’t see any way to “forbid” an editor from making such an edit. I’m sure most of the copies are done in good faith. - PKM (talk) 23:44, 20 January 2018 (UTC)
That's not about Polish only, I think it's about most languages. And I know that most (all?) such edits are done in good faith. But good faith is not an argument (cf. en:Wikipedia:Competence is required) and no information is better than wrong information. In most wikis adding such name would require a source; unfortunately, it's not possible in WD to add a source to label, description or alias (BTW that's one of several reasons, why Wikipedia communities reject WD being connected to infoboxes etc.; that is a reason in pl.wiki right now, because in a few attempts to connect infobox to WD, there were too many such 'good faith' labels and pl.wiki users simply do not want to correct this and reject WD as a possible source of data in infoboxes, as it's much easier to maintain such data in local wiki). Wostr (talk) 00:13, 21 January 2018 (UTC)
  Comment There are also again and again problems where names from various languages that need translation like “Master of ABC” or “XYZ the Younger” are incorrectly copied. For this a blacklist of name parts could help perhaps. --Marsupium (talk) 15:20, 22 January 2018 (UTC)

I just stumbled across this script: User:Jitrixis/nameGuzzler.js. Isn't the main purpose of this script to fill item names without any thinking with some self-proclaimed transcription? I ping @Jitrixis: the author of that script to say something here. Grüße vom Sänger ♫ (talk) 18:34, 28 January 2018 (UTC)

I can only confirm things @PeterisP: and @Edgars2007: are saying. We would prefer empty label instead of English, French or other variants added automatically. Someone copied all French names of communes in France and these are not correct versions, but sometimes it is hard to tell it. I invite to look at some fun pages in Latvian Wikipedia how the transcription is done for French, Spanish, Polish. --Papuass (talk) 16:08, 29 January 2018 (UTC)
I kindly ask to permanently delete 'pl' language from this tool. Many edits made using this script would be a reason of indefinite block on some wikis. Wostr (talk) 19:02, 29 January 2018 (UTC)

Dams and reservoirs

Hello. Here is a (probably poorly written) request that returns a list of dams with no reservoir created (P4661) set but with reservoirs nearby according to their coordinate location (P625). Would you be kind enough to check a few of them and add the missing property if they do have a match? Some merging and splitting will be necessary in quite a few occasions. Thierry Caro (talk) 21:24, 25 January 2018 (UTC)

That's a pretty damn fantastic query, Thierry Caro. --Tagishsimon (talk) 22:23, 25 January 2018 (UTC)
Quite a nice to-do-list. Seems a lot of them still have instance of set to both reservoir and dam, since most Wikipedias do not separate these two concepts. Ahoerstemeier (talk) 22:56, 25 January 2018 (UTC)
Thanks. I've been separating dams and reservoirs whenever I had time. Thierry Caro (talk) 08:35, 26 January 2018 (UTC)

Hi, I also thought about that problem and I wondered about the whole made of the dam and its reservoir. I think the entity composed of the dam and its reservoir could have its class, if it does not have yet. Actually the following query shows a number of candidates, by searching for each dam who are part of an entity the classes of that entity : 

select ?dam ?bigitemLabel ?class ?classLabel {
  ?dam wdt:P279*/wdt:P31 wd:Q12323  .
  ?dam wdt:P361 ?bigitem .
  ?bigitem wdt:P31 ?class .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
Try it!

Conversely this is type of bigger entities reservoirs are part of: 

select ?dam ?bigitemLabel ?class ?classLabel {
  ?dam wdt:P279*/wdt:P31 wd:Q131681  .
  ?dam wdt:P361 ?bigitem .
  ?bigitem wdt:P31 ?class .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
Try it!

The most interesting are « power station » and « irrigation system ». « river » seems too large for the complex formed by a dam and its reservoir, for example. river basin district (Q132017)      is not specific enough to capture that, and in the other hand too specific because it’s defined by and for the European Union for administrative purpose. We also learn that lakes was part of some Olympic games, which may be a little an overuse of « part of » :) @Thierry Caro: Pulling that string, are aware of a wikipedia article about the complex formed by a dam and its lake ? This would be of interest to have a class for its item, maybe a « dam/reservoir complex ». An instance of that class would have two parts linked with reservoir created (P4661). author  TomT0m / talk page 10:31, 26 January 2018 (UTC)

Yes, I do agree that we need an item for this. For the moment we can only cover Bonnie and Clyde separately when it comes to dams and reservoirs. Thierry Caro (talk) 10:34, 26 January 2018 (UTC)
Perhaps “reservoir complex” as a label for this combined entity? I did some work on dams and reservoirs a while back, before the new property was defined. I’ll try to find time to help with these. - PKM(talk) 19:31, 26 January 2018 (UTC)
On further thought, the "complex" name should handle both dams/reservoirs and dams/flood control basins (common here in California). - PKM (talk) 23:30, 26 January 2018 (UTC)
And I think “significant event” ought to handle the Olympics? :-) - PKM (talk) 19:36, 26 January 2018 (UTC)
I'm probably missing the obvious, but what property should be used on a lake/resevoir item to point to the dam which created it? I'm thinking Lake Kariba (Q1047206) should point to Kariba Dam (Q1367609). I see , but find that unconvincing. A dam is not a toilet block nor a car park. --Tagishsimon (talk) 19:56, 26 January 2018 (UTC)
@Tagishsimon: We have Wikidata:Property proposal/dam currently going on. Thierry Caro (talk) 23:11, 26 January 2018 (UTC)
@Thierry Caro: Is there somewhere we can discuss a few usecase and the result with the properties you propose ? I don’t like inverse properties as it duplicates information, but maybe « dam » is intended to use in the « complex » item. On the other hand « part of » fits very well to link the complex to its part, so I’m not sure this is any useful. In the end, I don’t like discussing properties one by one as the problem is more something like « which items and properties do we need to model the complex » ? author  TomT0m / talk page 09:55, 27 January 2018 (UTC)
To be honest, has facility (P912) could be the inverse property. It would have to be renamed to 'equipment'. And I'd use it on all Wikipedia articles about reservoirs and watercourses that are impounded by a dam through Q46849415 and Q46849340. Thierry Caro (talk) 10:25, 27 January 2018 (UTC)
Mmmm. I guess the trade-off is between inflation in the number of distinct properties, versus the extent to which we shoe-horn disparate meanings into fewer broader properties. Impounded lakes don't really have a facility of a dam; or really have a dam as equipment. I can see clear risks of ambiguity and failure to convey meaning from the shoe-horn method. I'm not sure what the problem is with the addition of new properties - presumably that the same concept can be expressed using different properties, which frustrates queries, etc. Meanwhile, fwiw, I deplore the little-endian camp which deplores inverses. Inverses are useful in that they allow different and sometimes easier queries; and they allow QA by reconciling across the inverse. (They also, of course, give us more work to do, but that seems fine.) --Tagishsimon (talk) 12:08, 27 January 2018 (UTC)
By the way, the California Department of Water Resoources California Data Exchange Center (Q47499723) has 3-digit identifiers for "stations", which encompass a dam and its reservoir. If we add item for "dam complex" (or whatever we call it), we could add a property for this identifiers. Example here. - PKM (talk) 23:24, 29 January 2018 (UTC)

Q529207

Ronaldo (Q529207) needs semi=protection, please, due to ongoing vandalism. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:05, 29 January 2018 (UTC)

@Pigsonthewing:   Done for one week. Sounds like the sort of request that goes in WD:AN. Mahir256 (talk) 14:56, 29 January 2018 (UTC)
Yes - loss of situational awareness on my part (aka wrong tab), sorry. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:09, 30 January 2018 (UTC)

Format of references referring to an external database and distinct values constraint

Hi!

I would like to request feedback about references to external databases in Wikidata statements. I have been using identifiers in them, like this on Heidelberg University (Q151510):

has subsidiary
  Heidelberger Institut für Radioonkologie edit
1 reference
GRID ID grid.488831.e
add reference


add value

I prefer this format (rather than imported from Wikimedia project (P143)GRID (Q27768150)) because it is more informative: anybody can directly click on the identifier and see the record that was imported. When Wikipedia pulls these references from Wikidata, it is also easier for them to translate them to usable references on their side. Do you agree? If so:

  • How should we promote this reference format? Are there any guides that should mention this?
  • The distinct-values constraint (Q21502410) that is present on most identifier properties applies by default not just to main values but also to references. This means that references following this format will be flagged as violating the constraint if the same id is used in references on different items. This can be solved by adding the constraint scope (P4680)constraint checked on main value (Q46466787) qualifier on the constraint (which I have done for GRID ID (P2427)). If it is fine to include identifiers in references, then we should either add this qualifier on most identifier properties, or ask the developers to adjust the interpretation of the constraint, so that by default it only applies to main values.

− Pintoch (talk) 19:26, 24 January 2018 (UTC)

(ec) See Help:Sources#Databases for a proper format of database references. Regarding the latter: there is typically, but not necessarily a 1:1 correspondence between Wikidata items and external database entries. How often does it happen that database IDs are used in references of other items? —MisterSynergy (talk) 19:43, 24 January 2018 (UTC)
From time to time I use other items’ IDs in references, probably most often for statements with child (P40) etc. --Marsupium (talk) 07:06, 25 January 2018 (UTC)
I agree it is not a very common use case (that's probably why the problem has not arisen before), but I find it quite useful when stating relationships. − Pintoch (talk) 12:13, 25 January 2018 (UTC)
  • The distinct value constraint should be set in a way that it doesn't trigger when it's used in references. In case that isn't true, we have to investigate what we need to do to make it true. ChristianKl ❪✉❫ 22:37, 24 January 2018 (UTC)

  Notified participants of WikiProject property constraints

  Comment The constraint currently only checks for duplicate values within the same “scope”: no violation is reported if two different entities use the same value in a qualifier and a reference. Additionally, the default scope of the constraint is to check it everywhere, though as Pintoch mentions this can be adjusted with the constraint scope (P4680) parameter. Both of these things could be changed, if desired – I could change the implementation of the constraint to check within different scopes as well, and/or I could change the default scope. But note that the scope only affects where the constraint is checked at all, not how it looks for violations: if the implementation is changed to check distinct values between different scopes, and a constraint has a scope of “statement”, then a statement with the same value as a qualifier on another entity will still be a violation. --Lucas Werkmeister (WMDE) (talk) 11:35, 25 January 2018 (UTC)
I think that the current separation of scopes is sensible. I would just change the default scope to constraint checked on main value (Q46466787). − Pintoch (talk) 12:05, 25 January 2018 (UTC)
It seems that use as qualifier isn't compared to use as main value, so indeed. Default scope constraint checked on main value (Q46466787) would be good.
--- Jura 12:36, 25 January 2018 (UTC)
  Comment seems like data duplication as the identifier is probably also in the subject item. On the other hand, the main purpose seem actually to link to a webpage, and use the identifier property as a kind of subproperty of « url ». As it’s a qualifier, it’s not the same as the property used as a main property in a statement (maybe a good thing could be, for the devteam, to actually have two wikipages per property to reflect that). Seems a little weird overall but no formal opposition. One possibility could be to just point to the page with a « url » snak, this would spares the need to extract the idenifier and that would probably be that I would actually do. author  TomT0m / talk page 17:46, 25 January 2018 (UTC)
In the cases where the constraint is reported as violated by the gadget, there is often no duplication at all as the identifier is generally used as main value on a different item. I don't think reference URL (P854) should be used instead: by doing that we would lose the benefit of being able to adapt the formatter URL if the third party service changes its URLs later on. There are a lot of other benefits of using identifiers in references. For instance, if we realize that some identifier was added by mistake to an item, it is easy to check all the claims on the item where the same id is used as a reference and see if they should be deleted as well. Or if an external database contains duplicate records which were matched to different items, we can merge the items and easily see which claims come from which record. By tracking the use of the identifier property in references we can get useful statistics about the use of a particular database as a source across the entire Wikidata (which is harder if we need to rely on stated in (P248) or imported from Wikimedia project (P143) used in conjunction with GRID (Q27768150) or its editions such as GRID Release 2017-05-22 (Q30141628), or reference URL (P854) matching a particular regular expression…)
Overall I have to admit that I am quite surprised to find myself advocating for identifiers in Wikidata references - I thought the idea was much more widely accepted, and even encouraged actually. − Pintoch (talk) 18:53, 25 January 2018 (UTC)
I use identifiers in Wikidata references all the time, notably AAT IDs for items in the arts and GIS/Getty TGN IDs for geographic items. - PKM (talk) 20:06, 25 January 2018 (UTC)
@Jura1: tracked in phabricator:T185696 --Lucas Werkmeister (WMDE) (talk) 16:53, 30 January 2018 (UTC)

Historical visual diffs (beta feature)

Hi all!

Historical visual diffs were recently demoed to me and I propose that we enable them as a beta feature on wikidata.

You can find an example diff here: https://www.wikidata.org/w/index.php?diff=29023190&oldid=24804938&title=User:Addshore&visualdiff&diffmode=visual

They only work on wikitext pages (so not on items and properties) but I still really like the feature idea and would like to get it enabled here.

Thoughts? ·addshore· talk to me! 03:27, 26 January 2018 (UTC)

Well, it's really interesting and shows the diff more clearly (on the visual pov) - nice :) --Hsarrazin (talk) 10:27, 26 January 2018 (UTC)

Organizing consolidated city/counties

Large cities tend to be both a county and a city, e.g. San Francisco (Q62) and San Francisco County (Q13188841) (same issues for New York boroughs). They share the same border coextensive with (P3403), and usually have multiple instance of (P31) (e.g. city (Q515) and county of California (Q13212489)). This gets very confusing with the contains the administrative territorial entity (P150) - county of California (Q13212489) lists San Francisco County (Q13188841), but in reality it should list San Francisco (Q62)... or not. I propose, for the sake of clarity, the following system:

PROs: establishes a clear administrative chain of what is equivalent-to-or-a-subdivision-of what, makes data processing much more streamlined
CONs: makes automatic usage slightly more difficult in some cases, e.g. if a wiki article about California uses contains the administrative territorial entity (P150) to list all counties, clicking on San Francisco county would result in a missing page (except for a few botopedias), whereas the user expects to see San Francisco. The proper solution to this would be adding redirects to Wikidata - there should be a "San Francisco county" redirecting to "San Francisco". The redirect page should be listed in the county entity. --Yurik (talk) 06:38, 27 January 2018 (UTC)

Instance completeness

I did not realised it until know but its know possible with the query service and the quantity (P1114)   statement to find all the classes whom wikidata seem to know it knows all the instances. This is an interesting query so I put it here : 

select (count(?instances) as ?count) ?class ?number_of_instances {
  # this query computes all the classes Wikidata seems to know all the instances 
  ?instances wdt:P31/wdt:P279* ?class .
  ?class wdt:P1114 ?number_of_instances .
} group by ?class ?number_of_instances having (?number_of_instances = ?count)
Try it!

A derived query is interesting. It’s intended to find the instance incomplete classes we short miss of knowing all the instances : 

select (count(distinct ?instances) as ?count) ?class ?number_of_instances (?number_of_instances - ?count as ?missingnumber) {
  ?instances wdt:P31/wdt:P279* ?class .
  ?class wdt:P1114 ?number_of_instances .
} group by ?class ?number_of_instances having (?number_of_instances != ?count) order by ?missingnumber
Try it!

The result are quite unexpected : it seem the number of municipality of Japan (Q1054813)     _in wikidata is far bigger than it should be, for example, and that there is cases where the number is splitted wrt. a criteria, for example see Siemens ES64U2 (Q248035) by owner. Maybe the best way would be to put that in the owner item, I don’t know. Roughly filtering out the negative results and excluding qualified numbers (thanks to {{Query no qualifier}}) gives : 

select (count(distinct ?instances) as ?count) ?class ?number_of_instances (?number_of_instances - ?count as ?missingnumber) {
  ?instances wdt:P31/wdt:P279* ?class .
  ?class wdt:P1114 ?number_of_instances .
   ?class p:P1114 ?class_number_statement .
 ?class_number_statement ps:P1114 ?number_of_instances .
 FILTER NOT EXISTS {
    ?class_number_statement ?classnumberpq_pred [] . # a triple involving our statement ...     
    [] wikibase:qualifier ?classnumberpq_pred .                                                        # ... with a predicate ?pq_pred that is a qualifier
}
} group by ?class ?number_of_instances having (?number_of_instances > ?count) order by ?missingnumber
Try it!

And it gives interesting results like Municipi of Milan (Q610625) who lacks when I write these lines only one municipis it seems :) A nod to Wikidata:Property_proposal/fully_divised_into as this example is the same kind of example that lead to this query :) author  TomT0m / talk page 12:46, 30 January 2018 (UTC) EDIT : like below, added a « distinct » makes some cases looks way less problematic as it seemed. There may be several paths connecting instances to the class in « ?instances wdt:P31/wdt:P279* ?class .» for some items which made them being counted several times. author  TomT0m / talk page 18:22, 30 January 2018 (UTC)

And part completeness

A similar query, but for parts instead of instances if we know type of its parts with has part(s) of the class (P2670)   and the statement is qualified by the number : 

select ?item ?itemLabel ?type ?typeLabel ?number (count(distinct ?part) as ?numpart){
  ?item p:P2670 [ps:P2670 ?type ; pq:P1114 ?number ] .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  ?item (wdt:P527|^wdt:P361) ?part .
  ?part wdt:P31/wdt:P279* ?type .
} group by ?item ?type ?number ?itemLabel ?typeLabel
Try it!

Maybe also highlight some problems with the model for « catholic church » :) no problem actually if we remove duplicates in the queryauthor  TomT0m / talk page 13:52, 30 January 2018 (UTC)

Lack of admin attention

I posted on the administrators' noticeboard on 16 January, asking for deletion discussion to be closed. I posted a reminder on 20 January, and again on 27 January.

Not only has the discussion not been closed, but in the two weeks since my initial request, no one has posted a response on the noticeboard to say that they have looked at it, or that they will look at it.

Do we need more admins? Is there some other reason why admins (as a body; no individual is compelled to take any action, of course) are not attending to matters where their attention is requested? What can we as a community do, to ensure that matters needing admin attention are addressed in a timely manner? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:30, 30 January 2018 (UTC)

  • Several (of the more active) admins have participated in the discussion, so they might not feel to be in a position to close it now.
  • This is not an emergency situation where sysop tool use is required to prevent damage, so this might not seem to be very urgent. Maybe it is indeed a good idea to wait for the decision about the related property proposal (which also lacks admin attention) before this “ended” discussion at WD:RfD can be sent to the archive. The number of questionable items is <10 meanwhile anyway.
  • We do indeed need more (active) admins. IMO we do even need substantially more admins. There would easily be enough to do for 10 or 20 active extra admins, with specialization in different fields.
—MisterSynergy (talk) 16:05, 30 January 2018 (UTC)

Towards aligning Wikidata and Wikipedia infoboxes

We are currently proposing a project (GlobalFactSync) for a Project Grant that intends to synchronize Wikidata statements with the contents of infoboxes of respective Wikipedia articles. The idea is to use DBpedia as a middleware here. Since DBpedia has profound knowledge about the templates used for Wikipedia infoboxes, this allows us to compare the statements contained therein among multiple language versions and with Wikidata. One goal is to transfer missing statements to Wikidata, the second goal is to support Wikipedia editors when creating infoboxes. By doing so, the project targets improving the quality of information in both Wikidata and the Wikipedia universe. An early prototype already demonstrates the differences between language versions. For supplement the data of Wikidata we will enrich the statements with references from the articles and use the Primary Sources Tool to add the statements and primary sources. As Wikidata is heavily involved in the project, we are looking for your feedback. Feel free to discuss here or at the project talk page. You may also endorse the proposal or join the project as a volunteer. --Mgns (talk) 22:45, 19 January 2018 (UTC)

One of the big problems at Wikidata is that users keep importing stuff from Wikipedias: often enough this is seriously harmful. - Brya (talk) 08:52, 20 January 2018 (UTC)
In the proposal we outlined the problem of using Wikipedia as a primary source. Often Wikipedia is the only or the most obvious available source for facts when an item gets redacted by a user. Hence, in Wikidata it is quite common that simply a Wikipedia is given as the reference for a claim. For these statements and those without any reference given, we also seek to find valid primary sources. The risk remains that inappropriate data exists in Wikipedias and gets imported to Wikidata, but as the tool makes the data and its variations visible it can be fixed at multiple points and editors can make better informed decisions. --Mgns (talk) 20:57, 20 January 2018 (UTC)
I'm not very aquainted with DBpedia (Q465). But as far as I know infoboxes at enwiki were the main source of DBpedia. So how does the proposal helps to increase trustability of WD statements? --Succu (talk) 21:33, 23 January 2018 (UTC)
Since 5 years, we are extracting infoboxes of 120 languages and also have mappings, so all the data is extracted using one schema (which can be switched from DBpedia properties to Wikidata properties easily). This figure from Wikicite 2017 starting on slide 15 shows that 2/3 of WD is not well referenced. I see the greatest potential in transferring existing infobox references, which have been added by Wikipedians to WD. This would also build more trust. --SebastianHellmann (talk) 10:40, 31 January 2018 (UTC)
Unfortunately this proposal seems to ignore what Wikimedians usually mean with "quality". It's about data being sourced. Any data that passes through DBpedia loses those sources and thus it's "quality". ChristianKl ❪✉❫ 23:44, 24 January 2018 (UTC)
Fortunately thanks to this proposal we can address true quality issues. Quality means that there is agreement that a fact is likely to be good. This proposal allows us to focus our attention where we *know* the problem is problematic. This kind of quality beats sources hands down. I do not need a source to know that a difference is problematic. I do need reports that point to these differences. By concentrating on differences we improve true quality and not what is erroneously understood to be quality. Thanks, GerardM (talk) 10:26, 25 January 2018 (UTC)
@ChristianKl: A large part of the proposal is about extracting all references from infoboxes in 120 language version (plus find a few more with other tools). So it fixes exactly the problem you are mentioning. --SebastianHellmann (talk) 10:45, 31 January 2018 (UTC)
I think, we moved that part into the project plan. I will try to mention that more prominently, so there is no confusion. --SebastianHellmann (talk) 10:48, 31 January 2018 (UTC)

Ontology of general elections and local results

Hello! I am currently working on 2017 Austrian legislative election (Q19311231) and the adjacant objects (currently parties and personalties of the XXVI Austrian general elections). I am using 2010 United Kingdom general election (Q215622) and 2017 German federal election (Q15062956) as references in order to edit this object correctly and it works quite well for the electorial event itself and the results on a national level. Unfortunately I am experiencing difficulties finding the right way to add voting data on a constituency level. The Austrian governmental open data initiative supplies us with information about the electoral districts and how many ballots were cast for each participating party but I don't find a proper way to relate this data to the election event itself. There is an example of an electoral district of the German Bundestagswahl - Hamburg-Eimsbüttel (Q264130) - which connects the successful contestant to the current term. It does not include an example how to express how many votes were cast for each party during the election. I searched for election ontologies and found a promising example by the UK parliament but unfortunately as far as I can make out it also doesn't allow results in this granularity. Could you point my into the right direction in order to solve this problem by any chance? Thank you very much in advance --WalternativE (talk) 20:42, 23 January 2018 (UTC)

I can't point to any prior art on wikidata; don't know if such exists, one way or another. I suspect not. The UK Parliament ontology does, to my eye, provide for the granularity you're interested in - in its relations Candidacy Result & voteCount. That still leaves us with the problem of deciding how we're going to represent such information within wikidata. I'm afraid the right direction to be taken is probably to put up a suggested data model for discussion, which will identify properties we lack. By way of example, here's the Wikidata:WikiProject every politician/Political data model which seems to have done good things w.r.t. the representation of elected representatives and officeholders. --Tagishsimon (talk) 14:38, 24 January 2018 (UTC)
The the example you provided is indeed a very good guide for modelling the political landscape of a country within a term. If there is a way to get to a consensus regarding a similar data model regarding the electoral process I'm sure the richness of political information storable within Wikidata would be greatly improved. Unfortunately I'm quite a Wikidata (and for this regard: a linked data) novice. I could ask my professors for assistance in this matter and I'm naturally willing to contribute myself but I'm quite positive that it would be beneficial to have an experienced Wikidata user to organise the efforts. Would you like to take the lead in this endeavour? --WalternativE (talk) 21:04, 24 January 2018 (UTC)
Here's some of the prior art I suspected didn't exist, @WalternativE: Wikidata:WikiProject Finnish Elections & Wikidata:WikiProject Politics infoboxes. The former strikes me as the makings of a model, and the latter more biased to what information might be included in election infoboxes on wikipedias, than setting out a clear model. I'm not sure how moribund either project is. We should note also Wikidata:WikiProject Politicians, such as for the links to other projects at the foot of that page. I'm going to duck your 'like to take the lead' question entirely; but suggest if you're determined to proceed with this, we might put up an ad hoc Wikidata:WikiProject Austrian Elections playpen, steal nice things from the Finnish & Infobox projects, and use the EveryPolitician model format. --Tagishsimon (talk) 23:43, 24 January 2018 (UTC)
Thank you for the info! Looks really promising. I'll try to coordinate efforts with the local experts of the Austrian Wikidata chapter. Having this playpen would be nice - can we just start it and see where it leads us? We had an election related Wikidata workshop last year and this project would be a wonderful basis for additional ones I imagine (apart from beeing very useful) --WalternativE (talk) 20:52, 25 January 2018 (UTC)
Yes, start it, see what happens. That's the way. --Tagishsimon (talk) 21:01, 26 January 2018 (UTC)
Poking some political experts: @Stefankasberger: and @Andrew Gray:. Jean-Fred (talk) 18:20, 31 January 2018 (UTC)

Better property for frequency of first names

A few years ago, frequency of first names in the Netherlands, 2010 (Q21644845) and most frequent first names at birth in Rotterdam (1811-1913) (Q21142040) were added to significant event (P793) on given names, including the list position, number of uses and something that should be moved to subject named as (P1810) nowadays. I don't think they belong on that property though. The qualifier suggestions aren't that ideal due to this use. What is a better property for them? Sjoerd de Bruin (talk) 13:41, 28 January 2018 (UTC)

Hmm. ranking (P1352) has fairly narrow constraints, to limit the situations it can be used in. Widening its scope might or might not be desirable. series ordinal (P1545) says it is different from a rank, but offers only P1352 as an alternative. Also P1545 is only licensed to be used as a qualifier, not a main property. This is something we should address, because it is useful to be able to record rankings of things, without having to create a new property like Alexa rank (P1661) -- eg to record rankings of sub-territories by area or population. Jheald (talk) 13:58, 28 January 2018 (UTC)
(ec) proportion (P1107) also is only intended to be used as a qualifier. incidence (P2844) is for number of medical cases per year. So neither of these would work as a lead property. And besides, it's possible our source might only give rankings.
For reference, the item Marcus (Q4642) gives an example of the existing mechanism, using significant event (P793) as a lead property to hang qualifiers like series ordinal (P1545), quantity (P1114) and quotation (P1683) from. (As Sjoerd notes, the latter might better be handled by subject named as (P1810)). Jheald (talk) 14:29, 28 January 2018 (UTC)
Perhaps "ranked by" with an item for either a study or an organisation as value ? Jheald (talk) 14:33, 28 January 2018 (UTC)
So something like charted in (P2291)? (not sure if we should broad the scope of that one) Sjoerd de Bruin (talk) 16:22, 28 January 2018 (UTC)
For the "ranked by" property we habe review score by (P447) --Pasleim (talk) 21:38, 28 January 2018 (UTC)
Not really. Firstly that's a qualifier on review score (P444), not a top-level property. Secondly relates to a score (ie a number of stars) for a work, not for where an item appears in an ordering from most to least. Jheald (talk) 21:45, 28 January 2018 (UTC)
So, should we request some new propery or rename one? Sjoerd de Bruin (talk) 09:02, 31 January 2018 (UTC)

Changing the property of a statement

Tracked in Phabricator
Task T149905

Noobish question: Right after adding credits for The Hall of Presidents (Q3126038) using cast member (P161), I discovered the more appropriate property voice actor (P725). Is there any way to just change the property without having to reenter the data, or at least quickly export the data and reinput it with QuickStatements or something similar? Thanks! Trivialist (talk)

Hello, I'm afraid it's not possible to do this with the interface, and I'm not aware of any tool of gadget that could do this. Did you have a look at Wikidata:Tools? Lea Lacroix (WMDE) (talk) 14:35, 30 January 2018 (UTC)
There are no ready-to-use tools available for this task, one needs bot code to do such a move (e.g. with pywikibot via PAWS). For one item only with a couple of affected claims, it is likely much easier to move the claims manually to the other property. —MisterSynergy (talk) 14:43, 30 January 2018 (UTC)
Figured as much, so went ahead and just moved them manually. But thanks to all for the assistance. Trivialist (talk) 02:30, 31 January 2018 (UTC)

QuickStatements version 2 can remove (with -Q... at the beginning of each line) and add (with Q...) claims. You need to make a SPARQL query for your data, somehow get the data into a spreadsheet, and then it's easy to build the two lines necessary for the change. Since all this takes some time, it's only useful if you want to change more than a few items. --Anvilaquarius (talk) 08:50, 31 January 2018 (UTC)

That doesn't respect qualifiers or sources, though. Sjoerd de Bruin (talk) 11:24, 31 January 2018 (UTC)
That's right. I guess it could be done, but it's obviously very complicated. We do need a tool for that, but it's probably best to not have it available to everyone. --Anvilaquarius (talk) 11:50, 31 January 2018 (UTC)

Edit request at Property_talk:P356#EIDR

We have long waiting edit request at Property_talk:P356#EIDR related to DOI (P356) property. I was trying to make sense of it and it seems to be valid. Unfortunately I am not familiar with DOI (P356) property and can not evaluate it. Does anybody else understands it enough to evaluate it? --Jarekt (talk) 17:01, 30 January 2018 (UTC)

@Jarekt: this can be safely removed, as the official DOI resolver will redirect to this specialized resolver for these DOIs. See for instance https://doi.org/10.5240/FDBD-5E62-33F8-1876-7966-G which redirects to https://ui.eidr.org/view/content?id=10.5240/FDBD-5E62-33F8-1876-7966-G. In general there is no point introducing custom resolver (or properties such as EIDR content ID (P2704)) for particular DOI issuers: the whole point of the DOI system is to provide a uniform identifier on top of publisher-specific ids. − Pintoch (talk) 10:11, 31 January 2018 (UTC)

Query Helper on WDQS

Does anybody here ever use the "Query Helper" on WDQS? Am I right in thinking that it is positively unhelpful on anything but the very simplest of queries?

And yet, am I also right in thinking that the default is always to show it to someone, if they follow a link to a WDQS query that we may have posted, unless they have previously visited WDQS and dismissed it?

Would it be a good idea to make the default to be not to show the query helper, unless the query is so simple that QH is certain it can understand all of it?

Pinging @Smalyshev (WMF): for thoughts. Jheald (talk) 12:07, 31 January 2018 (UTC)

Hello,
The purpose of the Query Helper is to help people who are not familiar with SPARQL yet. It is not meant for people who are used to directly write queries, but it's an initiation tool for newcomers. It is possible to hide it, and this information is stored in the cookies, so next time one open a Query Service page, the helper won't show up again.
Also note that it is possible to share a direct link to the result of the query, if we want people to focus on the result and not the code (link icon, on the right side of the page) - one can anytime go back to the query by using the link "edit this query" at the bottom left.
The development team is continuing working on the Query Helper so it becomes even more helpful for newcomers, and handle better the complex queries. Lea Lacroix (WMDE) (talk) 14:33, 31 January 2018 (UTC)
@Lea Lacroix (WMDE): The problem is this. If I'm wanting to present WDQS on social media, it's not necessarily just the result I want people to see -- I may also want them to see how it was achieved with only a very few lines of SPARQL.
But this objective -- showing SPARQL can be simple and relatively straightforward -- is spoiled, if when the newcomer follows my link they see half the screen taken up with the query helper -- the query helper which didn't build the query, probably couldn't build the query, can't fully understand the query, and which will almost certainly break the query is somebody thinks they can try to use it to modify the query -- as well as taking the focus away from the text of the query itself, which if I am deliberately posting a link to it is what I want the readers to see.
The query helper needs to have much more of a sense of its own limitations. If it can't understand the query, it can't help, so it should not appear. Even if it does think it can understand the query, it should only offer to help, rather than pre-emptively take up half the query screen. At most, it should pop up an invitation to assist like en:Clippy. But it should understand its limitations -- it was not without reason that Clippy was generally despised and derided. Jheald (talk) 14:51, 31 January 2018 (UTC)
Thank you for the detailed feedback. I understand what you mean. I'm going to discuss about this with the developers. Lea Lacroix (WMDE) (talk) 15:46, 31 January 2018 (UTC)

Country of citizenship is Wales

Can country of citizenship (P27)Wales (Q25) be true for current day people? Sian EJ has been adding this claim to a lot of items using QuickStatements. As far as I know Wales isn't an independent country, but part of the United Kingdom. Example item: Peter Morgan (Q3376765). Mbch331 (talk) 12:27, 3 January 2018 (UTC)

The Wikipedia article on Wales says that it's a country in the first sentence "Wales (/ˈweɪlz/) is a country that is part of the United Kingdom and the island of Great Britain.". As long as the UK is still under the jurisdiction of the Lisbon treaty it's also arguable whether it's currently an independent country.
The subclass tree currently sees constituent country of the United Kingdom (Q3336843) as being states and thus there's no constraint violation. ChristianKl ❪✉❫ 12:36, 3 January 2018 (UTC)
Asking on Quora gave me the answer that there's no legal category of Welsh citizenship, so I'm now more doubtful. Especially the idea that place of birth automatically translates into a Wales citizenship is problematic and that seems how the QuickStatements are sourced. ChristianKl ❪✉❫ 13:12, 3 January 2018 (UTC)
Thanks both. To quote from the WP Talk page: The result of all these debates is that Wales is indeed a country. This has been confirmed in formal mediation. As it's a nation, a country, then the nationality of the people born there are 99% of the time Welsh. In the 1% I would include people who have moved in, and moved out again within a few years. But the defacto should be: if you're born in Wales, your Welsh - certainly up until the early 20c. Now, Wikidata does not have an item for 'Country of passport'; and I think it should. The UK did not exist until recently, therefore many 4 - 7c saints and Welsh princes and royalty will only have Wales as Nationality. Since the 1707 Acts of Union, when the UK came into existence, however, there is dual nationality - birth place gives Welsh, being part of that UK gives a second. Sian EJ (talk) 14:04, 3 January 2018 (UTC)
I think this is the same type of discussion as Netherlands (Q55) versus Kingdom of the Netherlands (Q29999). It is agreed that The Netherlands is a country, however the country of citizenship is The Kingdom of the Netherlands. I wasn't talking about historic situations from before the UK existed, but only about the period the UK exists. (If you died before the UK existed you can never be a citizen of the UK) Mbch331 (talk) 14:51, 3 January 2018 (UTC)
Now I follow your original thoughts, and thanks! And after 1707, you can be both, and WD can be flexible enough to accept that there are two very strong beliefs (as with many conquered nations). For example, Cerys Matthews is a "Welsh singer": and nobody would say that she is a "British" or an "UK singer"! However she does technically hold a 'British / UK passport', so that also needs to be addressed. Wikidata can respect both points of view, using nationality and citizenship, or dual nationality (in a loose sense). I'll start ploughing through the last century, and revert those born in Wales who would not be called Welsh (the 1%). Can we also start a new statement: 'nationality by passport'? Sian EJ (talk) 15:37, 3 January 2018 (UTC)
You seem to be conflating "nationality" and "country of citizenship" - it is the former which Wikidata currently does not have; the latter is "country of passport". How many of these people hold Welsh passports? Also, beware using place of birth as a determiner for either nationality or (United States excepted) citizenship - consider temporary visitors such as holidaymakers, seasonal workers and diplomats. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:42, 3 January 2018 (UTC)
Yes, I combigned both, as I did not know nationality (Q231002) existed! Thank you Andy Mabbett! It would be better to use "nationality" = Welsh; country of citizenship (P27) = UK? People giving birth while on holiday + diplomats etc are part of my 1%. Welsh passport = no, that's why I suggested a second / new Statement; but country of citizenship = UK seems fine. Sian EJ (talk) 18:12, 3 January 2018 (UTC)
nationality (Q231002) is no property, so you can't use it in that way. In the last property discussion about a property for nationality, we decided against having one. ethnic group (P172) might be used to label Cerys Matthews as Welsh but it's nothing that you should add simply because someone was born in a certain area. ChristianKl ❪✉❫ 18:47, 3 January 2018 (UTC)

@ChristianKl: NB - If nationality is not a property then we have only one ("country of citizenship") and this clouds the issue, unless more than one property can be added (in this case "UK" and "Wales"); this is second best. It is very odd that "nationality" isn't a property (link please)! Note also that "place of birth = nationality" was very true in Wales until the 20c and people are often described as "Welsh by birth" etc on Wikipedia as in the real world. ethnic group (P172) is NOT relevant to Welsh people, as has been reiterated by a number of national bodies including the language Ombudsman, Welsh Government etc! We are a nation, and our nationality is Welsh @AlwynapHuw, Llywelyn2000, Dafyddt:. Sian EJ (talk) 07:35, 4 January 2018 (UTC)

Wales is a Nation. My nationality is Welsh. These are facts that are not a matter of debate. I accept, that despite not liking it, I am a subject of the Crown (not a citizen) of the United Kingdom. Because of my health I am not able to travel afar, so I don't have or need a passport. Passport holding thingy won't cover my multiple identities. In the case of Wales and Scotland we have "national" entities approved and acknowledged by the Crown. Scotland and Wales (England too) are nations acknowledged as nations by the unwritten rules of the United Kingdom. Loads of images are available on this site because the NATIONAL Library of Wales uploaded them having been given permission to do so by the NATIONAL Assembly of Wales. The fact that England, Scotland & Wales are Nations under one crown is not the problem; the problem is that France, Spain etc can't be as relaxed as the UK and accept that they are nation states made up of many other nations. Wikidata's problem is that Wikipedia is available in lots of non-Nation-State languages. Spain France etc might not like it but Breton, Catalan etc have an independent Wiki site based on the fact that they are old nations with languages that separate them from their modern nation states. Wikidata has to recognise that or give up nation state language wikis.  – The preceding unsigned comment was added by AlwynapHuw (talk • contribs).
AlwynapHuw, you can harangue the community on this matter, however, the clarity works better if you can come forward with clear definitions as prescribed by the UK laws for what is Welsh nationality; and how is that separated from Welsh people, and again separate from Welsh nationals who were born elsewhere, or those who are born in Wales but are not Welsh nationals. At this point of time, we track citizenship; we don't track this somewhat vague feeling of identification of nationality. If there is no exact definition or means to clarify beyond what a person believes themselves to be, then it is probably not data for Wikidata.  — billinghurst sDrewth 12:25, 5 January 2018 (UTC)
(e/c) Citizenship in modern times is a legal/juristic concept, while nationality can have wider meaning. Regardless of whether there is a "Welsh citizenship"-act that I have not been able to find, you are most definitely a citizen of britain (and likely, but not necessarily of the EU). en:History of British nationality law Before 1948, everyone in the commonwealth was a en:british subject, after 1948 it became possible for countries/nations of the crown to create their own citizenship and nationality laws, see also en:Canadian Citizenship Act 1946, while the rest went on governed by the en:British nationality law and it's amendments and evolutions, overseas acts etc. While you could possibly have 3 citizenships (eu, british, welsh), again I see no proof that there is a recognised set of laws that govern what citizenship of Wales would be, and thus you do not have such a citizenship. This is indeed not unique. Puerto Ricans are USA nationals, but not USA citizens for instance (see Q3543841). I do think that this is something that needs fixing at some point in wikidata.. we're speaking of a national identity here, which for most people is a rather specific, but well recognised form of our very wide 'ethnic group'. But 'ethnic group' is just a bad shorthand for saying "a group of people which share soil, ancestry, culture, society or language". It's not wrong, but it's not really recognisable for people either. Imagine replacing all usages of race, nationality, 'descent' etc with 'ethnic group' in Wikipedia. That wouldn't be very helpful. Then again, you could say: "Wikidata has country of citizenship, because it's clear what that is. We don't have nationality, as there is no consistent definition for what nationality constitutes".. Hmm. qualifiers ? TheDJ (talk) 13:13, 5 January 2018 (UTC)
Both billinghurst and TheDJ have missed the point: WD does not have to use a criteria based on laws to define groups of people. The laws in Wales, since 1535 have been created by the oppressor, not by the people of Wales, who have NEVER accepted English dominance in Wales. Wikimedia believes in diversity - and that is about respecting all groups and minorities, not just the few. To say that we need to side with one group over another, just because they had sharper swords, and better killing machines, is rubbish, we are not the UN or state police, we are not bullies and oppressors but free thinking, inclusive Wikimedians. And as such, we need to find a way to accommodate all groups, and include a criteria whereby we can safely say that a person's nationality is Welsh. It is done on Wikipedia (nationality = Welsh); it is done in real life. If we can not do that on Wikidata then we are in conflict with other Wikimedia projects and we might as well give up Wikidata, and start our own projects which would respect diversity. What Sian says is quite correct, and manageable. Of course there is a clear "consistent definition for what nationality constitutes" (TheDJ), it is defined every day for many bodies such as sports teams. We would be 100% correct, in Wales, to say that a person is Welsh if they are born in Wales and have Welsh parents. If other countries can not say this, then they don't need to add 'nationality' on their biogs. We can. The second definition and criteria, which should be equally as clear is academic sources: if the National Dictionary of Welsh Biography lists and accepts that there is such a group as "Welsh people" then we must accept that. To say say that our nationality is "somewhat vague feeling of identification" (billinghurst) is ignorant and unhelpful. It is defined in Wales, we are Welsh, now move on. Llywelyn2000 (talk) 08:41, 6 January 2018 (UTC)
At this point the UK government defines what is their citizenship and we capture that data in that field ... in that English/Welsh/Scottish/Northern Irish people are UK citizens. Ethnic group exists where you wish to define "Welsh people", and many Welsh people are already so stated. There is no nationality property. The discussion is, and was, about citizenship fieluniod, and as such commentary should be taken in that context. The legal status of Wales or Welsh people within the United Kingdom, or the path to that paradigm, is not something that seems within the scope to resolve in this forum.  — billinghurst sDrewth 12:34, 6 January 2018 (UTC)
Wikidata doesn't store nationality but citizenship and citizenship is defined by laws. In addition we also have ethnic group that can be used to store that someone is a Welsh person. If you want to draw this information from Wikidata for your infoboxes you are free to show "Nationality = Welsh" for every person that has "ethnic group : Welsh" in Wikidata.
@Llywelyn2000: You are saying here that the definition of Wales nationality that Sian EJ used is wrong when you say "We would be 100% correct, in Wales, to say that a person is Welsh if they are born in Wales and have Welsh parents". This vagueness where the two of you use different ideas of what Wales nationality happens to be, is a problem that makes natioanlity a concept that's more vague than citizenship which is more clearly defined. For the sake of Wikidata that vagueness can lead to edit conflicts about thousands of items. Given that we want a property that's less vague we chose citizenship. We have also ethnic group for groups of people that aren't citizens and we require sources for that to have a way to resolve disputes.
You could even set some object has role (P3831)nationality (Q231002) qualifier on ethnic group, to say that in this case you are talking about nationality but that still means that you have to follow the sourcing standards we expect for ethnic group. ChristianKl ❪✉❫ 17:11, 6 January 2018 (UTC)
@ChristianKl: The difference between Llywelyn2000 and I does not exist; read what was said! To base nationality in Wales from 1707 onwards on place of birth gives us a criteria which is 99% correct - the remaining 1% being movement of people; until 1900 there was very little coming and going in Wales. Llywelyn2000, adds a second criteria - the birthplace of the parents, which gives us a watertight criteria without any "vagueness". He also adds two lists which could be added: names of Welsh people from the DWB and sports where nationality is defined. You may not think it important to create a property, just because one small country asks for it, but others do, as "citizenship" does not give the whole picture. If citizenship = uk, then Wikidata is in conflict with reality.
Secondly, as was said above, "WD does not have to use a criteria based on laws to define groups of people". Laws are created by one group; Wikimedia respects all groups (or should!) Both last two comments disregard our view on this point.
Ethnicity does not come into it, as I've already said.
Third, my suggestion of having two citizenship ("dual nationality - birth place gives Welsh, being part of that UK gives a second") has not been addressed by anyone. My suggestion is that the status quo (uk citizenship by law) is one, and that a second be created: "Local Citizenship" by birthplace of the person and recognised, sourced lists such as the academic, world standard of DWB.
A new property needs to be created - either "Nationality" or "Local Citizenship". Not to do so is not an option, otherwise Wikidata becomes unbalanced, does not recognise the real world and ignores academic work. And that means that an alternative project be formed - Llywelyn2000 might elaborate on this later, and measures taken to block Wikidata.
To quote responsiblecitizen.co.uk: "...many people will be offended if called British, but they consider themselves Scottish or one of the other nationalities. And the mistake of calling a Scot, a Welsh person or Irish person "English" is one which is even harder to forgive." If Wikidata continues to offend so many people, then that needs to change and needs changing now. Sian EJ (talk) 08:18, 9 January 2018 (UTC)
The same is for Catalans, Corse, Kurds, Bavarians, Basque etc. These are not citizens, a legal property, but ethnicities, a more subjective property without clear definitions outside the core. Is it of any legal difference if you are in, say Germany, whether you are Welch, Scotch or English? That's what citizenship is about. Sänger (talk) 13:36, 9 January 2018 (UTC)
Please note the heading of this thread includes the word Wales. If you want to discuss whether or not the Bavarians have a similar problem, please start a new thread. My request is regarding Wales alone. User:Sänger - you also use legality as your criteria ("These are not citizens, a legal property"), so I take it that you haven't read or you haven't understood my arguments that legality is not the sole definition of citizenship / nationality! I have defined clearly my alternative citizenship ("Local Citizenship") using academic and secondary sourced examples; what part of this is "unclear"? Your last question I take it, is an attempt to defend the existing country of citizenship (P27) property, which no one is attacking or querying in any way! Sian EJ (talk) 14:46, 9 January 2018 (UTC)
But why make some distinction for the Welsh, and not for the Catalan? Citizenship as a legal property regarding passports etc., is something quite good defined. Nationality, ethnicity, regionality etc. are not. OK, Cerys Matthews is quite explicit about her thanking the god she's welsh, but does it make any difference for her at the swiss border? The property country of citizenship (P27) is about just that, nothing else. The other stuff should be held completely separated from that. Dual country of citizenship (P27) is of course possible as well, as can be seen in the (for me quite unfathomable) disaster in the Australian parliament now, where unknown, involuntary and never used second citizenships now lead to baseless destructions of political careers. But Welsh, Catalan, Basque, Quebecoise, Frisian or Tyrolean are not country of citizenship (P27), but something distinct. Sänger (talk) 15:55, 9 January 2018 (UTC)
Some examples: Is Gerry Adams (Q76139) Irish, Northern-Irish or British? Is Abdullah Öcalan (Q191530) Turkish or Kurdish? Is Carles Puigdemont (Q4740163) Spanish or Catalan? And who is it to determine this? Grüße vom Sänger ♫ (talk)
Another fundamental issue to consider here, both for country of citizenship (P27) and more generally for country (P17) and other sub-properties of P17 is searchability, and the expectations of people constructing searches.
If somebody is constructing a search for all British people with a particular set of properties, eg all British painters with works in galleries within 100 miles of a particular place -- they don't expect Welsh painters not to be included; especially if they many not even realise if Welsh painters are not being returned.
That is why we are quite fierce about using quite a limited controlled vocabulary for P17 and similar -- and why suggestions to use what e.g. may be technically more correct values like United Kingdom of Great Britain and Ireland (Q174193) for UK citizens before 1927 can actually create considerable difficulties, even for the quite wary. Jheald (talk) 16:18, 9 January 2018 (UTC)
@Sänger: You say "The same is for Catalans, Corse, Kurds... " - Sian has asked for you to keep this thread focused on Wales; don't cloud the issue. You use "legality" once again as your sole criteria; yet, please read my comments (above). Citizenship is taught in the new Welsh Curriculum (introduced by the National Government of Wales) in every school in Wales whereby they are recognised as having Welsh citizenship as well as British and European. Alwyn (above) mentions that he has no passport, so please don't confuse the issue with points which have already been addressed in this thread. If you want to bring legality and passports, then create another property (Nationality by passport), as was suggested above.
Nationality / citizenship of people living in Wales from 5th century to the death of the Welsh kings were Welsh; no one disputes this. They continued being Welsh until the present day (discounting a few immigrants, who very quickly call themselves Welsh) in everyone's eyes - even the UK Governemnt; but their citizenship - on paper - was defined as being UK/British. Lloyd George was always called 'Welsh' and so is Ryan Giggs. There is no mechanism on wikidata to achnowledge this. The fact that another country forces their citizenship on us (on paper) does not change the fact that we are Welsh. If Wikidata sides with one large group rather than minority groups then it is in conflict with its own aims and objectives in its support of diversity. That is not how data works. If a group exists, then we can give several titles / names to that group. I'm Welsh, British and proudly European. Several titles; no problem.
Your paragraph "If somebody is constructing a search for all British people" - is not a "fundamental" problem; there are several ways over this (and I'm sure you could find other solutions). For example, Wales is part of the UK as well as being a nation in its own right (different to some other stateless nations and minorities) and by connecting both properties / and making Wales a subgroup of the UK, a search for British painters would also give you Welsh painters! When we search for churches in the UK, it brings up Welsh churches as well as Scottish, English etc. There's no difference, from data perspective.
So far, there have been no reasons given for not going ahead with Sian's suggestions; either create a new property 'nationality=Welsh' or allow a looser definition of 'citizenship' which allows dual citizenship. We are here talking about Wales, and the new properties could be limited to Wales. Otherwise Wikidata is going against the National Library of Wales' National Dictionary of Welsh Biography etc etc which state categorically that there is a data group called Welsh. Llywelyn2000 (talk) 08:59, 11 January 2018 (UTC)
This can't be "focused on Wales", this is a fundamental question about the property. Is this property about citizenship only, or should it be widened to autonomous regions? Grüße vom Sänger ♫ (talk) 09:20, 11 January 2018 (UTC)
Let's look at them one by one, starting at our feet, with Wales. If 'Nationality' is acceptable in the real world then it needs to be defined and used on Wikidata. If the real world says that so and so is Welsh, then Wikidata needs to be able to do that, otherwise it becomes political, biased and falls flat on its face. The percentage of live feeds coming in from WD into the Welsh language Wicipedia (cywiki) is higher than any other WP language. For example, we have 17,000 Wikidata Lists, Lysteria updated (twice as many as all other WP languages put together) - and live infoboxes, which at present use "| suppressfields=citizenship" as a lock to keep out incorrect Wikidata information eg Gwen John.
I know very little about some of the other nationalities mentioned above, and will not be drawn into specific cases. Wikidata has a number of properties which only accept good solid references from sources such as National Libraries, Academia, National Governments etc. And if they state that person X is of Y nationality, then we must accept that; and if there are two different solid references with two different nationalities, so be it! One could be defined by the stateless state, the other by the recognised state, and each Wikipedia could decide whether to opt for one or the other or both. If we don't recognise both groups, if we don't change the existing void on Wikidata, then we're on a very, very nasty slippery slope towards political censorship. Llywelyn2000 (talk) 16:53, 11 January 2018 (UTC)

Exactly! And as we already have Q188353 / Welsh people (Q188353) with the description 'Welsh people' = 'nation and ethnic group native to Wales' then a similar property should be created. As Q188353 exists, then so should Property = Welsh people be created. All arguments above that nationality can not be defined for Wales fall down, as we already have that definition. Sian EJ (talk) 07:48, 12 January 2018 (UTC)

We generally don't create properties on Wikidata because there's a specific items. We have millions of items and allow any user to create new items but we keep properties more limited. ChristianKl ❪✉❫ 00:42, 16 January 2018 (UTC)
Wikidata properties don't get created because you want a solution to model a single case. If you want to have a solution for a single case, then you have to use the available properites and that's the ethnic group property to store information about who's Welsh. Saying that you don't want to speak about any cases besides the Welsh one, is basically saying that you don't want to contribute any effort for the creation of a property for nationality. That's okay, but it won't help you with getting such a property. ChristianKl ❪✉❫ 00:42, 16 January 2018 (UTC)

I've been thinking about this a lot in recent months. I think the suggestion for a new property is really the only workable solution here. Running through it all in order -

  1. This is a hard problem, but it's undeniably important for representing UK people, both now and historically; even without a legally defined concept, we know what it means to say that Gordon Brown was Scottish and David Lloyd George was Welsh, and we know why that's more significant than saying they were both British. We need to work out a good model for it.
  2. The situation for Wales and Scotland is directly analogous, so whatever solution we come up with will hopefully work for both. It will also hopefully work for pre-1922 Irish people, a similar situation that we don't represent well.
  3. Identities can be complementary - eg saying you are Scottish does not automatically mean that you reject being British, and different people may prioritise them differently, or use them in different ways at different times. In some cases, stating one and not the other is an explicitly political act which may not be appropriate for us to do. This means we need to be able to record both the formal citizenship and the national identity..
  4. The options discussed here are to use country of citizenship (P27), ethnic group (P172), or a potential new property.
  5. country of citizenship (P27) does not work well because of the complementary nature of the labels - if we do Wales or UK then we would be making an explicit decision to prioritise one over the other, which is inappropriate for some contexts. If it is Wales and UK then we get into all sorts of problems with double-reporting and preferred values.
  6. ethnic group (P172) is a minefield. These national identities are not the same as ethnic groups, although historically they did line up a lot, and things will get confused quite fast if we start treating them as such. The property does have "nationality" as an alternative label, but I'm really not comfortable with forcing this concept into a box labelled "ethnic group".
  7. So... perhaps the best approach is to restart something like Wikidata:Property proposal/Nationality - but with an explicit constraint limiting it to only certain specified values, so that we keep it for contexts where that makes sense, ie distinct and widely understood national identities like Welsh, rather than trying to force historic states into the model as well.

Thoughts? I think that the suggested constraint might be the key to making a new property work - it means that we can use it for situations where we know it's appropriate and push back on using it in others, which gets around one of the key problems with the earlier failed "nationality" proposal. Andrew Gray (talk) 12:59, 12 January 2018 (UTC)

That certainly makes sense! @Pigsonthewing, Jura1, Richard Nevell (WMUK), Mike Peel, RexxS: Llywelyn2000 (talk) 08:33, 13 January 2018 (UTC)
First of all, thank you, Andrew. I agree with all of the points you make.
Can I then recommend that further commentators take the time to read the en:Nationality article, especially the Nationality versus citizenship and subsequent sections, which carefully examines the differences between nationality, citizenship and ethnicity - based on reliable sources and with instructive examples? Please note this distinction: "Conceptually, citizenship is focused on the internal political life of the state and nationality is a matter of international dealings."
If you want to consider criteria for constraints, then I suggest that nationality is an appropriate property where international relations indicate that nationality. To be useful, this cannot be simply confined to matters such as passports (mine says "European Union United Kingdom of Great Britain and Northern Ireland"), but should take into account broader concepts such as representation. An obvious example would be international sports teams, which have a variety of categorisations. When I played rugby union (Q5849), I would have been eligible to represent my country by playing for the England national rugby union team (Q378628) and the British & Irish Lions (Q733600), but I was not eligible to play for the Ireland national rugby union team (Q599903), the Scotland national rugby union team (Q945016), or the Wales national rugby union team (Q822877). That eligibility to represent a country in an international sports competition depends on my nationality, not my citizenship, nor my ethnicity. --RexxS (talk) 17:46, 13 January 2018 (UTC)
@RexxS: Hmm, I think the "international relations" thing here is potentially a bit of a red herring. w:Nationality is focused very much on nationality-in-a-state, and in that context the answer would be "British" for us both. The concept of "nationality" we're considering here is briefly touched on in the fourth paragraph of the lead ("In English and some other languages, the word nationality is sometimes used...") but it conflates it with ethnicity, which is a bit confusing. I'm not sure if enwiki has a solid discussion of this or not. Andrew Gray (talk) 22:31, 14 January 2018 (UTC)
@Andrew Gray: The idea that Kant having German nationality is not a widely understood nationality is interesting (there was no historic German state when Kant lived). Could you expand your view of what you think a widely understood nationality happens to be?
What about Italy? Which historic people have Italian nationality and which have Padanian nationality? How do you go about figuring that out? ChristianKl ❪✉❫ 19:26, 14 January 2018 (UTC)
@ChristianKl: Apologies, I wasn't clear. I meant to say cases where "multiple nationalities within a single country" is a widely understood concept. I explicitly don't think we should try and deal with the historic concept of German or Italian 'nationality' in the same way as the contemporary concept of Scottish/Welsh 'nationality'. These are different things and I don't think the same model will work well for both - which is one of the reasons the last property proposal had trouble. Andrew Gray (talk) 22:21, 14 January 2018 (UTC)
Well, I hope somebody will come up with some proposal for such cases. I'm continually frustrated that when I have a major painter described as "Flemish", or even English/Scottish/British from the 17th or 18th or 19th century, I really have no idea how I ought to be represented, and tend to just walk away. Jheald (talk) 22:42, 14 January 2018 (UTC)
No, I think you are clear. You want a concept of nationality after which Kant doesn't have German nationality. I'm not sure there's an existing notion of nationality out there that's defined in that way. There's a good chance that it will be misunderstood.
When it comes to the UK nationalities might be more clear but when it comes to Padanian nationality there's a lot more room for debate. The same goes for many modern conflicts. Is South Korea a nation? What about Taiwan? ChristianKl ❪✉❫ 23:51, 14 January 2018 (UTC)
The thread opening is, like often in the wikiverse, a bit too anglocentric, thanks Christian for straightening this. There can't be a solution just for the UK, this ain't an english project, this is international, and solutions for such questions have to fit the whole world. And don't give sport as an example, as the UK is in some disciplines the (for me not really fathomable) exemption, as it has special status for it's second-level country entities. (Strangely enough there are all-irish Ireland women's national field hockey team (Q762483) and Ireland men's national field hockey team (Q1672803), including Northern Ireland.) A solution has to fit Cyprus (north and south, greek and turkish), Russia (lots of nationalities in different kind of autonomy status), Africa (many colonial borders without consideration of ethnicities (aka nations in Europe)) and so forth. No, this is not just about Wales. Grüße vom Sänger ♫ (talk) 00:14, 15 January 2018 (UTC)
I agree we need an answer for all of these things - I just don't think that the answers will all automatically need to use the same approach! The Wales/Scotland situation is pretty unusual and easily distinguished from most of the other cases, and I had hoped we could deal with that bit seperately and get at least one thing sorted out. Apologies if I've overstepped - I certainly never intended to suggest it was a universal answer for all cases!
So, hmm, let's think about the overall problem as a whole. What sort of ways can people have "nationality" that we might need to describe? I spent a bit of time going over this and I can think of a dozen different things we might want to describe - list below. Some of these we have an easy way to describe, some we don't, but all of them are things we'll probably want to decide how to model. Andrew Gray (talk) 20:10, 15 January 2018 (UTC)
People from modern states
  • Case 1: person from a modern state, with nothing unusual.
  • Case 2: person from a modern state, with an unusual type of legal nationality (eg Indian overseas citizenship)
  • Case 3: person from a modern state, who also has a distinct national identity within that state (eg Wales in the UK)
  • Case 4: person from a dependency/territory of a modern state (eg Faroe Islands, Isle of Man)
  • Case 5: person from a region disputed or claimed by two modern states (eg Crimea)
  • Case 6: person from a modern state which has declared independence but not been widely recognised (eg Northern Cyprus, Somaliland)
  • Case 7: person from a modern state who was previously from a historic state (eg East Germany > Germany)
  • Case 8: person from a modern state which has become independent, and was previously from another modern state (eg South Sudan)
  • Case 9: person who is stateless
People from historic states
  • Case 10: person from a historic state or territory that formed a modern state (eg Kingdom of France > France)
  • Case 11: person from a historic state or territory that is part of a larger cultural group which is associated with a modern state (eg Prussia > Germany)
  • Case 12: person from a historic state or territory that has many modern successors (eg Austro-Hungary, Joseon Korea)
  • Case 13: person from a historic colonial state or territory (eg British India > India)
  • Case 14: person from a period or context without a clearly defined concept of statehood or nationality

What other cases have I missed? Do we need to break any of these down further - particularly #11? Andrew Gray (talk) 20:10, 15 January 2018 (UTC)

What's #3 in regard of recognized national minorities, like the Sorbs (Q146521) or Frisians (Q106416) in Germany? What about Scotland as long as there was no national assembly? What was the difference to an autonomous region like Catalan (besides the for me just ridiculous British exceptionalism, they always pretend to be something else from the rest of mankind) and Wales? And there was no Germany before 1871, it was founded then and not before, so nobody from before 1871 could be German. Before were other entities, but not Germany, the thought of one Germany was revolutionary in 1848. The same could be said about a lot of regions/nations/ethnicities, there are no clear boundaries between the different case. The only valid case for p27 is #1, and the same for every year in history, a state or a territory that formed a modern state is completely subjective and sentimental, not encyclopaedic. Grüße vom Sänger ♫ (talk) 20:50, 15 January 2018 (UTC)
My (limited) understanding of the Catalan situation is that yes, this fits case #3 - a group with a defined identity as a nation, which is not necessarily ethnicity-based, who are part of a larger state. "Distinct" here is subjective and not legal; it doesn't imply political autonomy.
We don't have to use P27 for each of these these, but they are all things we should try and understand how to model in some way. "Subjective and sentimental" doesn't mean they're not important to people! To be clear, am I right in thinking that you're saying #1 and #10-13 should all be "use P27:country at the time", so eg country of citizenship (P27):Kingdom of Prussia (Q27306) for Kant? Andrew Gray (talk) 21:08, 15 January 2018 (UTC)
The concept of national citizenship (Q42138) has to be of relevance to the person at the time of their life in the region of their life. Otherwise such stuff makes no sense. So that may be a problem for most colonies: They are artificial countries, usually with borders with no connection to those who lived there and their identities. And I don't care about the legalese of colonist nations, that tried to justify their land robberies;) Grüße vom Sänger ♫ (talk) 21:35, 15 January 2018 (UTC)
I think I broadly agree with "facts on the ground at the time" for P27 and historic people - it's what I've been doing for the past few years - but it doesn't help us with "why won't Kant show up in a search for German philosophers?", one of the other key issues raised here. Maybe the solution is to make sure these predecessor states are linked in such a way that a query can pick them up? I'm not sure the best way to do that without being very complex, though. Andrew Gray (talk) 21:48, 15 January 2018 (UTC)
Kant wasn't German in relation to country of citizenship (P27), he has written in the German language and was Prussian, with Latvian ancestry. If you put him in a "modern state" (whatever that may be) he could be quite a lot: German, Latvian, Russian at least. Grüße vom Sänger ♫ (talk) 16:49, 16 January 2018 (UTC)
The list of cases doesn't say what "from a modern state" means. As far as case 3 goes, does Padanian nationality count? Is there something like an Arabian nationality? Are Han Chinese a nationality? What sort of nationalities does China have? What's the relationship to Taiwanese nationalities?
If you say that "distinct" is subjective, how do we solve cases where the subjective opinion of different people differs? ChristianKl ❪✉❫ 00:32, 16 January 2018 (UTC)
Also, to go with "facts on the ground at the time" for P27 and historic people, what about say Alfred the Great (Q83476) ? We're currently going with country of citizenship (P27) Kingdom of England (Q179876), but that's anachronistic by at least a couple of generations. For his father we say Kingdom of Wessex (Q105313), but is that still right for Alfred, whose dominion ran somewhat wider? In some ways one might like to call him an Anglo-Saxon king, but is that something we should code in?
And how does one manage search discoverability, if one wants to find him in a list of British people? Is country (P17) any part of the answer? My understanding is that, for places etc, P17 should generally point to a modern country. Does it make sense to write Kingdom of Wessex (Q105313)country (P17)United Kingdom (Q145)? (Though that wouldn't help us find kings in England). I see that at the moment Kingdom of Wessex (Q105313) has no value for P17, while for Q83476 we currently have Kingdom of England (Q179876)country (P17)Kingdom of England (Q179876). Jheald (talk) 20:08, 17 January 2018 (UTC)
Actually, looking at it further, P17 doesn't necessarily generally point to a modern country. But (I think) it should be acceptable for it to do so, even in cases where the relationship is an anachronism, if eg qualified with something like determination method (P459) "present-day boundaries". Jheald (talk) 17:12, 18 January 2018 (UTC)
  •   Comment Okay, I am looking forward now to the guidance that needs to come out for British subjects of Commonwealth countries prior to 1949—Canada, Australia, New Zealand, etc.—for the nations when they are nations, and for when they were colonies. Similarly how we will be handling territories, are they of the (mother) nation? or of the territory or both? For these colonies or territories, how will we define whether they are of the colony, or the mother nation? Is it a place of birth? Or a place of belonging, and how will that be determined? This has to be done if this is to be implemented successfully at Wikidata, otherwise we are going to be half pregnant.  — billinghurst sDrewth 03:55, 17 January 2018 (UTC)
Yes, that's one of the challenges. People from pre-independence US colonies are another thing I'm wondering about (presumably the pre-unification Canadian and Australian territories will have similar issues). Andrew Gray (talk) 12:21, 19 January 2018 (UTC)

Source based approach

We could just try to look at specific sources that state that some is (e.g.) Welsh (and possibly why) and then try to determine the appropriate way to add a statement reflecting that to Wikidata.
--- Jura 08:07, 16 January 2018 (UTC)

I'm not sure if the following is what you had in mind, but below is the full set of nationalities retrieved by running the following query on the British Museum's SPARQL endpoint at https://collection.britishmuseum.org/sparql (Note it's not on the federated list, so despite "Try it" below this will not run from our own endpoint). This is the kind of list of values from sources that (IMO) we ought to be able to benchmark ourselves against, and match.
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
SELECT (GROUP_CONCAT(DISTINCT(?nat)) AS ?list) WHERE {
# SELECT DISTINCT ?nat WHERE {
   ?subj <http://www.researchspace.org/ontology/PX_nationality> ?nat
} 
ORDER BY ?nat
Try it!
I am not sure that either country of citizenship (P27) or ethnic group (P172) really work consistently for this lot. Even when modern states exist, it seems to me that identities like "Afghan" or "Trinidadian-and-Tobagonian" do not cleanly match to times when those states might or might not have been in existence as independent entities. But I don't think this is what people talk about when they talk about "ethnic groups" either -- in the context of the USA for example, "ethnic group" I think would tend to indicate something quite different, at the very least something a list of hyphenated-Americans, rather than a single nationality "American-USA".
In my view, especially for people, we should cleave quite closely to sources. If a source claims to be giving a value for "ethnic group" we should use ethnic group (P172). If it claims to be giving a value for nationality, IMO the best way forward would be to use a new property "nationality". And if there is no source then neither should be used. That does seem to me to be the best way forward, and I don't think we can jam it into the existing properties.
Anyway, here's the list:
List of nationalities used by the British Museum

Afghan; American-USA; Anglo-Saxon; Arabian; Argentinian; Asian; Australian; Austrian; Belgian; British; Buddhist; Burmese; Canadian; Caribbean; Catalan; Chinese; Czech; Danish; Dutch; Egyptian; English; Finnish; Flemish; French; Georgian; German; Hindu; Hungarian; Indian; Indigenous-Australian; Iranian; Irish; Israeli; Italian; Japanese; Korean; Kushano-Sasanian; Malaysian; Maltese; Mesopotamian; Mexican; Mongolian; Moroccan; New-Zealander; Ottoman; Persian; Portuguese; Roman; Russian; South-Korean; Spanish; Sri-Lankan; Swedish; Swiss; Syrian; Taiwanese; Thai; Turkish; Turkmen; African; Albanian; Algerian; Angolan; Armenian; Austro-Hungarian; Bahraini; Baluchi; Bangladeshi; Barbadian; Bavarian; Bengali; Bhutani; Bohemian; Bosnian; Botswanan; Brazilian; Bulgarian; Byzantine; Celtic; Chilean; Colombian; Congolese; Coptic; Costa-Rican; Croatian; Cuban; Cypriot; Czechoslovakian; Dominican; Elamite; Emirati; Estonian; Ethiopian; Etruscan; Fijian; french; Ghanaian; Greek; Guatemalan; Haitian; Hawaiian; Hun; Icelandic; Indo-Parthian; Indo-Scythian; Indonesian; Iraqi; Israelite; Italic; Jain; Jamaican; Javanese; Jewish; Jordanian; Kazakhstani; Kenyan; Kurdish; Kushan; Kuwaiti; Latvian; Lebanese; Libyan; Liechtensteiner; Lithuanian; Luxembourgeois; Lycian; Macedonian; Maori; Maya; Meroitic; Meso-American-Pre-Columbian; Mongol; Mozambican; Namibian; Native-North-American; Nepalese; Ni-Vanuatu; Nicaraguan; Nigerian; Norman; North-Korean; Northern-Irish; Norwegian; Nubian; Omani; Pakistani; Palauan; Palestinian; Papua-New-Guinean; Paraguayan; Parthian; Polish; Puerto-Rican; Romanian; Salvadoran; Samoan; Saudi; Scandinavian; Scottish; Serbian; Sierra-Leonean; Singaporean; Slovakian; Slovenian; Solomon-Islander; Somali; South-African; Sudanese; Tajik; Tibetan; Tunisian; Ugandan; Ukrainian; Uruguayan; Uzbek; Venezuelan; Vietnamese; Welsh; Yugoslav; Zambian; Zimbabwean; Zoroastrian; Anatolian; Bactrian; Bahamian; Belarusian; Bruneian; Burkina-Fasoese; Cameroonian; Filipino; Gaulish; german; Guinean; Guyanese; Hebrew; Hittite; Indo-Greek; Kashmiri; Kidarite; Laotian; Madagascan; Malian; Manx; Mauritian; Native-Hawaiian; Numidian; Palmyrene; Panamanian; Peruvian; Phoenician; Phrygian; Qatari; Rwandan; Shan; South-Arabian; Trinidadian-and-Tobagonian; Urartian; Yemeni; Aramaic; Azerbaijani; Belizian; Beninoise; Bermudan; Bolivian; Cambodian; Cayman-Islander; Commagenian; Dalmatian; Djibouti; Ecuadorian; Eritrean; Gambian; Gibraltarian; Honduran; Ivorian; Kiribatian; Malawian; Maldivian; Melanau; Norse; San-Marinian; Saxon; Seychellian; Swahili; Swazilandian; swiss; Tahitian; Tanganyikan; Tanzanian; Tongan; Burundian; Cape-Verdean; Meso-American; Senegalese; Yuezhi

Jheald (talk) 00:47, 19 January 2018 (UTC)
@Yair rand, Pigsonthewing, Máté, Vladimir Alexiev, Pharos: You all !voted against the previous Wikidata:Property proposal/Nationality. IMO being recently called out by User:Fram on en-wiki (diff) for the statement Arnold van Westerhout (Q18511480)country of citizenship (P27)Belgium (Q31) (when the original source said "Flemish") is a wake-up call that it is way past time that we sorted out a way forward to deal with this. I also don't think it's sustainable, when a source says eg "English painter", to have to look up which one of a seemingly endless number of legal entities may have constituted the UK at that particular moment -- none of which actually code accurately what the source was saying. Do any of you have a better suggestion, or should we just bite the bullet and open a new Wikidata:Property proposal/Nationality (2nd proposal), as at least a workable, least-worse, closest-to-the-external-sources option?
Also pinging @Jarekt, ArthurPSmith, Blackcat, Marsupium: @PKM, Thierry Caro, ChristianKl, Pasleim: who all also participated in the previous property discussion; and   WikiProject sum of all paintings has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead., since we have particularly strong resources for painters, which may make a useful case-study group to consider what may or may not work. Jheald (talk) 01:15, 19 January 2018 (UTC)
I have some rough notes on how P27 is currently being used in practice that I'd like to put together first - could we hold off for a little bit on the new proposal? I think it might help us focus better on some of the issues here and while we do have to fix this, we can afford to take a little bit of time over it to avoid getting stuck in a confused discussion about two or three different things. Andrew Gray (talk) 12:19, 19 January 2018 (UTC)
Thanks, Andrew. Rather than jump straight to a proposal, what I was thinking was perhaps i) canvass for options here, then ii) put those options to an RfA, then iii) propose new properties (or not) based on what comes out of that RfA.
I was hoping people might come up with some alternative suggestions, hence the call above. Jheald (talk) 13:43, 19 January 2018 (UTC)
For the record what I said in the discussion of that proposal was "Oppose: I don't think there is sufficient clarity about how this would be used". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:41, 20 January 2018 (UTC)
  Support --Marsupium (talk) 12:23, 22 January 2018 (UTC)
  Support -- Thanks User:Andrew Gray for showing that something very complex can become clear (it took a better mind than mine to define it so logically!) And to User:Jheald for showing that Wales is also recognised by the British Museum as a nation (see list)! I have faith that a better situation than the status quo can be gained, by respecting all parties concerned, and working it out mathematically / logically and not politically! Thanks to all. Sian EJ (talk) 16:00, 25 January 2018 (UTC)
  Support 100%. It is standard practice in art history and literature at least, to describe artists or writers as “Flemish” or “Italian” when there was no state called Flanders or Italy at that time, but we have what our sources say and that has meaning in context. Also consider “Irish” pre-1921. We simply need to define “nationality” the way it is used in these disciplines. It has nothing to do with citizenship. - PKM (talk) 20:26, 31 January 2018 (UTC)
I am not opposed to a new property; if we have sources that clearly use this term in a way distinct from citizenship or ethnicity, then a new property does seem appropriate. ArthurPSmith (talk) 16:48, 26 January 2018 (UTC)
  • @Jheald: I couldn't quite figure why the data you mention assigns the values. I'm a bit reluctant to use a tertiary source that doesn't provide any references as a basis for this.
    --- Jura 13:07, 30 January 2018 (UTC)

Q7307505

Am I doing this correctly, I have Reformed Dutch Church of Blawenburg (Q7307505) "part of" Reformed Church in America church in New Jersey (Q47520018)? --RAN (talk) 18:28, 30 January 2018 (UTC)

I would use religion or worldview (P140)Reformed Church in America (Q2136834), this seems way too specific. Sjoerd de Bruin (talk) 19:24, 30 January 2018 (UTC)
@Sjoerddebruin: what means « religion » ? is Reformed Church in America (Q2136834) a set of beliefs Reformed Church in America (Q2136834) is comited to ? I mean, a member of Reformed Church in America (Q2136834) is supposed to believe in the texts written by Reformed Church in America (Q2136834) ? Or is Reformed Church in America (Q2136834) an organisation that is a member of Reformed Church in America (Q2136834), a bigger organisation ? This is definitely not a question of specificity :) author  TomT0m / talk page 19:51, 30 January 2018 (UTC)
Is there a better property to link this then? Sjoerd de Bruin (talk) 19:53, 30 January 2018 (UTC)
@Sjoerddebruin: Without any answer to my question, hard to know how to best represent the relationship between those two entities. « religion Search » seems like a catch all property used like this, but what is it supposed to mean ?
But actually, looking at it closely, it seems that Reformed Church in America church in New Jersey (Q47520018) is created by our good friend richard innocently as a class of church organisation belonging to a territory, so it’s not a whole/part relationship but a class membership one. I don’t think « religion » is any useful here. author  TomT0m / talk page 20:50, 30 January 2018 (UTC)
@Richard Arthur Norton (1958- ): That fits if the smaller church is an organisation that is a part of the bigger one. If there is a formal membership link maybe the organisation property should be used. author  TomT0m / talk page 19:51, 30 January 2018 (UTC)
@Richard Arthur Norton (1958- ): But actually no, it seems that you created this item as a subclass of churches, as you built it. Reformed Dutch Church of Blawenburg (Q7307505) is then just an example of such churches. The relevant relationship is then instance of (P31), and we get . part of (P361)_links stuffs of the same kind, physical objects to physical objects for example (my arm is part of my body) or kinds of physical object to kind of physical objects (« human arm » to « human body » for example), but never « my arm » to « human body » because « my arm » is a physical object and « human body » is a class of physical object. In the terminology of the type token distinction, tokens are parts of token, tokens are instances of types, maybe parts are parts of type (meaning if type A is a part of type B, an instance (token) of A is a part of a token of B, or A occurs in B maybe). But tokens are never parts of types. author  TomT0m / talk page 20:50, 30 January 2018 (UTC)
OK, let me see if that works. --RAN (talk) 22:25, 31 January 2018 (UTC)

What tool is available to automate ...

Taking the people listed in w:Category:Mayors of Schenectady, New York and populating their Wikidata entry with "position=Mayor of Schenectady, New York". I do not want to do it by hand.

Thanks! I will give it a try later tonight. --RAN (talk) 23:15, 31 January 2018 (UTC)
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat/Archive/2018/01&oldid=2119552825"