Wikidata:Project chat/Archive/2019/07

From Wikidata
Jump to navigation Jump to search

request merger

Sorry to ask, but I need to do this infrequently: could someone merge Aurebesh (Q56356221) into Aurebesh (Q13221551)? -- AnonMoos (talk) 17:41, 3 July 2019 (UTC)

Done. Scs (talk) 17:59, 3 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 07:46, 4 July 2019 (UTC)

Items merge

Please merge Q3766879 and Q3107110. Thanks! -- 21:22, 3 July 2019 (UTC)

I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 07:47, 4 July 2019 (UTC)

Please add ko:기독교 정교회 to Orthodox Christianity (Q3333484)

Please add ko:기독교 정교회 to Orthodox Christianity (Q3333484) – The preceding unsigned comment was added by 2001:2d8:e52d:9af6::ba08:1f00 (talk • contribs) at 15:55, 5 July 2019 (UTC).

Done by UnifyKorea. --Ksc~ruwiki (talk) 20:18, 5 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --Ksc~ruwiki (talk) 20:18, 5 July 2019 (UTC)

duplicates from BD Gest' import

It looks as though a recent MnM import for BD Gest' author ID (P5491), circa 28 May, may have created a number of duplicate items.

The database has separate IDs for pseudonyms, eg [1], identified as such in the fiche and linked to other entries for each author. New items seem to have been created via MnM for pseudonyms that MnM could not name-match to Wikidata authors.

I've found quite a few of these on the report page for people with matching dates of birth and death; but this will only catch authors who are (i) dead, and (ii) have full (day-specific) dates of birth and death.

Duplicates may also have been created where the name in the database was not an exact match for name(s) we have here; but these may be harder to catch.

I have a bit too much to do at the moment to go in in depth and clean this up, so if somebody could step in and sort it out I would be grateful. Jheald (talk) 13:34, 23 June 2019 (UTC)

Came across a somewhat similar problem with The Encyclopedia of Science Fiction ID (P5357) as the encyclopedia too has separate entries for pseudonyms and pen names (though they are often little more than "pen name used by author X"). Which led me to wonder in which cases it might make sense to create separate items for such pseudonyms and pen names. I guess for pen names used by just one person it's not necessarily neeeded, but we'd need to have a way to model which pseudonym the ID refers to. But there are also pen names shared by several authors or owned by a publisher, where I think it might be better to create separate items. --Kam Solusar (talk) 15:30, 23 June 2019 (UTC)
A applies to name of item (P5168) qualifier can be added to the identifier statement. Ghouston (talk) 02:58, 24 June 2019 (UTC)
@Ghouston: Thanks. I'd been adding named as (P1810) to the multiples I was finding. I can see that P5168 is useful as a qualifier for eg named after (P138), but I somewhat prefer P1810 for identifiers because for many identifiers people add it anyway for all items. Jheald (talk) 09:29, 24 June 2019 (UTC)
@Kam Solusar: Agreed that for eg house pseudonyms shared by multiple authors, or eg Nicolas Bourbaki (Q190529), in such circumstances it may well make sense to have a particular item. But not sure how one would link eg both the house pseudonym item and the actual author item in a author (P50) statement. The usual stated as (P1932) used to indicate the pseudonym that an author used for a book lets one specify a string, but not an item. Jheald (talk) 09:38, 24 June 2019 (UTC)
@Jheald: We also have the same problem with ghostwriters. There are a few discussions involving the topic of ghostwriters in the past (1, 2, 3) but there doesn't seem to be a good way to model that situation yet. --Kam Solusar (talk) 02:52, 28 June 2019 (UTC)
@Kam Solusar: Presumably, use author (P50) = <item>, with qualifiers object has role (P3831) = ghostwriter (Q623386) and sourcing circumstances (P1480) = "uncredited", listing the authors that are credited on the title page in the order that they appear first, before any additional uncredited (but sourced) authors. Jheald (talk) 09:31, 28 June 2019 (UTC)
@Jheald: That would work in most cases. But I think we might run into similar problems as with the pen names, as we don't have a good way to directly link the ghostwriter and the credited person. In the most common case where we have one ghostwriter and one credited person, we can infer the connection. But there are probably other cases involving multiple persons where it might not be that easy. Seeing as people creating something in someone else's name also exist in other arts like movies (Alan Smithee (Q734916)), music, etc., maybe we need a new property to properly model these kinds of relationships. --Kam Solusar (talk) 14:01, 1 July 2019 (UTC)

Wikimedia Space

Forwarding here since no one else has done it yet: Wikimedia Space is a new platform for movement news and conversations. See the announcement for more information. --Lucas Werkmeister (talk) 23:56, 28 June 2019 (UTC)

Personally, one way I can see this being useful for Wikidata is as a place for discussion or announcement of changes to the Wikidata data model that are especially relevant to other wikis. For example, I imagine that the recent-ish narrowing of Property:P143’s scope from “imported from” to “imported from Wikimedia project” could have simplified a lot of other wikis’ Lua modules (to determine whether a reference is a Wikimedia reference, you no longer have to look at the value of the P143, its presence is enough), but I’m not aware of a good place where this change could have been communicated to them (flooding all the Project:Village pump (Q16503) or Wikipedia:Village pump (technical) (Q4582194) pages via MassMessage might have been excessive). That’s not to say there wasn’t a solution for this before, but perhaps the Wikimedia Space can be part of an improved version? --Lucas Werkmeister (talk) 23:56, 28 June 2019 (UTC)
  • So, will there WMDE staff moderating it too? --- Jura 10:32, 1 July 2019 (UTC)

How to handle a page move which led to a new article

list of video games based on Marvel Comics (Q13644820) was originally a list article, but someone on English Wikipedia moved the old article to create a new one. Three other wikis followed suit; two did not.

Can someone familiar with the process explain how we can split this?

Thanks. Magog the Ogre (talk) 08:47, 30 June 2019 (UTC)

Presuming you're satisfied that the three 'articles' are now Articles rather than lists with a more detailed header, then the solution is to create a new item for Marvel Games, delete Marvel Games sitelinks from the list item and add them to the new article item. The subtelty is that those language wiki pages wikidata classifies as a list do not necessarily have to have a title starting 'List of ...'. --Tagishsimon (talk) 09:20, 30 June 2019 (UTC)
@Magog the Ogre: You didn't say what your problem is. I find the articles close enough if there is no other that better represent the game publisher. Leaving as is should be fine? It is however a problem that the WD item start to get conflicting properties attached to it.
In my personal opinion every "list of ..." should not be notable for wikidata. Is this "list itself" notable other than as a Wp article? IMHO turning the item into the actual publisher and make each game a separate item makes sense. Extracting the list from WD when needed should be better than maintaining an actual list. --Jagulin (talk) 19:46, 30 June 2019 (UTC)
Jagulin, possibly you overlook that sitelinks on wikidata items are the mechanism by which interwiki links are provided to wiki articles. For this reason, there is not really any question as to whether any list or article on a wiki is 'notable for wikidata'. They all are, so that interwiki links can be provided. --Tagishsimon (talk) 20:03, 30 June 2019 (UTC)
OK, I created a new item. Why is the interfacing not allowing me to enter the title in other languages? Magog the Ogre (talk) 21:39, 30 June 2019 (UTC)
You probably need to do something with Babel - --Tagishsimon (talk) 21:49, 30 June 2019 (UTC)
@Magog the Ogre: Adding new item Marvel Games (Q64918019) here for reference. You may want to have different from (P1889) statement to make sure they are not merged easily. There is Help:Split an item that may have been what you asked for, as a way to clone an object and move labels. Babel is good when you have several languages you want in the overview. If you occasionally want another language, there is a link below the label box to get all, saying "All entered languages" in English. Unfortunately it doesn't show language that is completely empty. Jagulin (talk) 11:53, 1 July 2019 (UTC)
@Tagishsimon: I did not overlook interwiki. When assigning the enWp "company with list of products" article into a separate item from "list of products, only briefly hinting the company" you will not get interwiki links between them. If there is no other article for the company, I see no error in leaving the list as the interwiki link. Seeing the complaints coming in I believe there is soon a request to merge the items just for this reason. If there is no "real world" notability for the list itself, each Wp should better adapt to the more notable concept. I know anything that have a Wp link is notable for wikidata according to #1, but when solving conflicts I think we should promote going for #2 instead. No consensus perhaps, but something to consider. Jagulin (talk) 11:53, 1 July 2019 (UTC)

Interwiki conflicts

Where is the Interwiki conflicts workflow documented? How many are usually working with the backlog? Is there an inter-language community group to explain language differences and suggest how to make those differences known on other Wikipedia language variants?

Wikidata:Notability mentions two main goals, but I feel WD documentation may need to shift to not get stuck on the inter-language-goal. There are three main criteria, the first gets most attention but the others show that the Wikipedia article structure isn't the rule for WD. Wikidata:Interwiki_conflicts gives three example types of conflict. As far as I see, only the first type (duplicate items) is a possible WD problem. The other problem would be when WD isn't granular enough to have each needed item. Most complaints to "interwiki conflicts" doesn't seem to say anything about which wikis that are in conflict, just pointing out two WD items that I suppose they think should be merged. Most cases (still in the backlog, I guess the easy merges are filtered quickly) to me seems to be a matter of how the local Wp should relate to WD structure, rather than WD to adapt. Did I misunderstand the goals?

An important part of validity of data and the notability criteria are source references. I see very little discussion and requesting of sources when one decides to link an article to an item. When merging, nothing other than personal decision seems to be required, no consensus or referenced sources? Jagulin (talk) 12:10, 30 June 2019 (UTC)

@Liuxinyu970226: Can you assist with some of these questions? Thanks! Jagulin (talk) 11:57, 1 July 2019 (UTC)

Inverse constraint question P1029

P1029 (Crew member) seems to be used to designate participants in spaceflight missions. It appears to have an inverse constraint of P5096 (member of the crew of).

However standard practice on data on astronaut/cosmonaut pages seems to be to use P450 (astronaut mission) to list their flights. Perhaps to avoid the constraint warnings or need for data duplication, this should be looked at. CanadianCodhead (talk) 13:08, 1 July 2019 (UTC)

We need a convention how a Wikidata external property context is explained for an end user

We are testing displaying Wikidata Properties of type External in a template on sv:Wikipedia

see test page Wikidatalight

To make it more user friendly we try to find the Wikipedia page that explains the "external property context"

Example of Wikidata Properties of type External

Salgo60 (talk) 11:10, 29 June 2019 (UTC)

For external identifiers, maybe it's COALESCE(?P1629value, ?P2378value) ..
Obviously, if you prefer items that have a Wikipedia article in a specific language, you might need to come up with a longer chain. By default, you could also link the property talk page, which does provide some information. --- Jura 17:26, 29 June 2019 (UTC)
I see in that list that you use an always the same icon. When available it might be nice to use the icon that's specific in the Wikidata property/subject item of this property (P1629)/issued by (P2378). That might encourage people in general to add more icons. ChristianKl❫ 09:56, 2 July 2019 (UTC)

weather properties

Hi, which property may I use to expres weather concepts like temperature (Q11466) or water balance (Q1148989) related with places, like natural àreas, deserts or just cities?. The temperature (P2076) is mandatory as qualifiers in chemical or biological reactions. The recently created maximum temperature record (P6591) is, by definition, only vàlid as a "maximus". IMHO, a lost opportunity to have a more flexible property, just naming it "temperature" or "remarkable temerature" + criterion used (P1013) with maximum (Q10578722) / average (Q202785) /minimum (Q10585806) etc. Any suggestion will be wellcomed. Thanks, Amadalvarez (talk) 07:17, 30 June 2019 (UTC)

I don't see why temperature (P2076) should not be useful for places or different from a chemical temperature. It does however say "Use only as a qualifier to indicate at what temperature something took place" and I don't know what the subject item of this property (P1629) means. helium (Q560) does use temp, in a clear context. Atlantic Ocean (Q97) has two temperatures, but as you see they don't really make any sense since there are no further clarifications. When, where and how? As subject, temperature measurement (Q909741) sounds promising, but I don't see it much used and there is a difference between measured and calculated temperatures.
maximum temperature record (P6591) was discussed first, though I agree that "extreme" or "remarkable temperature" would be a more general idea. I would suggest that it at least had "when where how" as a recommended qualifier. The ability to set constraints is a reason to not use temperature (P2076) directly, I suppose. Jagulin (talk) 17:17, 30 June 2019 (UTC)
Thanks, @Jagulin: I understand that temperature may be A) the "cause" or the "condition" that anything happens, or B) a result of something to document. For instance, "the forest fire started because the temperature rose to 45°C with a humidity of only 20%. The intensity of the fire produced temperatures up to 600ºC that destroyed several farms". It is, cause/condition and resultant temp. So, I agree with two different properties. The temperature (P2076) is to expres the necessary conditions or cause to get an event, and is correct as a qualifier. The other (P6591 ?) should be able to record what temperature got (or we use to a place); it must be use as a property to be able to have qualifiers ( "when where how"). The P6591 discussion was near "irregular". Only two days and 1 support to get approved. When I created properties, was not so easy.....
What do you propose ?. Try to modify P6591 ?. Try to create a "B) temperature" and let P6591 as a "property for trivial game" ?. Thanks, again. Amadalvarez (talk) 05:06, 1 July 2019 (UTC)
@Amadalvarez: Let me clarify that I'm not experienced with WD structure decisions, I was just responding from a general perspective. Sorry for not finding an obvious suggestion for you.
Causality is sometimes difficult to judge. As far as I see, it's not built into temperature (P2076) at all (see instead has cause (P828)). It however says that it should be "qualifier" only, an added information to another statement. If you have a temperature for the water balance (Q1148989), you should use the qualifier to register that. If the temperature itself is "the statement" it should not be in the qualifier but set as a "property". Maybe speed (P2052) can be changed to allow also use as property, I don't know why it was restricted.
@Manu1400, ديفيد عادل وهبة خليل 2, Jura1: You were involved in the maximum temperature record (P6591) creation process. I see that it is "subproperty" of temperature (P2076), which may be against the intention of that? Can you share your view of making it more general, using a mandatory criterion used (P1013) to mark the type of extreme? Instead of adding the "minimum ..." counterpart it would allow a more flexible use of the property. Is it against WD data design guidelines, making it more complicated? Did you use it somewhere to show how it's meant to be used? In Ohio (Q1397) it has no specific location and no clarification how it was measured. Jagulin (talk) 15:17, 1 July 2019 (UTC)
Thanks @Jagulin: to invite P6591 participants. Look my answer to Jura; probably talking about the specific case make easier to understand it and find a reasonable solution. Amadalvarez (talk) 16:56, 1 July 2019 (UTC)
  • If you want to add something "remarkable" about the weather and don't quite know what, there is weather history (P4150). --- Jura 11:53, 1 July 2019 (UTC)
That is for measurement series, I believe. In general I think WD idea is not to merely reference external data, but rather annotate it in graph structure. Otherwise creating a table for every location is a good idea if there is no external database to reference directly. Jagulin (talk) 15:09, 1 July 2019 (UTC)
It's neither external nor necessarily for series. --- Jura 15:20, 1 July 2019 (UTC)
  • Thank, @Jura1:. Tabular-data is not what I'm looking for. I do not have sources of mass data nor plan to upload historical series.
  • In the cawiki "infobox for natural spots", used for parks, lakes, deserts, seas, etc., we have a few "historical" manual parameters related with temp. I try to find the way to get them from WD, and I need a "quantity property" that can be use -with qualifiers- to expres "when, where and how", as Jagulin well described. Nowadays are: year average temperature, atmospheric temperature range, water temperature (for hot spring (Q177380)). I just want to eliminate the manual information moving it to WD and help editors to enter information in WD instead via manual parameters.
  • If we had a flexible formula, we probable will used for any other temperature data. I red Property talk:P6591 and I believe it may become a more useful property if we manage it as many other dimension properties, as:
Do you think we can reorient the P6591 and harmonize its ontology with this others?. Thanks, Amadalvarez (talk) 16:56, 1 July 2019 (UTC)
  • In general, a property should be usable without reading all qualifiers. I find the use of P2044 somewhat problematic.
P6591 is meant to store things like average and min/max temperature for every month in a year for a given location. What it actually stores is defined on Commons. The idea is that it would be too much data for Wikibase to handle directly in Wikidata. You don't need to store as much at once as it was done for NY. --- Jura 17:03, 1 July 2019 (UTC)
Sorry, yes, it should read weather history (P4150) not P6591. --- Jura 09:32, 2 July 2019 (UTC)
@Jura1: I think the numbers got mixed up in your reply. Did you mean criterion used (P1013) as problematic and weather history (P4150) for the temperature series?
"a property should be usable without reading all qualifiers" may be a good guiding idea. Can you point to any documentation of it for further understanding of reason and exceptions? The "qualifier" system is certainly adding value but you say that it's only meant for "useless" details?
Seeing that criterion used (P1013) is widely used, could you give some examples of what makes it problematic? The problem I can see is that it is difficult to document, thus less obvious for anyone to apply this "property". On the other hand I think having lots of properties with very specific is also making it difficult to use them properly. And weather history (P4150) is also not very obvious how to use the first time.
Even having criterion used (P1013) with "maximum" as in Q80294#P2107, seems to me like a fair use. The property is specific and the qualifier clarifies it. Is it problematic?
@Amadalvarez: Would it be possible for you to try weather history (P4150) in an example and learn more of the benefits and problems of it? Maybe it works well, otherwise it helps you to find arguments for a set of new, specific properties as is said to be the norm. Also see discussions (perhaps to find other users interested in the topic).
Thank you all for an interesting chat. I will now leave it all to you. Jagulin (talk) 07:05, 2 July 2019 (UTC)
@Jagulin:I did not propose the use of P4150. In fact, There is only 3 items with this property. So, it doesn't seem too much popular. In the discussion you linked (thanks), they talked about have a full series of climate and, finally, they created P4150. This references to external tracking events, probably may be solved with data stored on Commons. I don't know, but it was their decission and, today only 3 properties have been created that add 16 items among them.
Apart from that, I just want discrete values of something so usual as temperature in a place. Likely we have population (P1082) without having all the census month by month segmented by sex, profession, origin, incomes level, etc. etc. as it appears in official agencies. Thanks for your contributions. Amadalvarez (talk) 07:56, 2 July 2019 (UTC)
@Jura1: I find it interesting the doubts expressed by Jagulin; surely your answer would help us a lot. Additionally I affirm that the use of qualifiers to determine "when, where and how" for each one of the values, are not problematic. Qualifiers such as determination method (P459), criterion used (P1013) or applies to part (P518) allow us to handle the multi-values ​​of the properties. Obviously, the name of the property should not be too generic, except for the wildcard properties like P2670 or P793 that have saved us from creating hundreds of ultra-specific properties.
Speaking specifically on the P6591: his name is, right now, excessively specific. It is legitimate?, yes, but it would require creating as many properties as variants that someone could imagine (maximum, minimum, average, season of the year, etc.). Many of these combinations, or are not very useful or have few real values, therefore, I think they should not be admitted. I guess you'll agree on this point. Your last answer declares "P6591 is meant to store things like average and min / max temperature". I am with you in the concept and the scope. I wish to interpret that we will modify its name and its definition. Thanks. If you don't mind, I can incorporate the key points of this conversation in the Talk: P6591 to describe the change and ask for opinions in the new scope. I look forward to your reply, Amadalvarez (talk) 09:09, 2 July 2019 (UTC)
I mixed the property ids. It should read weather history (P4150). A more general property than maximum temperature record (P6591) was discussed in Wikidata:Property_proposal/average_yearly_temperature and thought to be problematic. So no, re-purposing maximum temperature record (P6591) to get there isn't a good idea. Do you see any problems with the data as entered? --- Jura 09:32, 2 July 2019 (UTC)
  • BTW, reminds me of phab:T147049. Maybe an update by the devs could help you. I think the general idea is that tabular data should be accessible on query server. --- Jura 09:32, 2 July 2019 (UTC)
Thanks, @Jura1: if you say that P6591 is questionable, I forgot my suggestions. I build infoboxes, so I think in terms of recovery full information about something, doesn't matter if it is in on item, in one monovalue property or in a multivalue property with several qualifiers to understand what each value means and represents. To me, tabular data is difficult to recover, because I get info via cawiki LUA module:wikidata and I do a few SPARQL queries. I think (may be I'm wrong), tabular data are no directly recoverable as the data value of properties. In addition, I don't need to work with series, by now. If, finally, do not delete the P6591, please consider to carry on with this discussion, thanks.Amadalvarez (talk) 16:57, 2 July 2019 (UTC)

Editing and discussions

I've been contributing to WD "by hand" for a few days now. Maybe more advanced users have a tool-assisted set up, but I'd like to share some impressions from a "new user" perspective.

  • There is a "quick insert tool box" below the editor in edit mode. It's mostly made up of symbols. I've been missing {{Ping|}} or {{re|}} for discussions. Even more important should be {{Q|}} and {{P|}} - or some kind of subst to handle either case. How to go about suggesting additions to the set? Or is there a plugin to be used?
  • Discussion pages for items rarely draw enough attention to have discussions started. Seems to be the case for help pages as well. Is the rule to have all discussion in the chat or are there certain things to be discussed on the item talk page? Maybe the talk page should be dedicated to tool output and all discussion redirected to chat? Having a search pull out all mention of the item so one knows if there has been consensus made already?

Editing WD is one thing, but learning the procedures and seeking assistance or consensus involves discussions. Making this more efficient can help improve WD overall. Jagulin (talk) 16:35, 1 July 2019 (UTC)

The edit tools below the editor are set at Template:Edit tools, but they require an admin to mark the version for translation before it goes live. --Yair rand (talk) 21:28, 1 July 2019 (UTC)
✓ Done --Matěj Suchánek (talk) 07:25, 2 July 2019 (UTC)
Thanks! I was just going to ask if it was possible to test the change in a sandbox, but I've now tested it live. Works great! There is a space before </charinsert> that I don't think is relevant, but it doesn't seem to have any effect so no problem.
Was there opposition for including Ping, or was I just not fast enough? Jagulin (talk) 07:41, 2 July 2019 (UTC)

River became a mathematical object last night

By the magic of some mysterious edits, it seems that river (Q4022) became a subclass of « history » last night, and is currently a subclass of mathematical objects. I just hope this was good faith edits. author  TomT0m / talk page 19:41, 1 July 2019 (UTC)

This morning it's mysteriously gone. --Matěj Suchánek (talk) 07:26, 2 July 2019 (UTC)
You awoke my interest... I see no trace of this in the item history and you talk about mysteries. Was it vandalism that have been hidden in history or would you think it was some kind of bug in the underlying database? Or write it off as a misunderstanding in the first place, if nobody else noticed? Jagulin (talk) 07:48, 2 July 2019 (UTC)
Probably refers to the addition of path (Q12799272) to watercourse (Q355304), where the former is a mathematical object. Ghouston (talk) 09:47, 2 July 2019 (UTC)

Wikidata Graph Builder - where can I find a description of the parameters?

I sometimes use Wikidata Graph Builder tool, but I have not been able to find any description of the parameters in general and Iterations and Limit in particular. I understand that they can be used to limit the result, but what are the exact function of these two parameters? Can anyone please enlighten me? --Larske (talk) 10:44, 2 July 2019 (UTC)

Bad constraint error

On Mecosaspis mapanjae (Q14823309) there is a bogus constraint error, because the image file name includes the string "map". Is this a worthwhile trade-off, or should the constraint be removed? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:48, 2 July 2019 (UTC)

The regex includes a word boundary before "map". If it also included a word boundary after "map", there would be fewer false positives. But it is somewhat strange to constraint based on a file name which can be in any language. --Dipsacus fullonum (talk) 14:58, 2 July 2019 (UTC)
It's a very hopeful constraint; I've seen a complaint that it gets annoyed with any .*poster.* which is not affixed to a film item, because theatre posters don't exist. --Tagishsimon (talk) 15:54, 2 July 2019 (UTC)

Politico y aristócrata chileno José Nicolás de la Cerda <=> político y aristócrata chileno José Nicolás de la Cerda de Santiago Concha

José Nicolás de la Cerda de Santiago Concha and José Nicolás de la Cerda are looking awfully like duplicates to me, but a merge is blocked by two different Spanish wiki articles. Also, the birth dates are (slightly) off. If anyone from Spanish wiki could have a look that would be great. Moebeus (talk) 12:58, 5 July 2019 (UTC)

Posting a note at Wikidata:Café is a good way to find a Spanish speaker to help out. —Scs (talk) 00:21, 7 July 2019 (UTC)
I have merged the two articles, waiting for a sysop to proceed with history merge. As of the birth dates, I couldn't find any source to indicate which one is the correct. Esteban16 (talk) 01:04, 7 July 2019 (UTC)
Done now. Esteban16 (talk) 03:39, 7 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 08:29, 8 July 2019 (UTC)

English As She Is Spoke

I was editing the entry for Battle of Waterloo (Q48314) and was surprised to see an "English" and "British English" entry. I have trawled the archive to this page and came up with these discussions:

The ignorance of some of the contributors to the previous conversations are surprising, how for example can someone not be aware that there is a dialect of English called w:Indian English? I find the discussions surprising and some of the comments bigoted, particularly as these were issues addressed more than 10 years before the first archive entry above and was settled inclusively and harmoniously.

Either the two outliers English-GB and English-CA need to be scrapped, or all of the different types of English need to be included. This includes the English used in the USA at the UN and within the EU. (The English used by the UN and EU are different but based on variants of British English).

As has been pointed out in the previous conversations English on Wikipedia is dialect neutral until a dialect word is introduced at which point that dialect spelling or word becomes the default version of the language used for that subject.

It is not as if the data within articles often provides this information:

Or in the case of some where the version of English is obvious to native speakers, other templates that can be indicative for machine reading:

There is a list of the use English dialect templates at w:template:Use X English. If there seem to be too many consider using w:Use Commonwealth English and incorporate into that all the other dialects apart from American and Canadian English.

This is supposed to be a database project, adding or removing attributes ought to be easy, why has this issue not been addressed?

-- PBS (talk) 12:33, 19 June 2019 (UTC)

We probably should add more types of English labels; there are 12 variants of Chinese available here. Though written (and spoken) Chinese probably varies considerably more than English does. To see the list of available languages for labels, try creating a new item and look at the language dropdown. The process of adding languages to the list however is a bit complex right now. Items do exist for American English (Q7976) and other varieties though. ArthurPSmith (talk) 19:22, 19 June 2019 (UTC)
There was a very sparsely attended RFC on this way back in 2013 - Wikidata:Requests for comment/Labels and descriptions in language variants. My personal feeling is that we should just get rid of English variants. It can very occasionally defuse an argument in some items, but in practice it's just more maintenance overhead 99.99% of the time. Andrew Gray (talk) 20:18, 19 June 2019 (UTC)
Thanks for writing this! I agree that more variants should be supported. I'll be happy to support requests for that. --Marsupium (talk) 17:44, 21 June 2019 (UTC)

I guess this has been said before, but: I'm glad this issue hasn't been "definitively resolved", that it doesn't receive much debate, that an RFC on it was "sparsely attended". Truly, labels (and descriptions) on Wikidata are secondary, just for convenience. (Because obviously doing everything based on raw Q numbers would be a nightmare.) But we don't have to get wrapped around the axle trying to find the "right" label or description, or complaining that some entity has the "wrong" label -- and this is as it should be. (The Wikipedias get wrapped around this particular axle all the time, in part because of their visibility, in part becaue of the requirement that each article have exactly one title. I'm really glad Wikidata doesn't have that problem.) --scs 11:05, 26 June 2019 (UTC) [My point is that distinguishing between British vs. Canadian vs. American vs. Australian vs. Indian English -- important though the issue can be in some contexts -- is just really small potatoes here.]

The IETF shares your attitude, and uses a "the shorter spelling wins" rule—as a joke, no centre in RFCs, centers will do.:-) Of course copying an enwiki SHORTDESC, even it uses en-GB spelling, should be okay. – 12:31, 3 July 2019 (UTC)

Featured pictures

On the subject of images, shouldn't we allow "featured picture" or "Valued image" as a reference type? Adam Cuerden (talk) 10:49, 24 June 2019 (UTC)

@Adam Cuerden: so after a bit of investigation, I think the best option here seems to be a object has role (P3831) qualifier - it's currently used in various places, though about 500 image statements have one. It's a bit cryptically named, though....
So we'd add this qualifier to the image statement, and give it a value similar to Featured article (Q28801937) (there is currently no item describing the concept of a "featured picture", but it would be straightforward to create one). Add a reference (to the FP discussion?) and bang, we're done.
A couple of caveats, though - firstly, this wouldn't be able to differentiate between Commons FPs or Wikipedia FPs (or which Wikipedia, for that matter) since we can't put qualifiers on qualifiers. If you wanted to do that, you'd probably need items for "Dutch WP featured image" and "Commons featured image" and so on.
Secondly, you'd have to add these qualifiers to each appearance of the image - although that would probably be an easy job for a maintenance bot.
In general, I don't see any reason not to do this - it's useful contextual information, and I can easily imagine that someone would be interested in being able to say "if there are several images, use the FP" or similar. Ultimately, it might be that the Commons linked data project (which is oriented towards information about that exact file) would be able to handle it better. But perhaps the two approaches could complement. Andrew Gray (talk) 21:09, 28 June 2019 (UTC)
On commons FP can be certified nonsense, just tweak the "scope" until only one file fits. – 15:38, 3 July 2019 (UTC)

Wikidata Bridge: edit Wikidata’s data from Wikipedia infoboxes

Hello all,

I’m happy to announce that the Wikidata Bridge project (you may have heard about before under the name “client editing”) started. The goal of this project is to offer a way to Wikipedia editors to edit Wikidata’s data more easily. This will be achieved by an interface, connected to the infoboxes, that users can access directly from their local wiki.

The project is now at an early stage of development. A lot of user research has been done, and will continue to be done through the different phases of the project. The next steps of development will be achieved by the development team working at Wikimedia Deutschland, starting now until the end of 2019.

Here’s the planned timeline:

  • From June to August, we will build the setup and technical groundwork.
  • From September to November 2019, we will develop the first version of the feature and publish a test system so you can try it and give feedback.
  • Later on, we will test the feature on a few projects, in collaboration with the communities.
    • We will first focus on early adopters communities who already implemented a shortcut from their infoboxes to edit Wikidata (for example Russian, Catalan, Basque Wikipedias)
    • but we also welcome also communities who volunteer to be part of the first test round.
    • Then we will reach some of the big Wikipedias (French, German, English) in order to see if the project scales and to address their potentially different needs.
    • Even later, we can consider enabling the feature on all the other projects.

On Wikidata, from the technical side, not much will change. When the feature to edit Wikidata’s data from Wikipedia will be implemented, a tag for edits coming from Wikidata Bridge will be added, so the Wikidata editors can watch and filter these edits more easily.

If you want to get involved, there are several ways to help:

If you have any questions, feel free to ask them here or on the on the dedicated talk page.

Cheers, Lea Lacroix (WMDE) (talk) 13:05, 24 June 2019 (UTC)

Hi Lea. Some thoughts and questions. First up, this is called a bridge, but it seems more like a shield -- trying to shield Wikipedia editors from the fact that it is Wikidata they are editing by not showing the Wikidata item, but instead a pop-up dialogue.
According to the pitch (mw:Wikidata_Bridge#The need), Wikipedia communities currently find issues over "lack of control over the data, not enough information about when and where the information is updated, differences in the way Wikipedias and Wikidata structure information".
It's not clear how the shield will help this. If information is not stored in the wikitext of wikipedia pages, bad edits on Wikidata (which will continue to be possible whether or not the shield is present) will continue to show as vandalism in infoboxes, without there having been any change to the wikitext, or any change in the page-history or in watchlists (if Wikidata notifications have been turned off). The shield if anything seems less apparent where information is held and updated, since its intention is to hide wikidata pages from casual editors. And there will continue to be differences in the way Wikipedias and Wikidata structure information, since for item-valued properties only a drop-down/autocomplete list of items will be available, rather than the free-text Wikipedia editors expect. There also remains the issue that most vandal-fighting and batch-editing power tools on Wikipedia are based on wikitext, which wikidata-based fields in infoboxes and templates will continue to sit apart from.
That's not to say that the shielded edit-interface may not be a good thing in its own terms for casual editors -- more accessible, more immediate, less scary than being thrown into a wikidata page; so the initiative I think is quite welcome. But when communities say I want the information that appears on Wikipedia to be editable on Wikipedia, it's not just the accessibility issue by/for casual editors that that sentiment is code for.
With luck more causual accessibility will mean easier good edits. But the point made by Vojtěch Dostál on the mw talk page is significant: it will also mean easier bad edits and easier vandalism, including quite subtle vandalism that may be hard to pick up (eg minor falsifications of physical constants, and other statement values that may be deliberately wrong, but not wildly implausibly so). His idea of requiring editors to be autoconfirmed before they can make shield edits is a good one.
But a further issue is that, for high-visibility high-vandal-risk edits like this, referencing and verifiability is particularly important. But there seems to be no provision for this, in the apparent development roadmap. Perhaps understandably. Creating new references on Wikidata can be quite tricky, particularly if new items for new sources would need to be created. An alternative might be to create a wikitext reference on the local wiki, and Visual Editor does have some quite developed tooling for that; but then (i) some option would be needed to allow or combine with a reference that already exists on the page; (ii) the reference would not be available on other wikis where the value might suddenly be appearing via Wikidata; (iii) there would be a danger of the value subsequently being changed on Wikidata, but the local reference for the old value still being applied to it. So local references become difficult.
There's talk of VE independence but also potential VE integration. This idea of à-la-carte access to VE-component features is quite interesting. I'm not sure if there are gadgets to give access to the VE reference-editing component without VE being chosen for the main editor. But maybe worth remembering that VE uptake on wikis is often low, sometimes very low. (I've seen figures of fr-wiki 23% use of VE, ru-wiki 19%, es-wiki 11%, de-wiki 11%, en-wiki 4%). I don't know how reliable those figures are; but on some wikis there is definite resistance to VE.
Finally, any thoughts about Commons? Is there any thought to make SDC data available editable from templates on description pages in this way? Jheald (talk) 15:56, 24 June 2019 (UTC)
@Jheald: I hope you won't consider this to be diverting your remark, and if you do please just tell me to take it somewhere else, but as I've said before mainly on Commons: I continue to believe that the best solution to this is serialization of Wikidata into (and deserialization of Wikidata from) something with at least approximately the same syntax as a wikitext template. The wikitext user would have basically the same experience as when using a template that offers name-value pairs; the content corresponding to the template's name-value pairs would actually reside in Wikidata. This is not necessarily exclusive of the current proposal, which may be easier for less techincally oriented users, but should be available for those of us who prefer to work with wikitext. - Jmabel (talk) 22:55, 24 June 2019 (UTC)
@Jmabel: I once thought that's what should be investigated, particularly for editing SDC via something close to wikitext edits on existing templates. But I am not sure it's so easy. Serialisation, from wikidata/SDC to wikitext, might be relatively easy. But the reverse is not -- data reconciliation from strings to wikidata items is a real grind, for which I really don't see any shortcuts. Any idea that one could just make arbitrary wikitext edits in the template, and leave it to the software to convert that into wikidata updates is just not realistic. I don't see any way round manual disambiguation (or manual confirmation that really no wikidata item exists). Without such intervention the best one could generally probably automatically get to might be somevalue stated as (P1932) "string". But that's not really good enough. Alternatively, one could accept a wikitext update to a local value, but then tag if it didn't match SDC/Wikidata, in some way that could be automatically retrieved at scale. That might be more promising, but it probably does make sense to try to get casual manual editors to directly choose a wikidata value if possible instead. Jheald (talk) 23:28, 24 June 2019 (UTC)
It probably can't be done universally, but for a lot of properties it should be easy:
  • Image links.
  • Things that actually are strings (e.g. captions; URLs).
  • Things with a reasonably limited domain, e.g. dates, simple numbers.
  • It's only a little tougher to do amounts of money (because a currency qualifier is needed) and physical areas (unit of measure qualifier)
  • We could have a syntax available so that any item that already has a Wikipedia article of its own wikilink can be referenced via the name of that article, or Wikidata items could, of course, be referenced directly. Especially for the latter (and perhaps for the former) it would be nice to have a search tool available during editing: not trivial, but not prohibitive.
That right there is easily the majority of cases.
This wouldn't be trivial to do, and obviously, there is a bit of calculation involved in the deserialization, and it would doubtless introduce a step of coming back to the user to deal with anything unparseable or ambiguous (much as Commons already has to warn about various upload problems), but I believe it would be tractable. Certainly tractable enough to merit some investment into looking at the potential. - Jmabel (talk) 04:06, 25 June 2019 (UTC)
I think it's only easy for those when you ignore the fact that good statements are sourced. This is in particular true when the big Wikipedia's don't want to read unsourced date values from Wikidata. ChristianKl❫ 10:09, 25 June 2019 (UTC)
Interestingly, infobox values at Wikipedia have rarely a source indicated ;)
Reminds me to repeat: make sure not to overwrite or "correct" sourced Wikidata statements when editing from Wikipedia. --- Jura 13:17, 25 June 2019 (UTC)
Even if it's true, they will still hugely complain if we want to install a Wikidata Bridge on their Wiki that brings unsourced claims with it. ChristianKl❫ 13:48, 25 June 2019 (UTC)
On wvwiki, infoboxes about persons are often unsourced as such. But they are often regarded as secondary to the text. And the text have to have sources. Complaints often comes when the infobox and the text tells different things. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 13:54, 25 June 2019 (UTC)
For most wikis that is probably true. Supposedly some even discuss diverting dates/info. Still in some wiki, user mention the source only in the edit summary ;)
Anyways, if the bridge is set up for editing/adding, it could be less an issue. Besides, many of the larger wikis may feel that they can't do all updates locally. --- Jura 14:03, 25 June 2019 (UTC)
All info in the lede and an infobox has to be covered by RS in the body, otherwise it should (or for a BLP must) be removed. For info from Wikidata "unsourced" is clear (=unusable, remove). The R in RS is undefined, e.g., reference URLs can be anything. – 12:54, 3 July 2019 (UTC)
c:Module:Creator, c:Module:Institution and c:Module:Artwork do something quite similar to this. They compares the templates' fields to Wikidata's data and if it's missing here composes a link to QuickStatements that allows adding them. That functionality is open to extension. Currently, only Commons will be added as a source. --Marsupium (talk) 10:55, 25 June 2019 (UTC)
You have to be aware about one important difference here. On Wikipedia we tend to update data by simply replacing it. Here we add new data and change the ranks. Its doable, but the developers have to know what they are doing. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 13:07, 25 June 2019 (UTC)

Hello all, and thanks for your feedback. Let me answer on a few points:

  • About Commons and other projects: if everything goes right, the system that we will develop can work on projects using similar templates. Because the need emerged primarily from Wikipedia communities, we'll focus on them for the first versions, but at any moment we can consider experimenting with other projects, if the template builders are willing to.
  • As we are aware that references are extremely important for the data quality but also the acceptation of Wikidata's data by Wikipedia communities, we'll make sure that they are present, at least in a simple form, since the first version. Later we can develop more complex checking systems.
  • Just to make it clearer: nothing will be enforced on Wikipedias. We will offer to the template builders a possibility to update the infoboxes to make them have the Wikidata Bridge option. We will encourage the communities to go step by step (eg start with only one Wikiproject) and see how it goes.
  • About replace vs update data: yes that is one of the tricky pieces, and we hope that we can boil it down to a few standard cases that lead to distinct edit flows on Wikidata (add new value, change rank, ...). We hope we can then let the template creator chose which of those standard cases applies to a particular template parameter. The editor would then just have to say if they are fixing a typo or adding something new as shown in the mockups. We'll test it in the next versions and see if it works out.

Cheers, Lea Lacroix (WMDE) (talk) 08:27, 27 June 2019 (UTC)

@Lea Lacroix (WMDE): Something is missing in the description of the new tool (or perhaps I didn't find it): the rules of the process used in data selection.
How the tool selects statement when several values are available ? When a preferred rank is used among other normal ranks, this is simple, but when several preferred rank values are present or several normal rank values without a preferred one, how the tool will select a value to be updated/changed ?
If I understood the principle, the tool will change one value but not necessary the value displayed in the infobox: if the extraction rule of data from WD is managed by other rules than the rank one (for example, statement selection is based on reference type in WP side by using a lua programmation), then wikipedians will spend time trying to update the value but not the one used by the infobox. How do you plan to overcome that situation ?
What happens if someone changes the value of a referenced statement ? Does the reference stay or is deleted ? If a referenced statement is wrong or outdated, then the value shouldn't be changed but a new statement created with rank lowering for the existing statement.
Someone see a data in an infobox which is not the usual one because the data selection of the Wikipedia template was defined in a particular way. Contributor will create new statement which perhaps already exists (duplicate) just because he is not able to see all existing statements in WD.
For me that kind of tools is not a good idea because he is not linked to what the contributor is reading: there is no link between the data which can be modified by the tool and the data displayed by the template. Rule for data selection is not only based on rank, date or reference type.
Unless a better description of what is done by the tool, I think we are going in the wrong direction by developing that kind of tools which are not able to provide all information about what is displayed in the infobox and which statement will be modified in WD. Snipre (talk) 12:25, 28 June 2019 (UTC)
@Lea Lacroix (WMDE): To be a little more constructive, I can only advice you to consider a complex case with several statements for one property, with different values, and several references and several qualifiers for each value:
Example, heat capacity of a chemical with different values, different units, different qualifiers (temperature, state), et different reference data
value unit qualifier 1 (temperature) qualifier 2(state) reference data 1 (stated in) reference data 2 (page)
2.01 kJ/kg °C 25°C liquid CRC Handbook of Chemistry and Physics 3-21
Determination of heat capacity of XXX (scientific article)
2.015 kJ/kg °C 20°C liquid CRC Handbook of Chemistry and Physics 3-21
1200 Btu/lb °C 25°C liquid New determination of heat capacity of XXX using quantum chemistry (scientific article)
1.75 kJ/kg °C -100°C solid The latest values in chemical engineering 150
This table should be displayed by you tool before wikipedian can only try to modify something in WD. Snipre (talk) 13:35, 28 June 2019 (UTC)
  • Let's see how it goes. The default should probably be "add the new value with preferred rank". --- Jura 13:31, 28 June 2019 (UTC)
@Snipre: Thanks for your feedback. We've been thinking about this, and we're still trying to figure out the technical details, and how to display all of this information in a nice and readable way in the interface. Lea Lacroix (WMDE) (talk) 08:15, 2 July 2019 (UTC)

Disambiguations in aliases

Hello. Some people add Wikipedia article names to aliases (diff) to make item searching easier for them. I have tried to find if it is allowed or disallowed in rules (Help:Aliases, Help:Label), but failed to do so. Please help me to understand consensus on this topic and modify rules accordingly. — Vort (talk) 08:29, 28 June 2019 (UTC)

I don't know that there is a consensus, one way or another. The full search will, AFAICS, return the item at the top of the results list if the sitelinked title is searched for - example: - but the search box dropdown does not list the article when the sitelinked title - John Smith (architect) - is input; and it would if the sitelinked title were included as an alias. So a case can be made that it is a useful thing to do. Again, AFAIK, labels, descriptions and aliases are there for the convenience of users, whereas statements and qualifiers are the rigorous structured data; and so my view, fwiw, is that we should be relaxed about users exercising the option to include what might from many perspectives be thought of as superfluous aliases. In the scheme of things, I don't think any harm is done. (Obvs, because this is a wikipedia project, others may hold different views.) --Tagishsimon (talk) 10:33, 28 June 2019 (UTC)
I see that item appears in search. Better example is "Ивановка (Теренкольский район)". But there are no appearance when user adds item as property value (most likely, the same mechanism is used as in search suggestions). So I think better solution is adding of titles as virtual aliases for all usages inside Wikidata engine. — Vort (talk) 14:36, 28 June 2019 (UTC)
Let me clarify the example to save others from doing the investigation to understand the details:
"Ивановка (Теренкольский район)" article is page for Q16654862. Still, the item does not appear when searching for "Ивановка (Теренкольский район)".
Item does appear in dropdown and search or "Івановка (Теренкольський район)" (which is a label in 'uk') or "Ивановка (Kazakstan)" (which I added as "also known as" for 'sv').
Item appear does not appear in dropdown, but probably in search, for "Siedlung in Kasachstan" (a description text for 'de')
Dropdown shows the description text too, and matching "alias". Description text included only from the current view language or from English (e.g. Q1374309 in 'ru'). Left out if both blank. If item label is blank for current language, item title taken from English (e.g. Q1374309 in 'ru').
Searching seems to match any language, regardless of set language. Q-number is an alias automatically.
The advanced search method "pages in this language" doesn't seem to have any use on WD. Trying both 'ru' for Q16654862 and 'de' for "Siedlung in Kasachstan" give zero matches. It thus neither matches items having Wp for a specific language, nor the description in a specific language. --Jagulin (talk) 17:14, 29 June 2019 (UTC)
I don't think that adding Russian Wikipedia article names to Russian aliases is a problem — it breaks nothing. However, I consider it useless to add foreign names to obviously inappropriate labels like Special:Diff/969490993 or Special:Diff/120755985. Сидик из ПТУ (talk) 14:06, 28 June 2019 (UTC)
Duplication is always a problem. If one instance is changed, second should be changed too. And almost always no one cares about it and incorrect duplicate stays in data for years. — Vort (talk) 14:21, 28 June 2019 (UTC)
I agree that duplication causes a maintenance problem. Time spent when pasting the "alias" in addition to the time spent by the community to maintain it needs to be considered if there is a guideline about this. A better solution would be a change in the software, to include each language URL in the search and dropdown list generation. Did anyone create such a suggestion?
The second problem to consider is how to make item selection in dropdowns efficient. "John Smith" gives lots of matches and there is a lead text. Having even more items matching in the list, if adding several alias, is perhaps not optimal. --Jagulin (talk) 17:14, 29 June 2019 (UTC)
Titles can be used as aliases if other data sources generates no (or small amount of) suggestions. — Vort (talk) 19:59, 29 June 2019 (UTC)
@Vort: If you still meant that it should be done automatically by the system, rather than by volunteers spending time on it, I agree. Did you create a tracker for the suggestion, to get developer feedback? Or did you man that it's already working like this? Jagulin (talk) 07:03, 3 July 2019 (UTC)
@Jagulin: I have created phab:T227170. — Vort (talk) 07:25, 3 July 2019 (UTC)

F1 races module


Is possible to create a module, as Cycling race, for F1 race, to not edit the result for all Wikipedias, but include the results and edit once for have the one result for all Wikipedias, and Wikinews.

Thanks, Aabbccddeeffabcdef (talk), the 15:32, 30 June 2019 (UTC).

Post scriptum : I'm French, so my English is not very good (level 2).

@Aabbccddeeffabcdef: Is it possile ? Yes, but for that you need to ask help in the wikipedia you want to implement that module. Wikidata is not responsible of the data use in Wikipedia and as each WP has different rules and different tools for data implementation from WD, there is no unique solution. Snipre (talk) 04:44, 1 July 2019 (UTC)
Avoiding the need to maintain raw data in each language Wp sound like a good idea if you have some users with you. Even if not easily solved, if you have a "Cycling race" module you should try to make contact with anyone in charge of that. They could probably assist in technicalities and clarify how much problem they've seen. Even if each Wp is different, it should be possible and beneficial to make a module generic enough to be used across language and style. "Commons" may be a place to work independently from Wp. Each Wp then picks it up when they see fit. Jagulin (talk) 12:05, 1 July 2019 (UTC)
Even if I agree with the comment of Jagulin, I can only remark a point: a lua module require the use of general template to extract data from WD and to format them (example: references). The main general templte for that is module:Wikidata. But this module is not the same for each Wikipedia. So you have 2 possibilities: use an existing general module and then you are limited by the module used in your Wikipedia or creat your module integrating all basic features for data extraction from WD. The best is to contact the person who wrote the module Cycling race to see how this module is independent of other modules. Snipre (talk) 15:59, 1 July 2019 (UTC)
Many people have contributed to Module:Cycling race. It is working indpendent of other modules so it can be copied to and used in any Wikipedia. To do something similar for some other area, you or the persons writing the module will need to be able program in lua, know how lua is used in MediaWiki (the Scribunto extension), and know how to access data from Wikidata. The module will need to access data from many different Wikidata items to compose its tables. That can use a lot of resources, so it is important to do this as efficient as possible to avoid timeouts or other resource limits. Also design the module to have translations and configuration in separate submodules. Module:Cycling race have some problems caused by some less than good design decisions in these areas. I would be able to give some advice on how to do such things. Keep in mind too that it only works if someone input all the needed data to Wikidata. That can also be a lot of work in it self. --Dipsacus fullonum (talk) 16:48, 1 July 2019 (UTC)
Thanks for sharing, Dipsacus fullonum, it was exactly the helpful guidance I expected to see. I never meant to say it was easy, but if there is a strong community interest it could be done. In the end, it should save effort to not maintain language-independent data on each Wp, and having even a first draft module may be the only way for people to take that step. At least, I hope that's the result seen in Cycling. There is of course a chance that "the normal wikipedian" stops contributing since WD and interwiki is too scary for them. A quick look to the Cycling module discussions it looked generic to other type of seasonal racing, but I understand it's unfortunately more work to be more generic too.
@Snipre: Even if there are historic Wp-specific modules, I don't see any reason to make a new model anything less than generic. Do you know if there has been plans to introduce a new "basic module" (abstracting some f the lower level programming work) that would (or could easily) be available on all mediawiki? It would still be a community decision to accept the specific inclusion, use and configuration of each template, but new modules would have a base to start from. Jagulin (talk) 09:24, 3 July 2019 (UTC)
@Jagulin: To create a module in WP using data from WD you need 3 things: a set of extraction tools, a set of data format tools and the module itself which organizes the data in the wikitext. The main proble is that extraction tools and data formatting tools are different between WP, because each community develops code according to its need. But this influence the way you access the data in your module code so you can't create a generic module because the interfaces between the module and the extraction/formatting tools are different.
There is no plan to generate generic code for all WPs because WPs want to keep the lead on that code which influences their templates. Snipre (talk) 10:05, 3 July 2019 (UTC)

Wikidata weekly summary #371

Quick The DJ List artist ID (P6935) plausibility check on Sasha Grey (Q2709) with the expected result spam.:-( 13:38, 3 July 2019 (UTC)

P6546 requires constraint on data

There should be a constraint on this preventing it from being a negative number, which is not possible. Whether by data entry or import error, it seems possible to be entered as one, such as here Q1852981 CanadianCodhead (talk) 15:57, 2 July 2019 (UTC)

You could try and see if format constraint (Q21502404) and format as a regular expression (P1793) works on penalty minutes in career (P6546). Multichill (talk) 17:29, 2 July 2019 (UTC)
Or you could use range constraint (Q21510860) (documentation). --Lucas Werkmeister (WMDE) (talk) 10:34, 3 July 2019 (UTC)

I used a format constraint (Q21502404) to set a regex pattern career plus-minus rating (P6547) values must conform to. Two things now: 1) can someone double-check my regex? I've used "-?\d+" to catch any positive or negative integer; 2) I'm seeing an error now on pages where this property is used, "Properties with constraint 'format constraint' need to have values of type 'String' or 'Monolingual text.'" Does anyone know how to change the value type to string? The data type for career plus-minus rating (P6547) is currently "Quantity" and I can't figure out how to change it. On a related note, I used a range constraint (Q21510860) to specify that the value for all the other properties I've employed for hockey statistics -- total goals in career (P6509), total assists in career (P6545), total points in career (P6544), penalty minutes in career (P6546), and total shots in career (P6543) -- can only have a minimum of zero, and thus shouldn't get mixed up with +/- values. LesserJerome (talk) 12:27, 3 July 2019 (UTC)

entries for social media user names

It seems like the social media user names for Greta Thunberg and Luisa Neubauer in their wikidata entries are changed again and again - but only the capitalization of some letters is changed. As far as I know the capitalization of letters in user names is irrelevant. Only the aricle history gets longer and longer. Would it be possible to have a filter that does not allow edits that only change the capitalization of social media user names? --C.Suthorn (talk) 17:27, 2 July 2019 (UTC)

Sounds like a textbook job for protection. Circeus (talk) 02:33, 3 July 2019 (UTC)

Creating an item for an author of multiple items

I was gonna create an item for Li Dongrong 李东荣, a Chinese central banker, then I found five papers: special:search/"李东荣". Is there a neat way (a script or program?) to create the item and migrate author name string (P2093), or create the item from the P2093?--Roy17 (talk) 11:34, 3 July 2019 (UTC)

You can try [2] --Stevenliuyi (talk) 14:11, 3 July 2019 (UTC)

Change in the name pattern of new Wikidata RDF dumps

Hello all,

Starting on July 15th, the name of the RDF dumps will be changed to remove the "beta". For example, the file that would have been named wikidata-20190717-all-BETA.ttl.bz2 with the former name format will be named wikidata-20190717-all.ttl.bz2.

This will only impact the new generated dumps, not the previous ones.

If you have questions or issues, feel free to ask on the Phabricator ticket. Cheers, Lea Lacroix (WMDE) (talk) 13:30, 3 July 2019 (UTC)

No JSON dumps for the weeks of June 24th and July 1st

Hello all,

Due to technical issues, the JSON dumps for this week and last week couldn't be properly generated.

  • 20190624 because of a commit that was fixing another bug (phab:T226601)
  • 20190701 due to an issue with qualifier hashes (phab:T227207)

We apologize for this inconvenience. The problem is about to be solved, so we expect the situation to be back to normal next week. However, we will not generate the previous failed dumps.

Thanks for your understanding, Lea Lacroix (WMDE) (talk) 17:37, 3 July 2019 (UTC)

instance of(P31) or subclass of(P279) pseudoscience(Q483677)?

Querying everything that has been labeled instance of(P31) pseudoscience(Q483677) there is 1 result: Bach flower remedies

Querying subclass of(P279) pseudoscience(Q483677) there are more results. Regardless of how many hits I get with any of these, should I do instance of or subclass of for this item? blood type personality theory.

Also having read the Help:Basic_membership_properties I didn't completely get yet the difference between instance of and subclass of an item. Ελλίντερεστ (talk) 20:32, 2 July 2019 (UTC)

Bach flower remedies (Q788110) would be a class. One particular glass of such a solution sitting on your table would be an instance. Whether it's really a subclass of pseudoscience I'm not sure: I think that's saying it's a field of pseudoscience, in the way that organic chemistry (Q11351) is a field of chemistry (Q2329), and has been made a subclass. potassium compound (Q12547983) isn't a subclass of chemistry, or of science in general. Ghouston (talk) 04:55, 3 July 2019 (UTC)
It could be a subclass of homeopathic drug (Q50092815). Ghouston (talk) 06:58, 3 July 2019 (UTC)
In most cases subclass of (P279) would be more appropriate here. When adding claims as something being a pseudoscience I would recommend that you add sources for that classification. ChristianKl❫ 21:01, 3 July 2019 (UTC)

New item: central banker?

Should there be an item of the occupation(?) central banker (akin to investment banker (Q2883465))? There is Category:Central bankers (Q8353786).--Roy17 (talk) 20:26, 3 July 2019 (UTC)

This data is stored using a combination of central bank (P1304), office held by head of the organization (P2388), and position held (P39). --Yair rand (talk) 20:52, 3 July 2019 (UTC)
See Mario Draghi (Q294460) and President of the European Central Bank (Q605440). Xaris333 (talk) 21:02, 3 July 2019 (UTC)
As Yair points out, "Central banker" is not really an occupation (economist (Q188094) would be more typical) and this is usually handled via position held (P39). That property doesn't work too well when defining category contains (P4224) (you can't structure it as you can for, say, Category:School principals and headteachers (Q8709447)). Was that the actual issue you were trying to get at, @Roy17:? I think skipping the "human" level for it can work: category contains (P4224) director (Q1162163) / of (P642) central bank (Q66344). Circeus (talk) 21:11, 3 July 2019 (UTC)
I know they are usually regarded as economists, but IMHO central banker could be a profession, being a special type of government economists as well as a special type of bankers. They are economists that have real power, such as influencing monetary supply. They are bankers who dont work for profits of a corporation but the well-being of a country's economy. I was just hoping to see objection to this item.--Roy17 (talk) 21:49, 3 July 2019 (UTC)

Date of birth reference

We now require a reference for date of birth, but not, for instance, date of death ... was their a discussion for that? I can see requiring it when we have contradicting dates, as sometimes happens, especially for early historical people. --RAN (talk) 23:56, 29 June 2019 (UTC)

It's just a suggestion, see: suggestion constraint (Q62026391): status of a Wikidata property constraint: indicates that the specified constraint merely suggests additional improvements, and violations are not as severe as for regular or mandatory constraints. Multichill (talk) 09:46, 30 June 2019 (UTC)
Coming from Wikipedia traditions I'd like to stress verifiability. Help:Sources allows unreferenced sources as an exception only. References are not to prevent criticism, but to allow it.
Spreading knowledge is of course always useful. @Tidoni: Can you share background to the change in date of birth (P569)? Recently copied to date of death (P570) too. Why are only some properties marked like this? Jagulin (talk) 13:00, 30 June 2019 (UTC)
The lack of a references last year led to a purge by some people of "ethnicity" and "religion" in entries. They were applying English Wikipedia rules to Wikidata. Again, I think it only needs a reference when we have two conflicting dates, each date should be referenced so reader can judge which one is canonical. A quick search shows we have >5,000 with multiple birth dates in the field. There already are several bots that scrape other databases for matching birth and death dates and add in the reference here at Wikidata. --RAN (talk) 19:03, 30 June 2019 (UTC)
No, I don't think they were applying Wikipedia rules, but the Wikidata policy on Living people as they may and should. --Dipsacus fullonum (talk) 08:58, 1 July 2019 (UTC)
If this is about living people, then why was it just expanded for the death date? --RAN (talk) 23:05, 1 July 2019 (UTC)
RAN, obviously having both verifiable birth date and death date is useful to judge if a person is living, but don't mix "reason for deletion" with "reason to have source". Thank you for highlighting that date of death was missing the symmetric enforcement. WD is, as most community driven projects, evolving in an iterative manner. I think it's very good that WD is at a point where promoting quality is possible. To me it's not at all dependent on "living person" policy, but person category is a large and important part of WD, so a good place to build quality and consensus. Jagulin (talk) 07:25, 2 July 2019 (UTC)
which was copied from BLP, QED. the venue may change but the deletionism is the same. you will not increase referenced statements by deletion. Slowking4 (talk) 12:00, 1 July 2019 (UTC)
@Richard Arthur Norton (1958- ): I was not promoting purge, but saying that Help:Sources and Wikidata:Introduction are clear: WD is supposed to be "A secondary database. Wikidata records not just statements, but also their sources". This is WD consensus, not enWp. Out of curiosity: Could you point at any Wikimedia project that have the policy you describe?
  • A source alone would require a fairly large amount of work to asses. For an automatic process it might be very difficult to judge the statement validity. There are however properties you can set on a contended statement (e.g. disputed (Q18912752)). A perfect knowledge system is perhaps a utopia, but we should not actively make WD guesswork.
  • What guideline says: only needs a reference when we have two conflicting dates? When that second date is entered it would be harder to find the source for the first date since we can't know where you found it.
  • You say bots add in the reference and I can only see that as a support for the need of references. Whenever there is no reference (added by bot or directly) the statement is IMHO unsourced and assumed to be speculative. Also if bots are good enough to actively add sources to correct persons, they could also add the date themselves. No need for you to do only half of their job. And "date of birth" is only a special case. The general case is that there are no bots that qualify all statements in WD. The contributor should do that.
I hope you now agree to the reasons and WD consensus for documented sources. -- Jagulin (talk) 10:27, 1 July 2019 (UTC)
  • Most claims should have a reference, especially P570. There just some that must have a reference already when first added. --- Jura 09:34, 1 July 2019 (UTC)
"Most claims should have a reference", then why this field and why now? --RAN (talk) 23:03, 1 July 2019 (UTC)
  • Wikidata has Living people. In contrast to Help:Sources, it's a policy. It's worth reading the policy as it specifies what classifying a property as property that may violate privacy (Q44601380) does and the process for challenging it's application to a property. As long as it's on date of birth (P569) I can understand having the constraint there. If you don't think it should be there, voice your opinion on the talk page.
I see no good reason to have the constraint for date of death (P570) as I don't see what makes date of death (P570) specially worth protecting. ChristianKl❫ 11:29, 4 July 2019 (UTC)

Took the place 2,5 years after the elections

Q64959563#P39. See the statements 1 and 2. In the first one he (the item) was elected at the day of the elections. In the second one, he wasn't elected the day of the elections. He was runner up. After 2,5 years the person who won resigned (he became a minister) and the item took his place in the parliament. Reading the statement, there is a gap between the elections new parliament start day (1996) and the start day of him as a parliament member (1999). Am not sure if someone can understand why. Xaris333 (talk) 21:17, 1 July 2019 (UTC)

  • You might be better served with a normal statement structure, one where the "start date" indicates the start date and the "end date" indicates the actual end date. --- Jura 21:26, 1 July 2019 (UTC)
But, this structure is already served. The "start date" indicates the start date and the "end date" indicates the actual end date. Xaris333 (talk) 21:47, 1 July 2019 (UTC)
As the guy is known to be immortal and stick to his seat? --- Jura 21:52, 1 July 2019 (UTC)
Sorry, I can understand the comment. Xaris333 (talk) 22:23, 1 July 2019 (UTC)
Instead of adding the start and end dates to the term, you add it to the position statement(s) of the person. --- Jura 22:39, 1 July 2019 (UTC)
Example? Xaris333 (talk) 22:47, 1 July 2019 (UTC)
I think all but one date. If the guy holds the position from 1991 to 1996 and from 1999 to present, there are just two actual start dates and one actual end date. --- Jura 22:57, 1 July 2019 (UTC)
But this has nothing to do with my first question. Xaris333 (talk) 23:07, 1 July 2019 (UTC)
If you wouldn't use fictional end dates, I don't think you would ask yourself the question. Election date and actual start date is rarely identical. --- Jura 23:09, 1 July 2019 (UTC)
You misunderstood or I wasn't clear, sorry. I don't use the election day as a start day. I use as start day the day they new parliament member starts. For example, election day was 22 May 2016 and the start day as parliament member was 2 June 2016... The "problem" is that the new parliament start at 1996 and he became parliament member at 1999. I am asking is that a problem the way I added the statement. Xaris333 (talk) 23:38, 1 July 2019 (UTC)
Did this person take the office because they were second in this years-earlier election? Or was that basically coincidence (or even a motivating factor in, but not legally the cause of them taking office), and this is like any time someone is appointed to a normally elective office? - Jmabel (talk) 05:19, 2 July 2019 (UTC)
The party won 1 seat in the parliament to that district. The person came second. So he didn't elected. After 2,5 years, the person who came first became a minister. So he quit from the parliament. According to the law, the seat is going to the person who came second in the elections. Xaris333 (talk) 06:47, 2 July 2019 (UTC)
We have a similair system in Sweden. I know from svwp, that such persons are not considered "elected" at all. Instead thay are seen as "Ersättare", . They have a seat in the parliament for shorter or longer period, but they often seem to lack the legitamacy an "elected" parliamantarian has. A user from svwp would then probably not add elected in (P2715) at all. But I do not know if that would be right here. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 07:19, 2 July 2019 (UTC)
The gap should not cause any problem. I would agree that it's an option to leave out elected in this case, yet having it as is seems perfectly fine, but not familiar with they specific system. If the source shows the person elected to be successor, that's all good. The way it was described (as well in Sweden) I would not dispute legitimacy, the person was still appointed by vote.
Possibly off topic: No related property found. Q10488457 and Q10684100 have hardly any links but would be variants of "ersättare". substitute (Q3504856) (french) and deputy member of the Parliament of Norway (Q16159375) seems to be intendedly country specific, though I don't see exactly why. I was surprised not to find any "successor/replacement" item. Relevant to make sure if it is a strict successor or also temporary stand-in when needed: substitute (Q29863201). runner-up (Q3275569) might indicate a successor-type of role, but most likely not. Jagulin (talk) 08:54, 2 July 2019 (UTC)
Q12009014 is a substitute for a member of a parliament as described in nowiki. --Dipsacus fullonum (talk) 10:01, 2 July 2019 (UTC)
Now merged with "Ersättare" from svwiki Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 08:20, 3 July 2019 (UTC)
Brought the off topic part up in noWp already. Jagulin (talk) 07:56, 4 July 2019 (UTC)

Thoughts about Wikidata

(Sorry for my English). I often realizes that Wikidata is a place were every user can do anything he/she wants. For similar subjects, they add different properties to say the same things. Everyone uses different structure and models. We don't have a common politic, manuals or examples... There is a mess... But we have thousands of users that contributes every day, we have bot and useful tools. But still we have a lot of problems... (And that problems make problems to Wikipedia also with the use of data of Wikidata to Wikipedia templates). That is Wikidata we want? In my opinion there are two ways:

1) Continue the same way...

2) Try to create common politic, manuals or examples. To discuss and find finals models. I know, there will be disagreements. But if the community agree to a model about a subject, then the users who have disagreements they have to accept the model... And of course we will need time to correct the items according to the model, but at least we will have a model that we can base on it to make corrections, to complete new items, to give examples to new users... And with the help of bot and tools, one day everything will be OK. I hope so...

Just some thoughts. I am not so expert in Wikidata and I don't use English language with perfect way, so I can't help as much I want, but I want to try... Or maybe I am wrong and there is no problem at all... Your opinion? (We need practical suggestions, not only theoretical). Xaris333 (talk) 22:48, 3 July 2019 (UTC)

We do have common politic, manuals or examples, such as Wikidata:WikiProject ShEx or Wikidata:Showcase items; or Wikidata:WikiProject Books#Work item properties and Wikidata:WikiProject every politician/Political data model, to give just two examples. We also have users who find it easy to criticise established practices, but who are slower to come up either with reasoned arguments to support their criticism, or with practical improvements; and users unaware both of extant models and of work done to promulgate such models, and who would wish us to reinvent what we already have. --Tagishsimon (talk) 23:04, 3 July 2019 (UTC)
  • There are many cases where we don't have established standards. The way to get established standards is to write them on the relevant Wikiproject and seek consensus there. Afterwards when you have consistent documentation of a standard on the Wikiproject it's possible to talk to the people who violate the standard. Sometimes this will mean that the people will reopen discussion about standards but the process of talking to people about the standard is the way you get people to adhere to a common way to model a domain. This means that there need to be people who care about enforcing standards for a given domain. For cases of more global standards it requires people writing RfCs. ChristianKl❫ 11:32, 4 July 2019 (UTC)
This keeps coming up. See the previous discussions "schema again" and "Glossary of topics on how to structure data". I'm sure Shape Expressions are the way of the future, but I suspect it'll be some time before they're fleshed out & in widespread use. I've been doing some work to try to come up with simple tables describing the common claims and qualifiers to be used to describe common entities (people, cities, rivers, etc.), but my work got interrupted by a computer crash so I don't have anything to show for it yet. —Scs (talk) 14:30, 4 July 2019 (UTC)

Wanted - someone to adopt a discontinued bot

The en:Wikipedia:WikiProject Women in Red is looking for someone who'd be prepared to run a bot, the owner of which has recently retired. The bot is described at d:Wikidata:Requests for permissions/Bot/Emijrpbot 6, which points to code here. The function of the bot is to add new wikidata items for new biographies and/or to add human and/or gender statements to existing wikidata items, based on articles found on en:Special:UnconnectedPages. WiR bases all of its metrics (& these) on wikidata records for articles, and since end April the project's stats have become increasingly hard to compile. We'd be more than grateful if someone would consider picking up this thankless task; thx. --Tagishsimon (talk) 17:44, 3 July 2019 (UTC)

(Really, I should not be taking more such responsabilities but here I go) If that would be welcome then I may be up for that − unless you find someone else of course :) Jean-Fred (talk) 21:24, 3 July 2019 (UTC)
If Jean-Frédéric can't do it, then I might be able to persuade Pi bot to help. Thanks. Mike Peel (talk) 21:33, 3 July 2019 (UTC)
Thx both; I have to leave it to you to decide. J-F - you know it makes sense; what could go wrong ;) --Tagishsimon (talk) 21:43, 3 July 2019 (UTC)
  • Personally, I do some of them once in a while, but, as I'm working with PetScan, this isn't possible any more. --- Jura 15:50, 4 July 2019 (UTC)
@Tagishsimon: It's nice code! I've merged it into one file at [3] and set it running, see Special:Contributions/Pi_bot. Unless there are objections, then I just need to know how often you want it running. If @Jean-Frédéric: wants to take this on, though, then I have no objections. :-) Thanks. Mike Peel (talk) 18:04, 4 July 2019 (UTC)
Could it add full dates instead of just years? e.g. at Q65032839, Q65032842, both linked to articles with actual dates. --- Jura 18:14, 4 July 2019 (UTC)
@Jura1: It currently uses the categories to get the birth/death dates, not the text. It should be possible to change that, though, I'll investigate. Thanks. Mike Peel (talk) 18:27, 4 July 2019 (UTC)
@Jura1: done, as long as the enwiki article uses en:Template:birth date and age. Thanks. Mike Peel (talk) 19:21, 4 July 2019 (UTC)
Excellent! Thanks. --- Jura 19:27, 4 July 2019 (UTC)


albedo (P4501) has image (P18) as a constraint. It is probably rarer that we have (downloaded) pictures of a celestial body than that we don't have. Could it be downgraded to a recomended property instead? Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 07:05, 29 June 2019 (UTC)

@Arbnos: You added this constraint. Any comments? item requires statement constraint (Q21503247) description says "should". Is mandatory constraint (Q21502408) enforcing it more strongly?
I don't know astronomy, and have hardly any wisdom of WD structure design, but I also noticed this as a bit tough. There must be lots of sources for albedo where there is no free image to use. Is the image meant to be a graph, such as in albedo (Q101038)? Maybe even "allow" is enough?
parent astronomical body (P397) is also required. Maybe meant to be the star related to the item, but AFAIK the albedo is not strictly bound to having a "parent". And the Moon (Q405) has albedo, and parent Earth.
Were the constraint meant to be on "albedo qualifiers" rather than on "item properties"? And should they be mandatory in that case? Jagulin (talk) 16:03, 1 July 2019 (UTC)
@Jagulin: Hello! The image I added when the property had limited use and it seemed that the absence would be an exception, so it can be removed.--Arbnos (talk) 16:20, 1 July 2019 (UTC)
@Arbnos:I take that as support for removing the image constraint entirely and I agree. I believe you also added parent astronomical body (P397) constraint and the existing exception. I suggest remote those too. What do you say about it? What is the meaning of the constraint? Jagulin (talk) 05:48, 2 July 2019 (UTC)
@Jagulin:Yes, you can delete the image. As for parent astronomical body (P397): the description of the property, like all asteroids, for example, should be the parent of warm Sun, even a "star" to be synonymous worth it.--Arbnos (talk) 17:57, 2 July 2019 (UTC)
We have objects about thousands of minor planet and comets. Since (with one single exception) all of them have Sun as parent astronomical body (P397), it maybe looks a little redundant to add this statement to them all. But as it is doable, so why not? A bot/script could add it everywhere for every comet or minor planet without causing any problems at all. Yes, there are many comets who do not circle around the Sun today. But at some point they have at least been doing that. Not a single one of them (yet) have seemed to have a foreign background. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 19:35, 2 July 2019 (UTC)
@Arbnos: I removed the image constraint, as agreed. About my other suggestion:
  • "parent astronomical body (P397): major astronomical body the item belongs to" says nothing about "parent of warm Sun" (nor "child" which I guess you meant). I already mentioned Moon (Q405) as a counter-example, but there are plenty.
  • "albedo (P4501): ratio of reflected radiation to incident radiation" says nothing about being dependent of a Sun or a parent. It is an instance of "Wikidata property for astronomical objects (Q21451142): Wikidata property" but that doesn't say "child" to me.'
  • Structurally it doesn't IMHO make sense to have "parent astronomical body" as requirement of Property:P4501#P2302. The planet (Q634) needs a parent even if nobody entered albedo value. "mean anomaly (P2325): element of orbital definition" has the requirement, but in this case it makes sense due to the "orbital".
  • On the other hand, it's not a problem and I said I don't know astronomy... If consensus is for the requirement to stay, I won't remove it.
@Sextvå.tvånoll.ettsjunoll.sjufyra: I may have misunderstood the context of your recent comment (sorry for mixing another topic into your original question). I totally agree that parent astronomical body (P397) should be set when known. Interestingly, you seem to answer a question I asked in another place (last bullet of this). If you can bot those, it would be nice! Or can it be inherited (e.g. from asteroid belt (Q2179)) and remove the effort entirely?
  • I do however question if you say "the Sun" meaning Sun (Q525). Naturally we know more about our current solar system, but I'm sure there are planets identified for other stars as well and as mentioned above "parent" is not always a sun. "minor planet and comets" is perhaps difficult to find at a distances, but are they really conceptually defined only to our Sun? You can bot now, but later it may be untrue. asteroid belt (Q2179) is well defined, but I would suppose there are "other asterid belts" both in our system and for other stars. Jagulin (talk) 08:36, 3 July 2019 (UTC)
@Jagulin: We have found exoplanets around other stars, but they are neither minor planets or comets. (Minor planets is the sum of asteroids in the inner part of the solar systems and the transneptunian objects and everything there between.) And we have found belts of asteroids around other stars, but not identified single objects. We have maybe detected swarms of comets in other solar systems, but not single comets. So there are comets and minor planets in other solarsystems, but we should have not items about them. There are from time to time comets who are encircling Jupiter and Saturn and not Sol, but they have in the past been encircling Sol, so adding that to the items is never wrong. The only known exception to this is 1I/ʻOumuamua. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 08:52, 3 July 2019 (UTC)
@Sextvå.tvånoll.ettsjunoll.sjufyra: Thanks for the added insight!
* We agree that you're not talking about Albedo it seems. I think it would make sense for someone to add the "parent" constraint to the item classes that should have it, e.g. planet, asteroid...
* You suggest to inject "parent" value into specific kinds of objects. That's out of my league, so as long as you know what you do and have support, I don't mind. :) As a general rule I'd not inject data into WD if I had no source for it. I prefer leaving "gaps" and let the "reader" fill them in by any assumption they're willing to do, rather than add statements that the reader will have to question.
* I don't know the definition of "parent" really, but I would expect the data to be "current" (is that within 1000 years?), stable and that the basis is gravitational. That was only my guess however. You seem to indicate that comets are bound to the Sun, but occasionally spin around a planet before they head out to the Sun trajectory again. If there are some sources that say the "parent" is Saturn and some say "Sol" then they are both valid statements and you may need to Rank or timestamp those facts.
* 1I/ʻOumuamua (Q42313338) was interesting. Rather than relax the need of parent astronomical body (P397) for that item (such as in albedo (P4501)), maybe parent astronomical body (P397) could be set to "no value" or "it's complicated" and have the exception stated in parent astronomical body (P397) only? I think that would make the structure more rigid.
* To get out of this trouble is also the reason why I suggested albedo (P4501) not to have constraints for "parent". Jagulin (talk) 16:50, 3 July 2019 (UTC)
  • Comet Shoemaker–Levy 9 (Q3076) was for a period a "pseudomoon" to Jupiter before it collided with the planet. The dynamics of orbits is a very interesting subject. Most small objects that are trapped by the gravity well of large planets, tends to run away after some years. They may stay for decades, but sooner or later the kinetic energy they have let them run away. There are exceptions. Both moons or Mars are probably caught out of the asteroid belt. Many of Jupiters moons probably have the same origin. The exact definition of parent, I do not know if it is scientfically defined. My guess is that it is mainly a WD-template thing. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 17:25, 3 July 2019 (UTC)
After more thoughts, I'm not found of the "originating" use of "parent". I did however open for discussion in Wikidata_talk:WikiProject_Astronomy#Definition_of_parent_and_child? in hopes of more expert opinions. I suggest we take the discussion there from now, since your original Albedo issue was solved. Jagulin (talk) 10:21, 5 July 2019 (UTC)

Same person?

Arthur Devine (Q16943844) and Arthur Devine (Q64980647)? --Magnus Manske (talk) 07:45, 2 July 2019 (UTC)

  • I'd think so, though it's unclear where the P569=1860 comes from. --- Jura 08:15, 2 July 2019 (UTC)
I've asked Mpaa about 1860 here. --Tagishsimon (talk) 09:12, 2 July 2019 (UTC)
I can't recall but 1849 seems correct, according to this s:Page:The_Catholic_encyclopedia_and_its_makers.djvu/73. I must have mistakenly taken the 1860 from the author below in such page.--Mpaa (talk) 20:56, 2 July 2019 (UTC)
Without looking into the issue myself, I would point out a situation that I am familiar with - Q7376409 and Q7376408. Both named Ruby Wright. Both blond. Both singers. Born 25 years apart. Died 5 years apart. The internet has the pair so mixed... But they are different people. It's one thing for common names like "John Smith", but I have even found "oddball" names that were not the same people despite similar details. Quakewoody (talk) 21:10, 2 July 2019 (UTC)
I think they are the same. References indicate same works for both of them.--Mpaa (talk) 21:18, 2 July 2019 (UTC)
"one thing for common names" you say. I'd expect the "odd" names to be a bigger problem because people (as in "even otherwise legitimate publications") may easily assume the name to be unique. Once it's in a legitimate source it will spread. For "generic name" I'd expect a bit more care, but you're probably right that mixups are even more frequent, since those names are also frequent. Since WD is based on other sources, not here to decide what's right or wrong, I don't know how to get around the problem. Marking "deprecated" if found, but there is no way to stop it. It would help a lot if we at least can avoid merging items based on the mixup. On the other hand, how to make sure (when to create an item) that the already existing one with a similar name is actually the same, so maybe that's a reason for duplicates too? Jagulin (talk) 06:20, 3 July 2019 (UTC)


Spind (Q6515721) and Spind (Q61955092) --RAN (talk) 13:51, 4 July 2019 (UTC)

No! Add different from (P1889) instead. The former was the capital of the municipality with the same name. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 14:07, 4 July 2019 (UTC)
There is no settlement called Spind in Norway. According to enwiki the capital of Spind municipality was Rødland (no Wikidata item). Besides Spind (Q6515721) have an area of 43 km² which is much too big to be a settlement in Farsund (Q490462), so I suppose both are the former municipality and should be merged. --Dipsacus fullonum (talk) 16:33, 4 July 2019 (UTC)
You are right. I was tricked by Spinds church. I thought I saw a village around it, but that is something else. And next to that "something else" is Rødland. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 17:35, 4 July 2019 (UTC)
I originally split the item, because while Spind municipality existed until 1965, Spind definitely still exists as "something" even after the municipalities merged. Store norske leksikon and Kartverket both call it a parish of Norway (Q814691), so I suppose we could do the same at Spind (Q6515721). Einar Myre (talk) 06:39, 5 July 2019 (UTC)
I changed the wrong and confusing instance of (P31) human settlement (Q486972) to instance of (P31) parish of Norway (Q814691) for Spind (Q6515721) and updated descriptions for no, da, en and deleted probably wrong descriptions for ar, de, nl.
Probably a Geonames-ghost. (Hej!) 13:08, 5 July 2019 (UTC)

Update: deployment and migration plans for wb_terms redesign

Hello all,

This is an update regarding the progress and dates of migration of wb_terms table replacement solution in Wikidata production environment.

We have successfully put Wikidata production in the stage where property terms are stored in and written to both the old store (wb_terms table) and the new replacement store. Retrieving property terms is still being done using the old store.

The previously announced dates are no longer effective. No changes to tools are needed yet. Tools can continue to read from the old store (wb_terms table) for the moment. There will be a later announcement regarding the date when tools have to switch to reading property terms from the new store.

The next step will be to go to the stage of retrieving property terms from the new store, while we keep storing them in both stores. That step is blocked by a problem we discovered while testing that switch on beta cluster, and we are working on solving it at the moment (Task on Phabricator).

As for item terms, and in the light of the new information about switching master node to a better host (see Failover s8 (wikidatawiki) on Phabricator) that can actually host the migration of them till the end, we have as well decided to push out item terms migration on hold until after that failover is done and is stable.

The migration of all item terms will take weeks to finish, but it isn’t clear yet how long exactly. We will run it in several stages and there will be separate announcements regarding those stages to announce and inform about how to deal with it, in case it affects your work.

You can find more information regarding those dates and how to prepare for them in this task, and we have dedicated a board to receive and help with any questions from tool builders that need to update their tools accordingly.

In order to keep all discussions in one place, we kindly ask you to react or ask your questions on Phabricator.


--Alaa Sarhan (WMDE) (talk) 10:39, 5 July 2019 (UTC)


I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 11:05, 11 July 2019 (UTC)



UsernameReviewerBot (talkcontribsnew itemsSULBlock logUser rights logUser rightsxtools)
Operator: JJBullet (talkcontribslogs)

Task/s: Creates a list of users - in userspace - that may have inappropriate names or edits, and sends them to administrators for review.

Code: N/A

Function details: Creates a list of usernames (In userspace) that may be inappropriate or that may have made inappropriate edits / be offensive to another user, sends them to admins / crats for review. --~JJBullet~ {Talk} 09:10, 4 July 2019 (UTC)

why are you importing english wikipedia drama? is there any evidence of inappropriate user names? what is your criteria for inappropriate? what is your criteria for inappropriate edits? is this scope creep? Slowking4 (talk) 10:28, 4 July 2019 (UTC)
Do you need to be so rude? i am simply carrying on with my other request~JJBullet~ {Talk} 10:31, 4 July 2019 (UTC)
would you care to answer the questions? you do not have a consensus, see also Wikidata:Requests for comment/Adopt a username policy - Slowking4 (talk) 02:32, 6 July 2019 (UTC)
  • You don't need to have bot permissions to read the list of user names. I would propose that you do the first 5 cases manually and then we see whether that's something worth automating. Apart from that, while it's nice to have a link in the project chat, the actual request should still be at requests for permission. ChristianKl❫ 11:08, 4 July 2019 (UTC)
    The bot may have gain of the apihighlimits rights. Sextvå.tvånoll.ettsjunoll.sjufyra (talk) 14:11, 4 July 2019 (UTC)

The above discussion is preserved as an archive. Please do not modify it. Subsequent comments should be made in a new section.

Uncertain date

Anastasiya Pavlovna Knyazeva (Q44979759) have 1 July 2011 as birth date. However, is there various website that have 30 June, other 1 July. On the Knyazeva YouTube channel was uploaded today 4 July a video which celebrate her birthday... So, when Anastasiya Knyazeva was born? 30 June, 1 July or 4 July? -- 17:55, 4 July 2019 (UTC)

  • No idea. From a Wikidata perspective, you can add all dates, each with the reference that supports it. --- Jura 18:01, 4 July 2019 (UTC)
  • Day of video upload on YouTube is not the same the birthday. According post on her Instagram she clebrated birthday on 1 July. The same is in Russian magazine's article. --Ksc~ruwiki (talk) 20:32, 5 July 2019 (UTC)

dewiki:Frauenzentrum Westberlin and enwiki:West Berlin Women's Center

Moved from WD:RfD

These two objects both describe the same entity, but I can't merge them. Could someone please help me? -- 17:35, 8 July 2019 (UTC)

✓ Done --GPSLeo (talk) 19:31, 8 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 09:02, 11 July 2019 (UTC)


I notice that some people can add comments to edits, how can I do that? I would like to let people know why I may be reversing an edit, they should see the rationale so they do not add it again. --RAN (talk) 18:19, 10 July 2019 (UTC)

In history, when you hit "undo" or "restore", you can write the explanation. --Matěj Suchánek (talk) 09:02, 11 July 2019 (UTC)
This section was archived on a request by: --JJBullet❫ 09:22, 11 July 2019 (UTC)

Format fix property proposal?

Can someone fix Wikidata:Property proposal/begin and end of covered period ; covered period so the discussion section appears? There's something funky with the ping templates but I can't figure out what it is. - PKM (talk) 21:07, 10 July 2019 (UTC)

Fixed. Not sure what exactly it was... --Yair rand (talk) 03:48, 11 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 08:59, 11 July 2019 (UTC)


Shouldn't Q803626 and Q23932292 be merged? Sorry if this is not the correct place to suggest a merge. -- 08:14, 11 July 2019 (UTC)

✓ Done --JJBullet❫ 08:50, 11 July 2019 (UTC)
Also it would be best to suggest a merge at the RFD JJBullet❫ 08:49, 11 July 2019 (UTC) sorry, my mistake--JJBullet❫ 09:04, 11 July 2019 (UTC)
No, that's where deletions are requested, not merges. A merge can be conducted by anybody. --Matěj Suchánek (talk) 08:59, 11 July 2019 (UTC)
I was literally about to strike that out but you got there before me, thanks :) --JJBullet❫ 09:01, 11 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SilentSpike (talk) 09:16, 11 July 2019 (UTC)

John Pierce (Q65054514)

That person, en:John Pierce (tenor) - too many John Pierce - is the same as de:John Pierce (Sänger). Sorry, I failed to make the connection when creating. --Gerda Arendt (talk) 09:19, 11 July 2019 (UTC)

This is just my thoughts, but why would we need to deal with stuff from outside wikis? correct my if im wrong?--JJBullet❫ 09:20, 11 July 2019 (UTC)

✓ Done--JJBullet❫ 09:55, 11 July 2019 (UTC)

This section was archived on a request by: --JJBullet❫ 09:55, 11 July 2019 (UTC)

Edit filter to prevent edits that are obviously wrong?

Why don't we activate an editing filter that prevents edits like the following, that are obviously wrong, to be entered into Wikidata in the first place?

[4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

These are just examples on when a taxon (Q16521) or a film (Q11424) has been used for country of citizenship (P27), probably by mistake as the name is similar to the name of the correct object. One of the edits in the list above is almost three years old, and has not yet been corrected! --Larske (talk) 16:39, 28 June 2019 (UTC)

Edit filters have very limited checking features and their maintenance is restricted to sysops. On the other hand, the system for constraints was developed for Wikidata and is still evolving, unlike edit filters developed mainly for Wikipedias. --Matěj Suchánek (talk) 19:17, 28 June 2019 (UTC)
Sorry if I used the wrong terminology. It doesn't have to be an "edit filter" if that has a particular meaning in the Wikimedia context.
I rephrase my concern: Why don't we prevent, or automatically (by a bot) revert, edits that obviously violates the most obvious contraints like a country of citizenship (P27) must not be a taxon (Q16521)? It is very bad that such edits that can be detected at once remain uncorrected after months or years. --Larske (talk) 22:04, 28 June 2019 (UTC)
It's good that you've found these anomalies.
The first thing to ask is whether there's a proper, governing constraint. Clearly, the target of country of citizenship (P27) should be a state (Q7275) -- and, in fact, there is such a constraint at P27.
The second thing to ask, then, is: Why is the constraint not working, in that (for example) the P27 claim at Juan Carlos Mareco (Q3466333) is not flagged as violating it? [Never mind, the constraint seems to be working fine. I suspect the reason it seemed not to be working for me was due to an unrelated issue in my environment.]
The third thing to ask is, if the constraint were working, why have some of the items you mentioned been in violation of their constraints for so long? Are there "you can help" / "things you can work on" lists for constraint violations? Do volunteers need to be prodded to whittle down those lists?
And then, finally, to your specific suggestion: should there be a mechanism to simply prevent edits which supply new or modified claims which violate certain (or all) constraints?
Me, I don't know the answer to any of these questions except the first, but perhaps someone else here does. —Scs (talk) 22:56, 28 June 2019 (UTC) updated 11:31, 30 June 2019 (UTC)
Two answers: 1) there are reports for most/all constraints, listed in the talk page of the property; for example, here's a report for P39. 2) We should not prevent edits that break constraints. In general, constraints are simple and the world is complex; the consequence of preventing the addition of truthful information because it is outside the compass of a constraint, is greater than the consequences of addition of data that breaks the constraint. --Tagishsimon (talk) 12:04, 30 June 2019 (UTC)
Well, reports are fine if they are acted upon. But in reality faulty statements can be left uncorrected for years as you can see among the few examples I listed. I understand, that not all existing constraints may be appropriate to use for stopping edits, so they can not be stopped "in general". But what is your opinion about when people enter a taxon (Q16521) or a film (Q11424) as the value for the country of citizenship (P27) property just because they have identical or similar labels. I found it very strange that such entries are not prevented.
Maybe we need another type of constraint that do not list what is acceptable (because there may always be exceptions in "complex world" as you say), but what is not allowed (i.e. not as in never, with no exceptions), based on experience of common mistakes like the ones I listed. Or, if not stopped, at least "delayed" with a "CAPTCHA" that humans, but not bots, can successfully pass if they actively confirms that there is not a mistake that they want to enter an taxon (Q16521) (or subclass thereof) as the value for the country of citizenship (P27) property. Do you think that would be possible?
Or maybe such constraints already exist but are not used. We have country of citizenship (P27)/property constraint (P2302)/value type constraint (Q21510865) with six listed class (P2308) and relation (P2309) equals instance of (Q21503252), but I guess what I am thinking about is a number of class (P2308) with relation (P2309) equals "is definitely not instance of (Q21503252)". The class (P2308) listed here can of course not be all classes except those listed above, but just those where experience has shown that mistakes are common. I think taxon (Q16521) and film (Q11424) would be two of my candidates for such a list for the country of citizenship (P27) property. Suggestions for such "commonly mistaken" class (P2308) can of course be found in the constraint violations reports that we have today. Do you see what I mean?
I also have a questions on the reports you refer to. You refer to the report for P39 when my examples were all about P27. If I look at the Constraints violation report for P27, I can not find all objects in my example list. The report says the violation count is 10914 and not all violations are listed. At the bottom of the first list it says: "Too many results. 5913 records skipped.". And if I try to run the "SPARQL" mentioned on the talk pages it results in timeout, and the "SPARQL (new)" question only lists 299 items. @Tagishsimon: What am I doing wrong here? --Larske (talk) 16:00, 30 June 2019 (UTC)
Bad idea. Wikidata allready has implementations in the Wikibase extension which enforce constraints. Even if that is unfinished or has some bugs in it, then the extension should be fixed. Abuse filter is not the fix for that. Also, I protest to the use of the term "Edit filter" for the Abuse filter. That is just enwiki jargon, based on different policies than this project has.--Snaevar (talk) 17:42, 6 July 2019 (UTC)

Q482980 proposed to rename in Japanese language

As announced on wikidata:project chat in Japanese language, there is a proposal/request-for-edit in progress on Talk:Q482980 by a Japanese editor: rename Q482980 from (A) "作家" (ja) to (B) "著作者" (ja), or the copyrighted author of things. The proposer lacks editing rights, and the proposal is in two folds to "swap" page labels between two wikidata pages as:

  • Q482980 needs to be relabeled as (B) "著作者" (ja) and should be wikilinked on jawp to (B) w:ja:著作者 as well. Currently wikilinked to (A) or on jawp w:ja:作家, and enwp link w:en:Author should remain as is. The proposer is drafting and planning to rewrite jawp page w:ja:著作者 extensively from the viewpoint of law field.
  • Q11620855, currently linked to (B) on jawp (w:ja:著作者), needs to be re-labeled as (A) "作家" (ja). FYI, (A) includes ISCO occupation code, as a creator of things. Please join and drop your thoughts.  --Omotecho (talk) 18:36, 3 July 2019 (UTC)
If the page is protected, you could use {{Editprotected}} or ping the protecting admin. --- Jura 09:11, 6 July 2019 (UTC)

problems with query API

Dear Wikidatas,

in my Python-code I use the module "SPARQLwrapper" referring to for retrieving a bunch of features on authors of research articles. It worked for many weeks. However since about two weeks I get Error messages, mostly "HTTP Error 403: Forbidden" and some times some Error 402 (?) due to too many requests. Then yesterday afternoon (Berlin time) it worked again for half an hour.

Whats going on? How can I fix my code? Or do I just have to wait till you fix something behind the scene? I tried to join the Wednesday tech-chat but I couldn't manage to join the channel on freenode.

Looking for you reply! Thank you very much in advance! Cheers! Eva from Cologne — Preceding unsigned comment added by (talk) 07:42, 4 July 2019‎

I had this problem when I hadn't specified a user agent in SPARQLWrapper. When I added it to my tool, I stopped getting 403 errors. --Vesihiisi (talk) 10:52, 4 July 2019 (UTC)
Are you following the User-Agent policy? We’ve recently started to enforce this more strictly, I believe; specifically, the default python-requests user agent is currently blocked. (Even if the problem turns out to be something else, sending an informative User-Agent header is good practice anyways.) --Lucas Werkmeister (WMDE) (talk) 10:56, 4 July 2019 (UTC)
ha! that worked! thank you a lot! Eva

Can we create a wikipedia page for a new company?

I wanted to know, whether we can create a Wikipedia page for a new company. I had tried that a few months ago, but the page was deleted, and my IP also got blocked. And I specifically don't know why?

Since I do not know what the company is and I do not know which version of Wikipedia you were editing and which IP you were using, it is impossible to tell here. But most likely the problem was that the subject of the article don't fulfill the standards we would like to see at Wikipedia. en:Wikipedia:Notability for example tell: "Wikipedia articles cover notable topics—those that have gained sufficiently significant attention by the world at large and over a period of time, and are not outside the scope of Wikipedia. We consider evidence from reliable and independent sources to gauge this attention." New companies normally do not fullfill such demands. And articles created by persons close to the subject, often look more like marketing of the subject than an encyclopedic article. That can also be a reason why it was deleted. IP (Hej!) 07:12, 6 July 2019 (UTC)
List_of_bad_article_ideas Why_was_the_page_I_created_deleted%3F It occurs frequently that self-promotional articles are created of naivety or ignorance and deleted short time later. But you should not have got banned for that. Probably you tried to re-post your article several times or edited many pages to link to your page. As about why, there is a design flaw in wikipedia: when an article is deleted then it vanishes from both the recent changes and your contributions. But the act is visible in the deletion log. The sysop performing the deletion is supposed to give reasons for that. But your case is easy to guess: inappropriate self-promotion. The answer to the core question is NO. Do NOT try to create a wikipedia page for your company, do NOT retry with other public wiki or other IP. Even if your company becomes notable then you should NOT start an article or substantially enhance such if it already exists. Taylor 49 (talk) 09:08, 6 July 2019 (UTC)

Bot that changes [[Q12345]] to {{Q|12345}}

I frequently see it that people who are used to the style of links within Wikipedia use links like [[Q12345]] inside discussion on this project. Likely, because they don't know about our templates. I'm of the opinion that the template is always better then the interwiki-link. I feel it would be great to have a bot that automatically changes [[Q12345]] to {{Q|12345}}. How do other people feel about the issue? ChristianKl❫ 09:03, 2 July 2019 (UTC)

  • Not really, personally I use Q123 when the label isn't helpful. What I do find annoying are QIDs without any links at all. --- Jura 09:06, 2 July 2019 (UTC)
Even if the label isn't helpful, what use does it have to be shown the number? ChristianKl❫ 09:24, 2 July 2019 (UTC)
Normally, there should be some text around it explaining the item. In any case, I don't think a bot (or a user) should edit other users signed comments. --- Jura 09:35, 2 July 2019 (UTC)
Even if there's text around it explaining it I don't see how the {{Q|12345}} form is worse. I generally do believe that edits that fix spelling mistakes or formatting issues are okay. I spent a lot of time on StackExchange where that happens very frequently and the advantage of the Wiki system is that editing is possible. ChristianKl❫ 10:15, 2 July 2019 (UTC)
A wiki talk page isn't exactly a wiki article. Given the problems we have had around here with people doing it .. let's not. There is actually no way to be sure what users see when you add {{Q|12345}}. --- Jura 10:27, 2 July 2019 (UTC)
I agree that mass-editing discussions should be avoided and StackExchange culture and answering is different from here. If nothing else, the history would be a mess. It could possibly be a feature of the editor (remember Office Assistant (Q1042885)?) so inexperienced users would be asked to (auto)change before saving. Another possibility is to have a WD plugin or local browser plugin to show any matching ID (template or not) the way you perfer it: Something alike "Descriptions: Show the description of..." in gadgets. Anyone knows if it is available? The downside of local fixup is that there may be confusion in the discussion when you see different things. Jagulin (talk) 06:59, 3 July 2019 (UTC)
I kinda wish the template could pull in the number only where the relevant description is missing, as I agree in most instances the numbers are not that useful. Circeus (talk) 02:11, 3 July 2019 (UTC)
@Circeus: To me, seeing the number is quite important. The same description may be used both for Q and P and it's very relevant to separate those in a glance. When talking about possible duplicates, it's also needed to keep track of items with similar names but different numbers. I "always" use Q-template (e.g. Count von Count (Q12345)), but maybe that's annoying to someone. There us also Q'-template (Count von Count (Q12345)  View with Reasonator View with SQID) and Q+-template (Count von Count (Q12345): character on Sesame Street) and maybe more (one showing only number but having details when hovering?) so the idea is that the writer decides which they think is most relevant at each case. Jagulin (talk) 06:59, 3 July 2019 (UTC)
  • I used to use {{Q|12345}} first time in discussion in order to help other editors to know the meaning, but not for the further appearances. Thanks by the proposal. Amadalvarez (talk) 09:29, 2 July 2019 (UTC)
Using regular links and not {{Q}} is also a good idea when the discussion is about the identity or labels of the items themselves – using their current labels for them can make the discussion very confusing to read in the future (“should imported from be changed to be Wikimedia project-specific?” vs. “should imported from Wikimedia project (P143) be changed to be Wikimedia project-specific?”). See the topic right above this one (#Same person?) for another example – using {{Q}} for items with the same label isn’t very helpful. --Lucas Werkmeister (talk) 16:54, 3 July 2019 (UTC)
My first reaction is that this sounds like a problem with the template. Ideally, the template would show both the label that was applicable when the template was first used and the current label. In any case I'm not advocating that the form [[Property:P143|imported from]] should be edited by the bot. ChristianKl❫ 21:09, 3 July 2019 (UTC)
  • Perhaps this could be done with JavaScript, so one can choose to use it or not? Might be expensive though, don't know. Best would be if Navigation popups (Q11305696) displayed the label in the user language with fallback. Unfortunately, it's half-broken for Wikidata. --Marsupium (talk) 15:12, 5 July 2019 (UTC)

So again: How to link to a section if that is using {{Q|XXX}}?

If there can't have a user-friendly solution of this question, then I would consider nominating zh:Template:Q for deletion. --Liuxinyu970226 (talk) 23:50, 5 July 2019 (UTC)

  • This makes no sense at all. (1) You begin the header here with "again," but this appears to be your first comment on this thread. (2) You refer to a template on the Chinese-language Wikipedia, but this discussion has been about a template on Wikidata. (3) The fact that a template might not be entirely suitable for use in a section header is not a reason to delete the template. (4) You should still always be able to get a suitable link to a section header by left-clicking in the table of contents to go to that section, then copy-pasting from the URL bar. - Jmabel (talk) 08:53, 6 July 2019 (UTC)
@Jmabel: I was asked at Template talk:Q for several times, and until your comment, no answers. --Liuxinyu970226 (talk) 09:03, 6 July 2019 (UTC)
  1. The account you are editing under has no posts at Template talk:Q. Were you editing on a different account?
  2. What about my points 2, 3, & 4? - Jmabel (talk) 09:15, 6 July 2019 (UTC)
What the... there is no trace whatsoever of any such discussion by anyone at all either at [[[Template talk:Q]] or at zh:Template talk:Q... Circeus (talk) 22:22, 6 July 2019 (UTC)


Is it legal to have adjectives as items? There is an item about "small" Q24245823 but none about "big". Also we have two items about "case sensitivity" Q55121384 and Q257869 and one item about "case sensitive" Q55121148. Are adjective items desirable? Taylor 49 (talk) 16:55, 5 July 2019 (UTC)

  • If, as for small (Q24245823), Wikiquote has a page for an adjective, aren't we pretty much obligated to have an item? - Jmabel (talk) 19:58, 5 July 2019 (UTC)
  • Yeah, we pull items from a lot of different wikis that can cause apparent duplicates with lexeme. I've applied subject lexeme (P6254) to small (Q24245823). the other two seem to be strictly because of stackoverflow tags that happen to be written in the adjectival form. I would straight up merge those with the matching qualities' items. Circeus (talk) 22:35, 6 July 2019 (UTC)

Help with Q65052377

What needs to be changed for the reference to Haida people to work? Not sure if I am using the wrong property, or if something needs to be changed at the Haida people item. Also, what is the appropriate way to connect Northwest Coast art to this item? Right now it's "genre" which doesn't seem quite right. Thanks! Calliopejen1 (talk) 00:03, 6 July 2019 (UTC)

@Calliopejen1: Haida is defined as an "ethnic group" (as are most US indigenous peoples in WD). ethnic group (Q41710) is (ultimately) a subclass of group of humans (Q16334295). The property culture (P2596) wants a value that is a people (Q2472587) or a civilization (Q8432). There are two ways to fix this. Quick and easy, make Haida <instance of> both ethnic group (Q41710) and people (Q2472587) "plurality of persons considered as a whole, from a government perspective", which I think is correct. We could also get consensus to add ethnic group (Q41710) as a valid value for culture (P2596). I'd vote for doing both.
I'd also make Native American tribe in the United States (Q7840353) an instance of people (Q2472587), since a recognized tribe certainly has a political aspect, and eventually add <instance of> "people" to all other indigenous peoples in the US that are not formally recognized tribes.
Typically, "art of [place]" isn't a genre, but a subclass of art. However, tribal art (Q2864736) is an art genre, with subclass Indigenous Australian art (Q420716). So Northwest Coast art (Q7059995) could be a subclass of either "art" (in which case this mask in an instance of it or part of it), or it could be a genre, subclass of "tribal art", equivalent to Getty AAT's Northwest Coast Native American styles. I think either one is valid as long as we're consistent. We have very little coverage of Native American Indian art in Wikidata, so we can set an initial standard with what we do here. - PKM (talk) 20:39, 6 July 2019 (UTC)
(And we could of course have two items for "Northwest Coast art", one a subclass of art and one the genre linked to it), which I think is good ontology. - PKM (talk) 20:44, 6 July 2019 (UTC)
@PKM: I am so new at Wikidata that my ability to weigh in here is limited. I will say that the ethnicity issue is a common problem I've encountered not just for Native American/Alaska Native art but also for any art or craft that is associated with an ethnic group (e.g. Nomoli figurine (Q3343234), Kutch blouse E77824 (Q65088082)). Re: Northwest Coast art, Northwest Coast art (Q7059995) is associated with a geographic region, but it is a style of artwork -- I wouldn't call it "art of [place]". It probably could be a subclass of tribal art, though some non-Natives are creating artworks in this style as well (see e.g. [21], [22]). Calliopejen1 (talk) 21:17, 6 July 2019 (UTC)
@Calliopejen1: Thanks for those links. We do have Visual arts by indigenous peoples of the Americas (Q7936583) I missed it before. It's a <subclass of> visual arts, and has no subclasses of its own.
While we tend to use Getty AAT as a reference vocabulary for decorative arts, where we lack anything better, we don't follow it uncritically. We have more or less deferred the entire question of "styles, periods, and cultures" (hierarchy here). FWIW, Getty does not attempt to group "tribal art" across geographic regions.
We have the concept art style (Q1792644), <different from> genre (Q483394). art style (Q1792644) is used very loosely for a lot of different concepts, but it would certainly be appropriate for Northwest Coast art (Q7059995). We don't have a specific property that goes with this concept, but we might propose such a property - we have genre (P136), movement (P135), and architectural style (P149), all of which are adjacent to "style" but not exactly the same. For now, we can make Northwest Coast art (Q7059995) <instance of> "style" and reference the Getty AAT item. - PKM (talk) 00:49, 7 July 2019 (UTC)
Another property you may find useful is indigenous to (P2341) for crafts and so on (not individual objects). - PKM (talk) 01:01, 7 July 2019 (UTC)
@Certainstars: Do you have any thoughts on this? - PKM (talk) 20:44, 6 July 2019 (UTC)

"Collaborating on the sum of all knowledge across languages"

Hi! I really try not to spam the chat too much with pointers to my work on the Abstract Wikipedia, but this one is probably also interesting for Wikidata contributors. It is the draft for a chapter submitted to Koerner and Reagle's Wikipedia@20 book, and talks about knowledge diversity under the light of centralisation through projects such as Wikidata. Public commenting phase is open until July 19, and very welcome: "Collaborating on the sum of all knowledge across languages". --Denny (talk) 21:49, 6 July 2019 (UTC)

Thank you very much for share it, Denny! I have pending to review it and give some comments if I have something useful to contribute. Regards, Ivanhercaz (Talk) 00:09, 7 July 2019 (UTC)
Thank you! --Denny (talk) 00:10, 7 July 2019 (UTC)

Stated-as as a qualifier on dates

special:diff/975860943 was done because Chinese dates could be ambiguous. In this case, it can be assumed to be a Gregorian calendar date, but there's still a small chance it's a Chinese calendar date since the date is written in Chinese characters instead of Arabic numerals. But stated-as is not allowed on dates. Could you please help present this info in a correct way, or allow stated-as as a qualifier?--Roy17 (talk) 11:02, 8 July 2019 (UTC)

A similar issue occur with dates of the French Republican Calendar (Q181974) (the year starts in September, so if it's only down to a year...) Circeus (talk) 16:45, 8 July 2019 (UTC)
@Marsupium: thx!--Roy17 (talk) 08:45, 12 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --Roy17 (talk) 08:45, 12 July 2019 (UTC)

EntitySchema Sandbox?

Who can tell me that where to test this namespace? -- 07:37, 10 July 2019 (UTC)

EntitySchema:E123 is the sandbox entity schema page. --Lucas Werkmeister (WMDE) (talk) 11:29, 11 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Lucas Werkmeister (WMDE) (talk) 12:34, 12 July 2019 (UTC)

From a parliamentary group to independent or to other parliamentary group

Q54152815#P39 She was elected as a candidate of a political party. After she elected, she was part of this party's parliamentary group. But then she left the party and the group and she became independent. If someone reads the statement may think that she was independent all the parliamentary term. (Maybe someone realize what happen by reading Q54152815#P102?) Xaris333 (talk) 12:24, 3 July 2019 (UTC)

You could state position held (P39) twice, with end time 2018-02-02 for Citizens' Alliance (Q20773826) in the 1st statement, and start time 2018-02-02 for independent politician (Q327591) in the 2nd statement with another reference adopted from political party (P102). – 15:16, 3 July 2019 (UTC)

Q52773609#P39 He was elected as a candidate of a political party. After he elected, he was part of this party's parliamentary group. But then he left the party and the group and he became independent. One year later he joined the parliamentary group of other party. If someone reads the statement may think that he was in the second party all the parliamentary term. (Maybe someone realize what happen by reading Q52773609#P102?) Xaris333 (talk) 12:35, 3 July 2019 (UTC)

See, for instance Chuka Umunna (Q267648) - 4 statements for his P39 for his 57th Parliament membership, as he moved from one political grouping to the next. This, I suggest, is the model to be followed. --Tagishsimon (talk) 18:25, 3 July 2019 (UTC)
(By which I mean, the model to be followed on wikidata. As an IRL political strategy, possibly less so.) --Tagishsimon (talk) 18:26, 3 July 2019 (UTC)
Seems like these statement get more and more out of hand. We need to get this closer to a model with actual start and end dates for positions. We can't just mirror everyones database artifacts. --- Jura 18:33, 3 July 2019 (UTC)
Well, good luck with that. Presuming you want the electoral group, having multiple position statements seems unavoidable. This sort of organisation has not been found to cause issues with e.g. UK politician reports; and the handling of UK MPs & reports theron is in advance of any country on wikidata. "Seems ... out of hand" is an interesting but unempirical objection. Can you point to an actual downside - for instance, a report that's no possible because of the data structure? --Tagishsimon (talk) 20:55, 3 July 2019 (UTC)

@Jura1: I aggree. And the cases are not so many, we can easily find a model. Where can we open a discussion about it? Xaris333 (talk) 20:48, 3 July 2019 (UTC)

"Easily". I've ordered popcorn. On you go, Xaris333: easily find a better solution which represents the changing electoral group membership of an individual as part of their holding of a position. Alternately, per Jura, specify what exactly you think the practical problem with such an approach is, ideally with examples of the failure modes it causes. Those of us who have gone here before await the insight which evaded us. You could try discussing it at Wikidata talk:WikiProject every politician. I know Jura a) doesn't like the every bit of every politician; doesn't like subclasses of MP for British MPs; and now does not like multiple P39s reflecting membership of different electoral groups. So all this will probably cause them lots of pain; but there you go. --Tagishsimon (talk) 21:03, 3 July 2019 (UTC)
That's so disappointed... Xaris333 (talk) 21:53, 3 July 2019 (UTC)
Easily. --Tagishsimon (talk) 22:04, 3 July 2019 (UTC)
@Tagishsimon: These are just baseless claims. Please refrain from such accusations. Apparently some users don't like to hear that they are adding unsourced claims and that others ask them for references ..
Maybe it's time to learn how sourcing of claims works and not depreciate statements that don't appear suitable to you [23]. --- Jura 14:40, 4 July 2019 (UTC)
Here where I live, electoral groups are not so important as they maybe are in other nations. Technically you can be elected for a political party, without any membership in that party. On national level, you can technically leave a electoral group and become independent politician (Q327591), but you cannot technically not join another group, even if you cooperate with them. And you can definitely not form a new group midterm. Another way to group MP's are according to their constituencies. On national level, they often meet in such groups. They cooperate in regional matters and it is not unusual that an MP challenge the whip on matters that are consideered important on regional level. IP (Hej!) 06:44, 7 July 2019 (UTC)

Balustrading To Area Of Municipal Buildings Municipal Buildings (Q17551052)

Where did this title come from? Nowhere I know, especially when the linked Historic England page says totally different. I'm going to remove it from the Commons category. Rodhullandemu (talk) 17:48, 7 July 2019 (UTC)

Wikidata items for Women in Red's events and redlists

Women in Red is having a conversation regarding whether there would be value in creating Wikidata items for Women in Red's events, most of which are virtual but some are in-person. There are 120+ events to date; each has a "meetup" page on en-wp as a subpage to Women in Red. (Example: There are additional events (virtual & in-person) in the 23 other language communities of Women in Red.

Women in Red is also discussing whether there would be value in creating Wikidata items for our 400+ "redlists". These are subpages of Women in Red which include redlinked notable women. (Example: Some of these redlists are "crowd-sourced", others are Listeria-generated, others are based on biographical dictionaries, and so forth. Besides the en-wp redlists, there are those which have been created by the 23 other language versions of Women in Red.

If we were to go forward with creating these Wikidata items, we would include "on focus list of Wikimedia project" (P5008) : "Wikipedia:WikiProject Women in Red" (Q23875215) on the Wikidata items. I don't know if other groups/communities doing this. What are your thoughts about us doing it? --Rosiestep (talk) 15:25, 7 July 2019 (UTC)

I think that is what the property is created for. The examples at on focus list of Wikimedia project (P5008) are exactly the same as you want to use it. --GPSLeo (talk) 16:41, 7 July 2019 (UTC)
@Rosiestep: We have 43 items that are <instance of> edit-a-thon (Q16022392) going back to 2015, so that seems to establish a precedent that events meet notability requirements. I don't see that any of them have associated lists, though - how are you thinking of structuring those? - PKM (talk) 21:29, 7 July 2019 (UTC)
@PKM:, I thought Art+Feminism and/or Black Lunch Table had associated lists, so maybe I should check with them? --Rosiestep (talk) 21:56, 7 July 2019 (UTC)
@Rosiestep: That sounds good; I don't see any linked lists on their Wikidata items. - PKM (talk) 22:01, 7 July 2019 (UTC)

Merging request

Q65122684 and Q54604086 are one and the same disambiguation page in different languages. Francesco 13 (talk) 12:23, 13 July 2019 (UTC)

✓ Done I did done it Moebeus (talk) 12:39, 13 July 2019 (UTC)

I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Moebeus (talk) 12:39, 13 July 2019 (UTC)

Part of goverment

Q2792539#P39 -> Cypriot Minister of the Interior. Which qualifier can I use to show that was a minister in Demetris Christofias Government 2008 (Q1719844)? Xaris333 (talk) 06:41, 8 July 2019 (UTC)

Possibly cabinet (P5054), or member of (P463), both of which are listed as valid qualifiers for position held (P39)? Looking at the count of uses, we see:
fwiw, EveryPolitician seems to make do with a comparison of ministry dates (e.g. for Second May ministry (Q30178117) and position dates to provide e.g. Wikidata:WikiProject every politician/United Kingdom/Cabinet/Q30178117 so there's a sense in which membership qualifiers within a position statement are redundant. Equally, it does seem to me desirable that we have a standard for this, which I don't see in Wikidata:WikiProject_every_politician/Political_data_model#Cabinet_memberships ... it does seem like a natural qualifier. One to consider, @Oravrattas:? --Tagishsimon (talk) 07:48, 8 July 2019 (UTC)


Can we remove the commons category from Q63245258 as the page is not active and also the candidate info as he is not old enough. 2600:6C56:6F08:1CF:0:4945:4CD6:27CB 09:18, 8 July 2019 (UTC)

As the cat no longer exists, consider it done. --Tagishsimon (talk) 09:47, 8 July 2019 (UTC)
Is the person actually notable?--Ymblanter (talk) 18:21, 8 July 2019 (UTC)
Ymblanter - I didn’t think so but I was overruled during the deletion process. Datamaster1 (talk) 18:41, 8 July 2019 (UTC)

Wikidata weekly summary #372

Request from RfD

Moved from WD:RfD

Hi, there is an issue with the two pages: School of Management Fribourg (Q30594328) is the most complete section

It is therefore not possible to link the two translations in EN et DE:

Language Label Description Also known as English School of Management Fribourg No description defined French Haute école de gestion Fribourg No description defined HEG-FR German Hochschule für Wirtschaft Freiburg No description defined HSW-FR -- 10:55, 8 July 2019 (UTC)


Paulsboro Refinery (Q35287721) is listed in Geonames as an airport instead of an oil refinery, some refinerys have heliports, but I do not see a corresponding FAA airport code, for example see Bayway Refinery (Q3014399) which has Bayway Refinery heliport (Q61675718). Do you think it is an error at Geonames? --RAN (talk) 22:32, 8 July 2019 (UTC)

Yes and no. The AIA Directory of Heliports (admittedly for 1981) lists a heliport at Paulsboro Refinery, on page 147. Equally, I cannot find a heliport by eye using google satellite view; so perhaps its now defunct? What is odd, for me, at least based on a search in geonames for 'Paulsboro Refinery', is that there is not a record for the refinery coded as 'Oil refinery', a label that does feature in their ontology. That aside, presuming that there is no longer a heliport there might explain the lack of FAA code, presuming that US heliports must have such a code. --Tagishsimon (talk) 23:07, 8 July 2019 (UTC)
If there is, it wasn't licensed as of 2016. Circeus (talk) 23:11, 8 July 2019 (UTC)

Errors caused by ambassadors

László Trócsányi (Q1324589)'s ambassador of Hungary to Belgium (Q63247361) statement causes error (and other ambassador Qs too). Could somebody help me putting a fix to the correct place? Thanks in advance! Bencemac (talk) 11:53, 6 July 2019 (UTC)

I would expect that to work, since ambassador of Hungary to Belgium (Q63247361) is a <subclass of> ambassador (Q121998) and ambassador (Q121998) is an <instance of> position (Q4164871), which should satisfy the constraint. Can any constraint experts weight in? - PKM (talk) 20:52, 6 July 2019 (UTC)
Unless I am mistaken, that's not how the constraint actually work. Only subclass can "look up the tree". Instance must be marked directly on the linked item (i.e. unless ambassador of Hungary to Belgium (Q63247361) is an instance of the requisite items, it won't satisfy the constraint). The constraint probably fails because (I would assume) at not point does the tree put ambassador of Hungary to Belgium (Q63247361) as a subclass of position (Q4164871) itself. Circeus (talk) 22:44, 6 July 2019 (UTC)
Then it's a stupid constraint. And the solution is not, Jura, this since we have perhaps 195^2 of these positions, all of which are a subclass of ambassador (Q121998) which is a position (Q4164871). The solution is to make a constraint which works given the pattern ambassador of Hungary to Belgium (Q63247361) is a <subclass of> ambassador (Q121998) and ambassador (Q121998) is an <instance of> position (Q4164871). --Tagishsimon (talk) 23:29, 6 July 2019 (UTC)
The WikiProject property constraints might help you with this issue. I am going to ping them.

Lucas Werkmeister (WMDE)
Jarekt - mostly interested in properties related to Commons
John Samuel
Yair rand
Jon Harald Søby
Was a bee
Peter F. Patel-Schneider
ZI Jony
Pictogram voting comment.svg Notified participants of WikiProject property constraints

Regards, Ivanhercaz (Talk) 00:07, 7 July 2019 (UTC)
  • It's actually fixed now. To avoid any doubt, could you add reference to the position statement on Q63247361? We noticed that some people add fictitious positions. (I don't think this is true here). ----- Jura 00:53, 7 July 2019 (UTC)
@Jura1: I added official website (P856), it verifies the item. I hope it is suitable. Bencemac (talk) 06:57, 9 July 2019 (UTC)

Thanks for the help, but I am afraid of that ambassador of Hungary to France (Q62558339) (and probably others) still has the same problem. Any idea? Bencemac (talk) 07:01, 9 July 2019 (UTC)

Dupes from CBDB imports

I found some dupes from User:Fantasticfears's imports from CBDB. I believe there are more, but checking one by one manually is daunting. Is there some script or tool that might help?--Roy17 (talk) 18:42, 8 July 2019 (UTC)

This, maybe - items sharing the same CBDB ID. --Tagishsimon (talk) 18:49, 8 July 2019 (UTC)
Awesome! They should be merged. However, most of the ones I found do not have CBDB ID on one of them. I matched them by matching Chinese name and their 籍貫 and era. (It's almost impossible to have two persons having the same name from the same county in the same era.)--Roy17 (talk) 22:46, 8 July 2019 (UTC)
All entries in the query is fixed.--GZWDer (talk) 10:19, 9 July 2019 (UTC)

What can I do with a wikidata item that is an instance of a wikimedia category?

What useful properties should be added to a wikidata item that is an instance of Wikimedia category (Q4167836) and what properties should it be the object of?

In particular, en:Category:Uniting Church in Australia presidents corresponds to Category:Uniting Church in Australia presidents (Q25011267) and currently has 15 people in it, who each held the role for approximately 3 years. Is there a standard way to associate the person with the category in wikidata? <person> subject has role (P2868) Category:Uniting Church in Australia presidents (Q25011267) doesn't feel quite right. Should I somehow create an item "Uniting Church in Australia president" as an instance of role (Q4897819) and if so, how would that be associated with the category item? I could then give each person subject has role (P2868) with start and finish dates, successor, predecessor as qualifiers. Thank you. --ScottDavis (talk) 01:01, 9 July 2019 (UTC)

Category items shouldn't generally be used in statements. When there are several people in Wikidata who had the same job, creating an item like "Uniting Church in Australia president" is probably the thing to do, like President of Evangelical Church of the Lutheran Confession in Brazil (Q57628868). Ghouston (talk) 01:33, 9 July 2019 (UTC)
The other option, which is often used, is on items like Helen Murray Free (Q7476), where position is president (Q1255921) with a qualifier for the organization. Ghouston (talk) 01:36, 9 July 2019 (UTC)
Thank you. I shall attempt to follow the style of the first example as that reflects what I was trying to do but didn't have the confidence in my familiarity with WikiData. --ScottDavis (talk) 10:15, 9 July 2019 (UTC)
Update I have created President of the Uniting Church in Australia (Q65249159), added 15 officeholder (P1308) entries with dates and the matching position held (P39) on each of the 15 items. I wish I knew if I could have used tools to create the inverse statements and links instead of having typed everything including qualifiers twice. --ScottDavis (talk) 13:43, 9 July 2019 (UTC)
I'm not sure why they can be listed on both either. I'd tend to put the statements on the individuals, not on officeholder (P1308). Wikidata has a lot of such duplication though. Ghouston (talk) 01:40, 10 July 2019 (UTC)

Qualifying time of criminal convictions

Using qualifier point in time (P585) on convicted of (P1399) is a little ambiguous to me. I would assume it is the time of conviction and not the time at which the crime was committed? In which case, how would the latter be best modelled? --SilentSpike (talk) 23:06, 9 July 2019 (UTC)

Yes, since the property is about the conviction, any qualifier should apply to that. Where there's no specific property for an event, in this case the perpetration of a crime, key event (Q2245405) can be used. Ghouston (talk) 02:07, 10 July 2019 (UTC)
Although, I suppose convicted of (P1399) was created to link a person with a crime, so it seems a bit unfortunate that you'd need to add another statement about the crime to record its date (assuming it's not a crime with a separate item, like Murder of Julia Martha Thomas (Q2613392), where the date can be added). Ghouston (talk) 02:11, 10 July 2019 (UTC)

Guide to adding references

I figured it might be helpful to add references for information I've already sourced in articles I've written. Is there a guide for me? Figuring out how to add just the URL is easy enough but that feels incredibly incomplete. Best, Barkeep49 (talk) 18:50, 10 July 2019 (UTC)

@Barkeep49: Help:Sources should have what you need. - PKM (talk) 20:19, 10 July 2019 (UTC)
Thanks @PKM:. Having read that I have a few follow-up questions. I edit lots of books on EN Wiki so those and authors was where I was starting. Am I correct that a first edition should have a seperate item from the book itself? Is this in all cases or just in cases where we've referenced that edition somewhere? Also I'm still generally confused about best practices with references. Should I be using stated in and reference URL when I'm using a specific webpage as a cite or just reference URL? How in general does Wikidata deal with linkrot? Thanks and best, Barkeep49 (talk) 20:20, 10 July 2019 (UTC)
@Barkeep49: The best practice is to add both a "work" and an "edition" for a published book used as reference (and the reference should always point to the edition). Examples: Encyclopedia of the Middle Ages (Q42962936) (work), Encyclopedia of the Middle Ages (Q42963002) (edition), Ottonian architecture (Q1588393) (item with citation). Note in this case the book is online and sections have their own URLs, so I have included that info.
If a web site or database has a Wikidata item, used "stated in". If not, just use "reference URL". See Catherine Donovan (Q51077492), reference for "date of death". I'm sure there are other ways to do this.
To avoid link rot, you can add archive URL (P1065) and archive date (P2960), just like you would on Wikipedia. See Bernard Quaritch Ltd (Q65034936) property "official name" for an example. There is a bot that will (eventually?) do this for you. I forget what the status/schedule is. - PKM (talk) 20:44, 10 July 2019 (UTC)

To reuse WP references, I thought Mvolz was working on --- Jura 12:01, 11 July 2019 (UTC)


I was disentangling two concepts that used to be in one item (Vaishnavi (Q7908873) and Vaishnavi (Q65534195)) and may have made a mess. How do I move all of the multilingual descriptions for female given name from the Devi item to the given name item? I can only easily edit en, fr, and es (I'm guessing that's because of my language boxes on my userpage?). Calliopejen1 (talk) 17:03, 12 July 2019 (UTC)

@Calliopejen1: one option might be to delete all descriptions on the old item using User:Ash Crow/dataDrainer.js and then adding fresh ones on the new item using Wikidata:Namescript. (Afterwards, you’d have to manually restore any descriptions on the old item that shouldn’t have been deleted – I’m assuming that’s only a few if any, otherwise this probably isn’t the best approach.) --Lucas Werkmeister (talk) 17:33, 12 July 2019 (UTC)
@Calliopejen1:User:Ash Crow/dataDrainer.js seems to require rollbacker permission, and I'm not sure you've been active in WD long enough to have that ...
You might also restore the original item to the state before the split and then modify the newer item to be the "other half" of the split. - PKM (talk) 18:18, 12 July 2019 (UTC)
Nope, I don't have rollbacker yet. Yeah, it might be easier just to swap them. I thought it would be easier in this configuration because all the WP links could stay where they were... But I think changing the names around is way worse. Calliopejen1 (talk) 18:57, 12 July 2019 (UTC)
Okay, all sorted now! Calliopejen1 (talk) 19:03, 12 July 2019 (UTC)

Sequence of children/siblings

Which property/qualifier to use to present this information?--Roy17 (talk) 18:08, 8 July 2019 (UTC)

series ordinal (P1545) is used for pretty much the same thing for author order on works where author ordering is relevant see, say Speed change discrimination for motion in depth using constant world and retinal speeds (Q64082142) (example selected literally at random). Circeus (talk) 22:29, 8 July 2019 (UTC)
How to present one info in particular though? That is, John F. Kennedy (Q9696) is the second among his siblings. It's easy on his father's page, assigning 1 to n to them. It's also easy for his siblings on JFK's page, assigning 1 to his older brother and 3 to n to his younger siblings, but how to show he himself is no.2?--Roy17 (talk) 22:46, 8 July 2019 (UTC)
The data being present on related items is sufficient. --Yair rand (talk) 23:58, 8 July 2019 (UTC)
I am both my father and my mothers firstborn. My sister is also her fathers firstborn, but her mothers second. My youngest sister is her fathers second and her mothers third. So add that info to the father/mother instead. IP (Hej!) 05:54, 9 July 2019 (UTC)
Sounds like data that is hard to reference (1st, 2nd, 3rd). --- Jura 08:48, 9 July 2019 (UTC)
@Yair rand: could you please help input the following information? (Italics denote uncertain info.)
Tan Zhonglin (Q11632273) had 5 sons:
Tan Baozhen 譚寶箴 (no item yet), b. ?, 1st son
Tan Yankai (Q708781), b. 1880, 3rd son
Tan Enkai (Q45694336), b. ?, 4th son
Tan Zekai (Q45694204), b. 1889, 5th son
Another case whereby only the order is known: Yuan Shao (Q337013) had 1st son Yuan Tan (Q145671), 2nd son Yuan Xi (Q953232) and 3rd son Yuan Shang (Q1276938).--Roy17 (talk) 10:20, 9 July 2019 (UTC)
As there is no property for son, I don't see how sons can be numerated. child (P40) is for children of all gender. --Dipsacus fullonum (talk) 22:18, 9 July 2019 (UTC)
Let me summarise the problem.
Suppose there are three siblings A, B and C, each notable for wikidata. They meet the following criteria:
  1. They have no (biological, step-, adoptive or whatsoever) father or mother.
  2. Their years of birth are unknown.
  3. We know A>B>C.
Info #3 is surely useful, but how does wikidata store it? Is there a qualifier to say older/younger than?
(It's not common to find such people meeting both #1 and #2, but it's common enough to find ones that meet #1 because their parents are not notable enough for a wikidata item, unless you choose to create it just for this fact. #2 is also common for historical figures, even though I havent found a case in which their parents were not documented yet.)--Roy17 (talk) 14:18, 10 July 2019 (UTC)
Can these be used:
--Dipsacus fullonum (talk) 19:56, 10 July 2019 (UTC)
@Roy17: "They have no father or mother" isn't really a thing that can happen. Not only is it true that all humans have mothers and fathers, but siblinghood additionally implies not only that they have parents, but that the siblings share the same parents. Even if we don't know anything about the parents except that they're the parents of multiple particular people, the items can be created to fulfill the structural need, per WD:N. --Yair rand (talk) 03:55, 11 July 2019 (UTC)
Exactly, take wife of Saint Peter (Q22340337) as an example. The benefit of this item is that it connects two other items, St. Peter and his mother in law. IP (Hej!) 07:04, 11 July 2019 (UTC)
User:Dipsacus fullonum's answer is the best so far. Thx!--Roy17 (talk) 08:45, 12 July 2019 (UTC)
  • "They have no [..] father or mother.": clearly this isn't possible for Q5. In the sample by Dipsacus, "sibling" should be used instead of P1038. --- Jura 12:03, 13 July 2019 (UTC)

suggestededit-add 1.0

What is this functionality? special:diff/976058660 added wrong information. I dont know if the wrong info was suggested by something and then chosen by the user, or if the user was suggested to edit the description and s/he came up with the wrong info. If the former is true, this functionality is probably suggesting too much wrong info.--Roy17 (talk) 10:20, 9 July 2019 (UTC)

@Roy17: new here? You should leave people a message. It's kind of rude to start a topic here behind someone's back. You didn't even welcome the user! I'll leave a message at User talk:Leofreitasa. Multichill (talk) 21:45, 9 July 2019 (UTC)
@Multichill: I find it ruder to misread others' messages. I made it clear this functionality could be at fault. The fault would be the developer's rather than any particular user's.--Roy17 (talk) 21:52, 9 July 2019 (UTC)
  • I don't recall the post that seems to have announced it either. Odd that the tool expects that one would revert a partially correct edit rather than fix it. --- Jura 17:08, 10 July 2019 (UTC)

Thank you all for your comments and questions about Suggested Edits. I have added a FAQ page that gives more background on the feature, how it works, and how we have announced it in the past. Please do let me know if you've any more questions. CGauthier (WMF) (talk) 21:29, 10 July 2019 (UTC)

  • Could we try to improve the readability of Wikidata in mobile first? --- Jura 11:59, 13 July 2019 (UTC)


... is probably vandal. That account removes descriptions or property titles for many articles. I put messages on the talk page and reverted some edits. Block it if needed. --Obsuser (talk) 15:09, 10 July 2019 (UTC)

Stop. You are restoring incorrect labels and also removed fixes for Indonsesian language at Instagram username (P2003). You should be sonsidered as vandal in this case. Eurohunter (talk) 16:06, 10 July 2019 (UTC)
Link ratcher this User talk:Obsuser#IMDb. Eurohunter (talk) 16:08, 10 July 2019 (UTC)
One more time. You called me vandal? Look at my edits first. Eurohunter (talk) 16:16, 10 July 2019 (UTC)
@Obsuser: When you are going to restore fixed label for Indonesian at Instagram username (P2003)? Do you even checked edit before undo? You cancelled it two times without reflection. Eurohunter (talk) 16:24, 10 July 2019 (UTC)

You removed a dozen of correct designations. You restore what changed meanwhile. I reverted to last stable version. --Obsuser (talk) 16:27, 10 July 2019 (UTC)

@Obsuser: You removed already fixed Indonesian and restorred incorrect labels which I removed before. You made a mess and you will fix it now or I will restore everything again (everything should be restored anyway but let's wait with it before someone will explain you or you will understand). Eurohunter (talk) 16:31, 10 July 2019 (UTC)
No. You removed correct marks for most of the languages and meanwhile someone changes only few of them. You fix if it is needed. You can restore for a short span of time but someone else will revert because removing hundreds of designations for already properly translated properties for many languages is not acceptable. --Obsuser (talk) 16:34, 10 July 2019 (UTC)
Can you look what you have done with Indonesian? It's ridiculous because you can't proced like that (cancalling "suspicious" edits for you and everything else absolutelly correct). Eurohunter (talk) 16:38, 10 July 2019 (UTC)
@Obsuser: You see it? If you are not going to fix it now it's just unfair. If you want to cancell something do it carefully. Eurohunter (talk) 16:49, 10 July 2019 (UTC)

First at all. Wikidata is not yours and if some edits were incorrect for you, you need to prove it and achieve consensus and then you can remove them, not right now you have already removed them and now started discussion. Eurohunter (talk) 16:48, 10 July 2019 (UTC)

Can someone either point at actual diffs so that there is half a chance of people following this, or take it somewhere less public than Project chat if there is no need for anyone else to understand? - Jmabel (talk) 22:06, 10 July 2019 (UTC)

@Jmabel: I have removed incorrect labels and Obsuser not just only restored incorrect labels but he also removed Indonesian label which was fixed meanwhile. I told him probably 10-12 times he removed corrected label and he don't care about it. Eurohunter (talk) 15:48, 11 July 2019 (UTC)

Active admin?

Can someone restore my edits at Instagram username (P2003), SoundCloud ID (P3040) and IMDb ID (P345)? @Obsuser: removed my edits and restored incorrect labels (incorrrect labels in few languages + unfounded removal of label in Indonesian for Instagram ID). It's not Obsuser's Wikidata so I'm not agre with that and my edits should be restored as it was till consensus in eventual discussion. Eurohunter (talk) 19:56, 10 July 2019 (UTC)

You can see here that almost all or all of the designations for languages were removed by Eurohunter. That is not the way to edit Wikidata. "ID" is not neccessary if property is of a primary character, so designations added by other users are correct and applicable so should not be removed per se. --Obsuser (talk) 23:42, 10 July 2019 (UTC)
@Obsuser: is still vandalizing labels. Eurohunter (talk) 15:32, 11 July 2019 (UTC)
Clearly the Indonesian labels should be kept. As for the many (possibly better than nothing but not quite on the mark) labels in other languages I have no opinion. - Jmabel (talk) 16:04, 11 July 2019 (UTC)
I think it is reasonable to keep something what is added by other users. That designation that I have changed is hilarious and thus corrected into previous and standard form. --Obsuser (talk) 16:07, 11 July 2019 (UTC)
@Obsuser: You are not allowed to make mess bigger. Eurohunter (talk) 16:58, 11 July 2019 (UTC)
@Obsuser: Are you describing the Indonesian text as "hilarious"? If so, can you clarify what is wrong with it, rather than providing this nearly vacuous remark? - Jmabel (talk) 17:21, 11 July 2019 (UTC)
That applies for you. Not Indonesian but Serbian. It starts with noncapital and is long, incorrect and unusual for such marks (see other examples, "Фејсбук", "Твитер", "Инстаграм", one word). --Obsuser (talk) 17:49, 11 July 2019 (UTC)
Problem is that those removed were correct. --Obsuser (talk) 17:50, 11 July 2019 (UTC)
@Obsuser: You can't say they were corrrect bcause they had no logic and 4 of them was already fixed (naturaly were incorrect if were fixed by me/others). One more time. You need to notice that you have removed Indonesian and 3 other labels which were fixed by me/other users and you are obligated to restore it. It's vandalism and it just can't be left without action. About Serbian. Would you like to tell me that Serbian language has no logic? If we continue with your logic we can say you "have Wikidata"? Not "Wikidata account"? It just doesn't makes sense. You are talking about "slang"/colloquialism which some people use in every day talks but it doesn't mean it's correct, like some people makes many other errors which are ratcher not acceptable in Wikidata. Eurohunter (talk) 18:18, 11 July 2019 (UTC)
I am afraid for me this edit by Obsuser does not make sense.--Ymblanter (talk) 18:31, 11 July 2019 (UTC)
It again applies for you. We are equal editors. You need consensus before making such radical edits without a valid explanation. Fix them again, that edit makes sense Ymblanter because previous marks were restored and are valid without "ID" suffix or something else. It has logic, mark for Instagram ID is "Инстаграм", not long explanation that that is user ID for profile on Instagram because it is logical that it is for that (only) thing associated with persons having accounts there. --Obsuser (talk) 18:36, 11 July 2019 (UTC)
I do not see how "Instagram" is a valid alias for "Instagram username".--Ymblanter (talk) 19:01, 11 July 2019 (UTC)
  • As there is only one Instagram property, I think it is ;) --- Jura 11:58, 13 July 2019 (UTC)

Merging two Signage

Found these two pages, signage (Q24841283) and signage (Q1211272). I am not experienced enough on Wikidata to merge these but I think it should be done. Could someone help, please. --W.carter (talk) 08:21, 10 July 2019 (UTC)

signage (Q24841283) and signage (Q1211272), I think you are right. Ghouston (talk) 09:17, 10 July 2019 (UTC)
I speak some of the languages in both groups so I know they are about the same sorts of items. I was also searching for where the French and Spanish versions had gone (I'm pretty sure there are articles since it is a universal subject) but haven't found them so far. --W.carter (talk) 10:36, 10 July 2019 (UTC)
There is French fr:Signalisation routière, but only for road signs, perhaps corresponding to traffic sign (Q17480810). I also discovered a possibly duplicate category item Q22910716. Ghouston (talk) 10:46, 10 July 2019 (UTC)
The Q22910716 doesn't look like a duplicate to the two above to me, since it is a category covering a bit more than just signs. --W.carter (talk) 16:54, 10 July 2019 (UTC)
I found the French article, fr:Enseigne (signal) thanks to your clue about the road signs. It was misplaced at guild sign (Q229917) and I have moved it to the right place. I also found the Spanish es:Letrero (through the French 'neon sign') all by itself at signage (Q5973946). I've moved it and will propose the empty item page for deletion. --W.carter (talk) 17:03, 10 July 2019 (UTC)
I think we should wait for a bot to fix the links to signage (Q5973946), and then merge it with signage (Q1211272) (in case double redirects would cause problems.) Ghouston (talk) 22:32, 12 July 2019 (UTC)
No problem, I'm happy to wait. I've put in a request for a move/merge/deletion from someone who knows how to do this properly. --W.carter (talk) 19:39, 13 July 2019 (UTC)

Another merging problem of items on the same subject

Albanian sworn virgins (Q16002379) and sworn virgin (Q21027639) should be merged but there are some Wikipedia's that have two articles (some of them are a redirect). How should this be resolved? Keren - WMIL (talk) 08:06, 14 July 2019 (UTC)

  • If it's really the same: delete all redirects in sitelinks, move remaining ones to one item, move the statements to the same, mark the item with the remaining cawiki sitelink with P31=Q17362920, possibly also suggest to cawiki to merge the two.
The last two steps can be done with User:Matěj_Suchánek/markasduplicate.js.
A bot will also merge the items once cawiki merges them. --- Jura 08:14, 14 July 2019 (UTC)

President of a football team

Hello. Which properties should I use to show that:

1) Evgenios Hamboullas (Q65554755) held the position of President of Anorthosis Famagusta FC (Q65557188)

2) President of Anorthosis Famagusta FC (Q65557188) has as officeholder Evgenios Hamboullas (Q65554755)

Xaris333 (talk) 14:15, 14 July 2019 (UTC)

1 - Evgenios Hamboullas (Q65554755) - position held (P39) - President of Anorthosis Famagusta FC (Q65557188).
2 - President of Anorthosis Famagusta FC (Q65557188) - officeholder (P1308) - Evgenios Hamboullas (Q65554755) --Tagishsimon (talk) 14:33, 14 July 2019 (UTC)

Yes, I have already tried these two but I get constraint. Ok, I have found the problem. Thanks. Xaris333 (talk) 14:39, 14 July 2019 (UTC)

Linking two sets of articles on the same subjects, but where different language wikipedias use differing titles?

In 2003, Brian Wells, a pizza delivery man, was killed in a bizarre bank robbery where he was either part of a conspiracy or an unwitting victim, in which he robbed the bank while wearing an explosive device locked around his neck.

There are seven Wikipediae that carry articles on him. In four of the articles, the subject's name is the name of the wikipedia article:

On the English Wikipediae (both standard and Simple English), the article is named for the bank robbery event due to the ONEEVENT policy; one with a redirect from the subject's name:

The Japanese Wikipedia's title is based on a description:

Despite these articles all covering the same subject matter and being alternate-language versions of one another apart from the article naming, wikidata links them into two separate groups for purposes of the "Languages" sidebar: one (Q56461070) with the German, Spanish, French and Indonesian articles; and one (Q523883) with the English, Simple English, and Japanese articles. What's the best way to gather them together? They're really all on the same topic, and only the article-naming policies pull them into separate groups, not content. TJRC (talk) 22:05, 12 July 2019 (UTC)

I'd say that on Wikidata, Brian Wells (Q56461070) is a useful item, but death of Brian Wells (Q523883) is unnecessary since doesn't contain any additional information. I.e., it's only created supposedly for the benefit of Wikipedia, but in fact is only damaging the interwiki links. I think both Wikidata and Wikipedia would be better off if the two Wikidata items were merged. Ghouston (talk) 22:27, 12 July 2019 (UTC)
I think it's fine as is. As per the notability guidelines it is notable if "It contains at least one valid sitelink to a page on Wikipedia". Unless wikipedia changes their pages to be about the person then an item for the event seems necessary to me --SilentSpike (talk) 22:49, 12 July 2019 (UTC)
It makes sense to me to have an entity for the person (Q56461070) and an entity for the event (Q523883). And then the mapping from the differently-titled Wikipedia articles is natural and, not coincidentally, matches what we have now. (Clearly it makes sense to interlink the two entities using relevant property claims, too. participant in (P1344) and participant (P710) don't strike me as ideal for this purpose, but they're pretty close.) —Scs (talk) 23:02, 12 July 2019 (UTC)
Sharing item is not the only way to have interwiki between articles. If the wikipolicy allows it, you can use Template:Interwiki extra (Q21286810) to get interwiki from more than one item. IP (Hej!) 06:06, 13 July 2019 (UTC)
There should indeed be 2 items, but what is somewhat problematic here is that Q523883 was changed in 2018 from one into the other. @Bovlb: --- Jura 11:31, 13 July 2019 (UTC)
Thanks for the ping. In 2018, I found this as an item that seemed to combine two concepts, and I split it into two, creating a new item for the human. Not quite sure in hindsight why I chose to split it that way rather than the other. I was probably favouring the English sitelink. I agree that Template:Interwiki extra (Q21286810) seems to be the way to go for these cases. Bovlb (talk) 22:24, 13 July 2019 (UTC)
They're not really two subjects, though, are they? They all cover the same ground, and basically only the way the titles and first-sentences are written suggest that they're on different subjects. But they're not. All seven articles are on the same subject. TJRC (talk) 04:42, 14 July 2019 (UTC)
I guess we could periodically re-read all sitelinked articles, check if they are mistitled or its lead doesn't match the remaining article, should be renamed or moved to different items. But wouldn't that be interfering with the editorial choices of the various wikipedias? --- Jura 08:19, 14 July 2019 (UTC)
@TJRC: A death and a person are ontologically different things. I agree that writing an article about one or the other will tend to cover the same ground, and the the precise ontological type of Wikipedia articles is not always clear. @Jura1: I see it as supporting the editorial choices of the Wikipedias. The ONEEVENT policy pushes a specific editorial choice on the English Wikipedia. It looks like Template:Interwiki extra (Q21286810) allows us to avoid losing sitelinks in this case. Bovlb (talk) 16:10, 14 July 2019 (UTC)
Yes, they are different, but generally the event item wouldn't be required in Wikidata, since there are designated properties on the person's item to hold all the same information, so the event item is redundant. I doubt it would be much appreciated if somebody created an item "Birth of X" for every person X in Wikidata, to redundantly hold their birth date and location. Ghouston (talk) 21:17, 14 July 2019 (UTC)
The reference items for this type of case are murder of Kitty Genovese (Q18341392) and Kitty Genovese (Q238128). Multichill (talk) 11:24, 14 July 2019 (UTC)
I added Template:Interwiki extra (Q21286810) to w:en:Death of Brian Wells to bring in the additional Wikidata links automatically. Simple English Wikipedia does not have this template, so I could not add it there. I see that there were multiple cases in September 2018 where I split an item into a person and a death event. (I believe was trying to prepare a QuickStatements run, but ultimately decided that the task could not easily be automated.) I reviewed them and found no other cases that would benefit from additional links. Bovlb (talk) 19:11, 14 July 2019 (UTC)

S/2014 (130) 2

What to do with discoveries that showed up to be nothing? [24]

The P31:minor-planet moon (Q657829) statement should be changed to inferior. But do we need to add something instead? IP (Hej!) 14:03, 14 July 2019 (UTC)

  • By "inferior" do you mean "deprecated" or something else? - Jmabel (talk) 16:21, 14 July 2019 (UTC)
    • That is maybe the word, yes. It was thought to be a satellite to 130 Elektra, but it turned out to not exist. The statement should stay, but as deprecated. But can we have items with only a deprecated P31? Do we have to complementary add something like P31:"dust on the camera lens/science mistake/s**t happens"? IP (Hej!) 19:05, 14 July 2019 (UTC)
@Sextvå.tvånoll.ettsjunoll.sjufyra: note also debunked (Q59237924)

Time to review Wikidata:Alternate accounts?

This is not yet a formal proposal.

Wikidata's sock puppetery policy seems to be the strictest in Wikimedia projects. It requires public declaration of all alternative account used, and thus disallows clean starts, accounts for privacy purposes (previous discussion), etc.

However, there're many ways to edit Wikidata without going to the Wikidata website. Moving or deleting local pages automatically updates Wikidata; Wikidata description may be edited by apps; there're various 3rd party tools; and also there's the Wikidata Bridge project. In some time there's no indication that it is Wikidata that you are editing. Therefore, many people in people in local projects does not know the purpose of Wikidata at all, let alone being aware of the sock puppetery policy of Wikidata.

If some person decides to have a clean start, then it's allowed in most Wikimedia projects, but not allowed in Wikidata. It's not easy to avoid editing Wikidata while edit other projects normally (especially when most of the data in infoboxes comes from Wikidata). Admininistrators in Wikidata may choose not to block them per IAR or UCS, but this does not mean the policy don't needs to be amended.

Therefore, this is my idea:

  1. Users are encourages to, but not required to, declare alternative accounts. However, declaration is required if there's a strong accusation about misuse of multiple accounts. (How to enforce this?)
  2. Accounts that are not used in a productive way blatantly (including blatant violation of policy) may be blocked without warning.
  3. In other case, a discussion is required before taking any action.


  1. All alternative accounts (e.g. old accounts that is abandoned after a clean start) must be disclosed, either publicly or to a designated mail list that some trusted users will have access.

--GZWDer (talk) 10:37, 11 July 2019 (UTC)

  • Sound like a good idea to change this! --Marsupium (talk) 20:00, 11 July 2019 (UTC)
  • GZWDer (talkcontribslogs)'s interpretation of the policy is incorrect. Clean-start accounts are not expressly forbidden and would be allowed as long as the previous account is declared; that latter requireent is also the requirement on the English Wikipedia for clean start, and is necessary to maintain accountability. Nothing is wrong with the current policy, other than perhaps clarifying this point. Until we establish an arbitration committee or other community body that can be empowered to deal with private matters, we cannot allow undeclared alternate accounts for accountability reasons.--Jasper Deng (talk) 20:27, 11 July 2019 (UTC)
    • The enwiki policy does not require user to notify anyone of their clean start, so it is possible not to reveal the connection. For accountability issues, this is why I propose to set up a seperate (private) mail list, though who can access to the list may be discussed. (P.S. Chinese Wikipedia has neither ArbCom nor local CheckUser currently, but there's an admin mail list for discussing private matter.)--GZWDer (talk) 20:55, 11 July 2019 (UTC)
      • It would be odd and awkward to setup a private mailing list specifically for this purpose. We don't even have local CheckUsers, otherwise that would be the natural option.--Jasper Deng (talk) 21:07, 11 July 2019 (UTC)
        • It may be less odd if it is dubbed Wikidata-admin. (actually in Chinese Wikipedia we use the Unblock-zh mail list for this propose.)--GZWDer (talk) 21:16, 11 July 2019 (UTC)
          • I guess we could do that. I'm not too fond of that idea either but it could work, and if there's consensus for that then maybe we can change the policy.--Jasper Deng (talk) 22:08, 11 July 2019 (UTC)
            • The other issue that we get into is what if they decide to run for administrator in the future? Then we're stuck having to reveal the previous account, or endlessly arguing about whether to reveal the previous account (and if they edited under their real name, then that's another mess entirely). --Rschen7754 04:23, 12 July 2019 (UTC)
              • The enwiki policy discussed the issue adequately. I think we can just borrow it, replacing ArbCom with the mail list.--GZWDer (talk) 05:18, 12 July 2019 (UTC)
                • I don't think we can. How do we get a consensus to release info? There's 13 of them and 50+ of us. It would devolve into bickering, and such a list would be difficult to actually keep things private (as many large mailing lists leak even if they are "private". Not to mention that WMF has removed two of our own admins before, one of them for real-life harassment). How do we make sure the mailing list doesn't get abused and decisions don't start to get made on it? No thanks. I don't want this nightmare. --Rschen7754 06:21, 12 July 2019 (UTC) (To add to that: if we want an ArbCom, let's call it an ArbCom and elect them, let's not just make our own semi-private-but-not-really-private-because-61-people-are-on-it mailing list and pretend.). --Rschen7754 06:27, 12 July 2019 (UTC)
                  • If you think the number of subscriber is an issue, the list may be limited to "functionaries", i.e. bureaucrats and/or oversights. In my opinion being an arbiter have no inherent difference to administrator, especially in WMF's point of view. If WMF can remove an admin, they can remove an arbiter too.--GZWDer (talk) 12:00, 12 July 2019 (UTC)
                    • "In my opinion being an arbiter have no inherent difference to administrator, especially in WMF's point of view" - that's the point, enwiki ArbCom is specifically approved to handle nonpublic information, and goes through a lot more scrutiny than an administrator does (when really, to pass our RFA process you have to get 8 supports and not enough opposes to drag you below 75%). --Rschen7754 18:14, 12 July 2019 (UTC)
                      • Bureaucrats needs 15 support votes and oversight needs 25, so they are considered more trusted. By contrast, English Wikipedia ArbCom election theoretically only require 50% support and no mandatory support vote requirement.--GZWDer (talk) 05:34, 13 July 2019 (UTC)
                        • But in practice, there are over 2,000 ballots, only the top X candidates are picked, and SecurePoll percentages are always lower than voting done in public (because your vote remains secret, you are free to oppose without consequence). --Rschen7754 23:58, 13 July 2019 (UTC)
    • If the previous account is declared, then it is not really a clean start though. --Rschen7754 00:15, 12 July 2019 (UTC)
      • It's pretty much "clean" if it is not too prominent on the userpage. It's not that ideal but I do think accountability is very important if we don't have a body trusted with private information.--Jasper Deng (talk) 00:39, 12 July 2019 (UTC)
  • Pictogram voting comment.svg Comment not sure to understand... What part of this policy « disallows clean starts »? For me, when someone is doing a clean start then the old account is not used anymore, so the new one is not an alternate as we're back at the « Editors are expected to edit primarily from a single account, used by a single person. » scenario. If my interpretation is wrong or is not clear then maybe we could precise or reword this policy to make it clearer. For the other cases listed on the policy, I still think that "must" is better than "encourages to". Cheers, VIGNERON (talk) 07:00, 12 July 2019 (UTC)
  • Pictogram voting question.svg Question How can we have a policy of our own, if we do not have the tools to verify an abuse? IP (Hej!) 07:36, 12 July 2019 (UTC)
    • CheckUser is always useful as there's usually an steward online. I don't think not having a local CheckUser will have any significant difference.--GZWDer (talk) 12:02, 12 July 2019 (UTC)
      • I agree using the Stewards fits us very well here. But they have to follow global policies. IP (Hej!) 12:24, 12 July 2019 (UTC)
  • I agree that the current Wikidata policy is too strict. Most people who actually abuse multiple accounts don't follow the rules in the first place and it limits well-intentioned use of multiple accounts by people who want to follow the rules. ChristianKl❫ 13:57, 15 July 2019 (UTC)


Arbitrary signature for archiving. — regards, Revi 12:00, 15 July 2019 (UTC)

Wikidata weekly summary #373

Merge request

See Talk:Q24969494. If this is not the proper place to discuss this, please inform me where is. Renard Migrant (talk) 19:50, 14 July 2019 (UTC)

@Renard Migrant: You want to activate the Merge Gadget. Then you can merge items as you need. Circeus (talk) 02:35, 16 July 2019 (UTC)

Property for higher order metaclass

We currently have

  • A is instance of B (A is order one lower)
  • A is subclass of B (A is same order as B)

Is there a property that allows A to be one level higher? I think this would be quite useful. It allows one to define vague things (a higher order metaclass) in terms of more concrete things.

For example, we can say something like

"A dish is a class of subclasses of food."

Here "class of subclasses" is the temporary name of such a property. What this is saying is that instances of "dish" are subclasses of "food". So for example, "sauerbraten" is an instance of "dish". From the property above, we deduce that "sauerbraten" is a subclass of "food".

The current way these items are set up, this example doesn't work. But I think this is an extremely useful property to have. "Dish" is in some senses more abstract (here, it is an order 2 metaclass). On its item page, you'd want to know how it is defined in terms of things that are more concrete, in this case "food" (an order 1 metaclass, and a subclass of "physical object").

--Johnmichaelwu (talk) 14:08, 9 July 2019 (UTC)

Interesting idea - maybe you should create a property proposal for this where we can discuss in detail? ArthurPSmith (talk) 14:55, 9 July 2019 (UTC)
We also have metasubclass of (P2445) View with SQID for stuffs with the same order which are not subclasses of each other, but their instances are (taxon ranks, for example).
I think there is current practice for this is to use a of (P642) View with SQID qualifier sometimes, I would not be surprised to find
⟨ dish ⟩ subclass of (P279) View with SQID ⟨ class ⟩
of (P642) View with SQID ⟨ food ⟩
⟨ dish ⟩ instance of (P31) View with SQID ⟨ first order metaclass ⟩
of (P642) View with SQID ⟨ food ⟩
. But maybe a new property is better indeed. author  TomT0m / talk page 15:48, 10 July 2019 (UTC)
Orders are a tricky business. It's hard to reason about them for people who are new to ontology and modeling decisions are often subjective in nature (often there are decision to be made about which aspects should are most important to the modelling). It's not clear to me why subclass of (P279) isn't enough for the relationship between food and dish. ChristianKl❫ 14:44, 16 July 2019 (UTC)

Duplicate page

Q13183538 and Q56379325 are for same person. Please merge Q13183538 into Q56379325. ~SS49~ {talk} 15:38, 19 July 2019 (UTC)

✓ Done by VIGNERON--Ymblanter (talk) 19:07, 19 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 10:55, 22 July 2019 (UTC)

Sheila Atim

Hi, I'm pretty new to this, but it looks like Sheila Atim (Q61989847) and Sheila Atim (Q65115171) both refer to the same person. Thanks! BennyOnTheLoose (talk) 00:40, 21 July 2019 (UTC)

Yes; merged. Thanks. --Tagishsimon (talk) 01:05, 21 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 10:55, 22 July 2019 (UTC)

Adding missing labels or descriptions in one or multiple languages

Hi everyone, I notice on my watchlist we have a quite a few people working on filling missing labels or descriptions for Wikidata items. It's very good that this missing data is added, but it's a lot of single edits, this report is dominated by these edits. Would it be feasible to cluster this so that a lot of missing labels and descriptions get updated at the same time? Might take a bit more time to setup, but the robot could just be scheduled to run every once in a while so that new items get updated too. You would only have to setup it once. Multichill (talk) 09:09, 29 June 2019 (UTC)

Maybe we could move to more structured way of doing things, e.g. Wikidata:Property_proposal/default_description_for_instances.
Also, maybe a nl label isn't needed for an article with an English title, at least when it merely copies the English label. "an article" here means >20,000,000 million items. --- Jura 16:29, 29 June 2019 (UTC)
Every item on Wikidata should have a Dutch label and a Dutch description. The descriptions shouldn't be one dimensional (copy of the instance) most of the time. For example for artists we combine the country and the occupation to get something like "Nederlands kunstschilder" ("Dutch painter") and for works of art the type of work and the creator to get something like "schilderij van Vincent van Gogh" ("painting by Vincent van Gogh"). Often these kind of descriptions need to be tweaked a bit. Doesn't look like this new property is going to help a lot here. Multichill (talk) 17:07, 29 June 2019 (UTC)
In general, I agree. For articles, I don't think the nl-label is efficient or useful. Have a look at Q57665075. --- Jura 17:10, 29 June 2019 (UTC)
Every item on Wikidata should (ultimately) have a label and description in every language, IMHO. How we get there is a good question. - PKM (talk) 22:44, 29 June 2019 (UTC)
  • Why is that? Why would it be needed for items while it's not needed for other entities? Can you explain why the nl-label needs to duplicate the en-label at Q57665075? --- Jura 22:48, 29 June 2019 (UTC)
I share the sentiment that every item and property on Wikidata should ultimately have a label and description in every language (having the same question of how to get there coming to mind often), but I do agree with you, @Jura1:, that the duplication of the en label in the nl field is questionable and possibly outright incorrect. I think a better nl label would render the title of the article entirely in Dutch rather than preserving the original English. I wonder how many such nl labels are carried over from bracketed en labels (where the bracketing indicates that the label was translated from another language). Mahir256 (talk) 22:57, 29 June 2019 (UTC)
  • Perhaps there should be a label for every item in every language, but mass creating misleading or bad labels via bot is not the solution. See for example [25], just one example of an often-repeated problem. In this instance, a Polish translation of an Ancient Greek play is described as "edition of/from Sophocles". Compare that description with the English description "1916 Polish translation by Kazimierz Morawski". The Dutch description does not distinguish the item from any other "edition of/from Sophocles". It also (incorrectly) implies that Sophocles was responsible for the edition, or perhaps that there is only a single work by Sophocles, neither of which is the case. The data item is for a Polish translation made by Morawski, and the language into which the translation was made is of central importance for the identification of the data item, yet it is not mentioned in the Dutch description. As a result, we have a very poor description added en masse by bot. --EncycloPetey (talk) 23:46, 29 June 2019 (UTC)
Mass introduction of mistakes by bots should be avoided, but an occasional mistake can happen. This is still a wiki so you can always improve something.
Can we go in small baby steps? Take for example this edit by Liridon. We could invest some time to figure out between what languages names can safely be copied like for example this edit by Sjoerd. Multichill (talk) 11:29, 6 July 2019 (UTC)
Sjoerd, i'm not sure what the problem with that edit is? Would you kindly elaborate? --Trade (talk) 19:26, 9 July 2019 (UTC)
What's your opinion on the 20,000,000 items for articles and their labels in Dutch? --- Jura 11:40, 6 July 2019 (UTC)
The unused ones should be deleted? The whole concept of having items for citations was nice until some people hijacked the concept and turned Wikidata in an unused scientific article dumping ground. Multichill (talk) 16:25, 8 July 2019 (UTC)
@Multichill: The problem with this edit, is that it is hard for a software to do that. If the origin of the persons name would have been from russian or another non-latin alphabet, you cannot copy the label from one language with latin script to another. You have a similair problem with descriptions. We have added "instance of:asteroid" to thousands of items. And they also have that description in many many languages. But what should I do when I discover that this was not at all an instance of:asteroid? I can easily change the statement, but not the bad descriptions. Since I do not know many langauges, I can only delete them, but even that is hard work. So, please let them be! Concentrate on languages and subjects where people do not read English very well. I doubt you have such problems in Dutch-speaking areas. From experience I know that we do well in Swedish speaking areas. And you are not dumber than us. Au contraire! IP (Hej!) 19:46, 9 July 2019 (UTC)
Here is another example where it probably went bad. His name according to P1559 was Bengt Oxienstierna. Bengt Oxienstierna is most likely the better label, but instead we have copied a text that in practise adds the Patronymicon as a separator. IP (Hej!) 08:13, 10 July 2019 (UTC)
And here an example of bad script-adding of descriptions. IP (Hej!) 13:29, 10 July 2019 (UTC)
@Jura1: there's the solution for articles: I've proposed a month ago Wikidata:Project_chat/Archive/2019/06#Multilanguage_label. --Infovarius (talk) 12:17, 17 July 2019 (UTC)
It's somewhat different. The English article could still have a (translated) title in Dutch. Your proposal sounds much like one of mine you didn't support earlier ;) Wikidata_talk:WikiProject_Names#New_datatype:_monostring_item --- Jura 18:12, 17 July 2019 (UTC)

To disambiguate or not to disambiguate

If an item is a disambiguation page in some projects and normal article in other projects, how does Wikidata decide whether it's a disambiguation or not? Example: Q292934. Bennylin (talk) 18:15, 14 July 2019 (UTC)

@Bennylin: They're different concepts so should be split into different items. Someone needs to go through all the sitelinks and decided what concept they represent. Cheers, Bovlb (talk) 18:48, 14 July 2019 (UTC)
Sometimes they are indeed the same concepts, just some have template "disambig" and the others not. Randomly. --Infovarius (talk) 10:15, 17 July 2019 (UTC)

Religion and sources

Hello. I have removed religion (P140) ->atheism (Q7066) from Dimitris Christofias (Q57334). @Ghuron: undo it "because included in the Category:Atheists (Q8985993)". The statement has no source (imported from Wikimedia project (P143) -> Serbian Wikipedia (Q200386) is not a source). I have tried to find a source that 1) he said that was atheism or 2) importance sources that said that he was atheism, but nothing. So why to have the statement just because he is included in the Category:Atheists (Q8985993). If it helps, Dimitris Christofias (Q57334) was buried after an orthodox christian ceremony [26], but I have found no source that he said that was an orthodox christian or an importance source that claimed that he was orthodox christian. So, I believe that we can not add anything with religion (P140). Xaris333 (talk) 15:50, 16 July 2019 (UTC)

@Xaris333: why we cannot do this? --Ghuron (talk) 17:13, 16 July 2019 (UTC)
I am not contributing in Serbian Wikipedia (Q200386), I don't know the language. :) Xaris333 (talk) 17:44, 16 July 2019 (UTC)
Neither do I, but google translate is enough to understand that there are no sources there that support atheism of Dimitris Christofias (Q57334). Please let me know if you see any other my erroneous edits --Ghuron (talk) 17:57, 16 July 2019 (UTC)
Given that religion (P140) is privacy sensitive I oppose adding it based category pages of Wikipedia. It should only be added if you can provide a source. Via our privacy policy it's clear that when there's a conflict the burden of proof is with the person who wants the information to be listed. ChristianKl❫ 19:39, 16 July 2019 (UTC)
Maybe WD:BLP#Bot interaction with items for living persons should apply here given automatic nature of this job. — regards, Revi 10:26, 17 July 2019 (UTC)

Orcid Number

How is the number 4.506.903 at Property talk:P496 generated/retrieved? The orcid website says there are 6.792.120 IDs. Steak (talk) 12:50, 17 July 2019 (UTC)

@Steak: I think Module:Property documentation gets it from the ORCID iD (P496) number of records (P4876) statement(s). --Lucas Werkmeister (talk) 12:54, 17 July 2019 (UTC)
Ahh, right, thanks. Steak (talk) 13:15, 17 July 2019 (UTC)

How can one customize permissions in a wikibase instance?

I would like to know how to customize permissions on a wikibase instance created from docker. Is it possible? Any hint to some documentation?

What I would like is to have allow some groups of users to edit some items, while allowing other groups of users to only visualize/query the data.

 – The preceding unsigned comment was added by Jelabra (talk • contribs).

How can I import properties from an ontology in a wikibase instance created with docker?

Is it possible to import properties from an external ontology so I could use them in my own wikibase instance?

For example, if I wanted to use the property owl:disjointWith from OWL in my own wikibase instance, how could I do it?

I saw that I could create my own properties, but I would also like to use other properties from external ontologies.

 – The preceding unsigned comment was added by Jelabra (talk • contribs).

Unknown value plus stated as

How do I input this? I want to have "Lilian H. Zirpolo" as the editor at Historical Dictionary of Baroque Art and Architecture (Q65621317), similar to how it is done at The Welsh Academy encyclopaedia of Wales (Q56848446). But if I only put the qualifier and no value under editor, it doesn't let me save. What am I missing? Calliopejen1 (talk) 19:32, 17 July 2019 (UTC)

I'm not sure why you couldn't save this. I added the "stated as" qualifier with no problem. The best solution is to create an item for Lilian H. Zirpolo - she's in VIAF - and delete the existing author name string statement and change the editor to the new item. The Cradle tool author form and Authority Control gadget make this really easy. - PKM (talk) 20:10, 17 July 2019 (UTC)

Commons P180 and Wikidata:Notability

If a user uploads an image to Commons and uses the depicts (P180) field, is it acceptable to create a new item for a person (or thing) that would normally fall outside of Wikidata:Notability? U+1F360 (talk) 19:15, 15 July 2019 (UTC)

No, I would say this is not acceptable.--Ymblanter (talk) 19:18, 15 July 2019 (UTC)
Yes, for me clearly acceptable & generally highly desirable under criteria 3: It fulfills some structural need, for example: it is needed to make statements made in other items more useful. --Tagishsimon (talk) 21:24, 15 July 2019 (UTC)
This is equivalent to saying that every detail of every photo ever uploaded to Commons is notable. I do not see how this follows from our notability policy.--Ymblanter (talk) 19:49, 16 July 2019 (UTC)
Yes for the reasons stated by Tagishsimon. Also when paired with relative position within image (P2677) and/or instance of (P31) this allows for labeling items within the image which can be useful for a number of tasks. Example  ElanHR (talk) 02:29, 16 July 2019 (UTC)
I think we should be cautious here, because a commons photo can depict anything and anyone. So allowing automatic notability provides a wide open gate to anything. I generally favour inclusion on wd, but this looks too broad to me. I drafted the relevant paragraph on Commons, which encourages users to be as specific as they can be, given the existing set of WD options (rather than listing everything it is a subclass of), but this proposal would go even deeper, with no clear limit. --99of9 (talk) 05:11, 16 July 2019 (UTC)
I think the intention of the structural needs paragraph is to speak about the structural needs of Wikidata and not of Commons. ChristianKl❫ 07:30, 16 July 2019 (UTC)
I think if the person on a photo is not notable in Wikidata the photo can also get deleted on Commons, because it is out of scope there. If the identity of the person on the photo is not notable it dose not need the person as depicts (P180), then can human (Q5) just be use there. --GPSLeo (talk) 07:59, 16 July 2019 (UTC)
The trick, as always, is where to draw a line on notability. On the English-language Wikipedia, much to my amazement, members of the 9-member city council of the city of Seattle are not considered inherently notable, but anyone who ever competed in the Olympics is.
How notable does a band have to be before each of its members is individually notable? Is a person known mainly for always being in the lead of an annual parade in some city notable? On Commons, certainly both of those merit being individually called out in describing what the picture depicts, though as thing stand we might not give them a cagegory. - Jmabel (talk) 15:36, 16 July 2019 (UTC)
The notion that we explicitly “draw lines” w.r.t. notability in Wikimedia projects is not ideal in my opinion as there is a much more important condition: we consider an entity to be notable if there is serious coverage about it somewhere outside of the Wikimedia world; that usually excludes user-generated content sources, sources that are under control of the entity to be described, and some others. Wikipedias do indeed have to be a little more rigorous about notability, as the question about which source is serious enough can be debated. As a practical approach, they do it indeed by somewhat arbitrarily “drawing lines” as an auxiliary measure. Here at Wikidata, the notability policy is designed in a way that external coverage about an entity is accessible realtively easily from each item, i.e. by references, identifiers, or connected Wikipedia articles (which themselves are usually backed by sources). The structural need criterion is an exception, but this term not well defined and it is generally not a good idea to rely on it alone if notability is questionable.
The problem with Commons content is that practically all the auxiliary information in that project beyond the files themselves, i.e. the information which is now gradually being transfered to SDC/Wikidata, is not backed by any sources—contrary to most other Wikimedia projects. One could of course semi-automatically transfer a reasonable amount of information to Wikidata by creating new items based on existing information at Commons, just for the sake of having it described in a structured manner. However, those items would be unsourced, and it is in fact very difficult to work with them if they are not linked against external sources; from Wikidata perspective, unsourced content is pretty much dead and unverifyable content, as most community members are not able to work with such items. Some Wikidata editors also fear that the possibility to have unsourced content in Wikidata would be exploited for undesired purposes which would be much more difficult to control. —MisterSynergy (talk) 17:04, 16 July 2019 (UTC)

I'm a newbie here, so take my opinion with a big grain of salt, but my thinking is that it depends on whether the image uploaded at Commons is a work that is notable here. If a notable (by WD standards) painting depicts someone who is not notable, it may be worth creating an item about that person so that the item about the painting is more useful. If the image uploaded at commons is some random photo, I don't think we want an item about someone random in that photo. Calliopejen1 (talk) 20:25, 16 July 2019 (UTC)

That distinction seems like a reasonable one to consider, because it implies a direct structural need. --99of9 (talk) 03:55, 17 July 2019 (UTC)
I guess whatever is worth naming in a label on Wikipedia or Commons could be worth having items for here. And they do not label random people neither on Commons or on Wikipedia, only because they accidently can be found in i picture next to something notable enough to have a Commons-category or WP-article. IP (Hej!) 07:44, 17 July 2019 (UTC)

There are two ways to look at this: is something notable *because* it was depicted? No, that would be too much. Can you create a new item because you need it to add a P180/depicted? Probably yes, but not always. I would also suggest to only add P180 to pictures on Commons where it would make sense to search for, and it would be useful in a search result. A very small church in the far background of the image will not make sense in a search result for "church". Edoderoo (talk) 08:47, 17 July 2019 (UTC)

  • Do we have any examples of what people think are poor entries, that violate Wikidata notability? Is this about people sneaking themselves in, or making entries for their favorite coffee cup? Some examples would be helpful. --RAN (talk) 23:42, 18 July 2019 (UTC)

Letter-digits or numeral letters

How to model letters (usually of old alphabets) which were used as a numerals? The brightest examples are Coptic (Q65599830) and (Q65600159) which are even not used as usual letters. Other example is Cyrillic alphabet which letters were used as numerals some time ago (Cyrillic numerals (Q491668)) or Greek numerals (Q108645) or other alphabetic numeral system (Q4063087). I want to add numeric value (P1181) but it constrained to some quantity. I've added numerical digit (Q82990) to widen the constraint but I doubt now. Not all digits (ex. positional) have predefined numerical value. What are (Q65599830) and (Q65600159) after all? Some new class? --Infovarius (talk) 11:35, 18 July 2019 (UTC)

Additionally, shouldn't we have all these digits as items, separate from Arabic numeral (Q29961325)? --Infovarius (talk) 12:08, 18 July 2019 (UTC)
What's wrong with eg ג (Q15074) numeric value (P1181) "3", perhaps with qualifier subject has role (P2868) = numerical digit (Q82990); or L (Q9927) numeric value (P1181) "50" with qualifier subject has role (P2868) = Roman numerals (Q38918) ? Jheald (talk) 12:31, 18 July 2019 (UTC)
It might make sense to have separate items for the numeric meanings of these symbols? At least for the Cyrillic and Greek cases they do have specific numeric values (in the range 1 - 900). I'm not familiar with the Coptic case, do you have a reference on that? Jheald's solution seems fine to me also. ArthurPSmith (talk) 14:38, 18 July 2019 (UTC)


Familysearch now has a landing page for entries on people, for users who are not logged in, using the standard Familysearch ID: Should we change the url formatter to this new version? Here is the old version and what you see when not registered, or registered but not logged in: What do you think? --RAN (talk) 02:12, 19 July 2019 (UTC)

Symbol support vote.svg Support makes sense to me. ArthurPSmith (talk) 13:15, 19 July 2019 (UTC)
Symbol support vote.svg Support instead of replacing it, perhaps just add the new value with preferred rank (since the old formatter is still valid)? --SilentSpike (talk) 14:52, 19 July 2019 (UTC)
When I am logged in to Familysearch and I click on the url to the new style landing page, at the top is a link to login (despite already being logged in). When I click on the login button, it automatically directs me to the old url displaying the full data. So, it may not be necessary to have both. The new link redirects to the old link once logged in. --RAN (talk) 20:11, 19 July 2019 (UTC)

Inverse Properties

I would argue that child (P40) is not inverse property (P1696) of father (P22) and mother (P25). Reason: By "A child (P40) B." you can not conclude that "B father (P22) A.".

Also, father (P22) and mother (P25) are not inverse property (P1696) of child (P40), because inverse property (P1696) is a symmetric property. Wiljes (talk) 06:03, 19 July 2019 (UTC)

I think we also have a lot of subproperties which are not really subproperties at all. I think those statements are barely used in practice, so this is probably a big mess overall. author  TomT0m / talk page 16:35, 19 July 2019 (UTC)

Improbable birth date added for Augustus

In this edit Sergey kudryavtsev added the date 23 September 63 BCE as the birth date of Augustus (Q1405), and marked it as belonging to the Julian calendar (or accepted the default). Although this date can be found in sources, such as it is highly probable this date is in the w:Roman calendar. Due to irregular insertion of intercalary months and a lack of records concerning when these months were added, it is usually impossible to accurately convert a Roman calendar date to either of the two calendars that can be used in Wikidata, proleptic Gregorian calendar (Q1985727) or proleptic Julian calendar (Q1985786). I believe the edits should be reverted. Jc3s5h (talk) 22:18, 16 July 2019 (UTC)

As you described it, it looks to me like the main problem is the use of unspecified calendar; assumed Julian (Q26877143)? Why (instead of a complete removal) then not simply change it to something like "unspecified calendar, assumed roman" or only "unspecified calendar". IP (Hej!) 07:51, 17 July 2019 (UTC)
The only calendars that are supported by Wikidata software are proleptic Gregorian calendar (Q1985727) or proleptic Julian calendar (Q1985786). These are the only two calendars that can be specified in Wikidata whenever a date is required. Using unspecified calendar (Q18195782) would be appropriate if it is uncertain which of the two allowable calendars is meant by a source. But it is not appropriate when the most likely calendar is one that is not supported by Wikidata.
If we look at w:Augustus we see that paragraph 5 footnote a of Suetonius is cited as the source. I personally would not want to cite a 2000-year-old source, and I have not had a chance to visit the library to find something better. The translator's footnote says in relevant part "The 23rd of September was Augustus's birthdate in the calendar of the time. It was, however, out of whack with the seasons, as is proved by the frequent statement in ancient authors that his birth sign was Capricorn; some modern writers state that Capricorn was the sign under which he was conceived." Since the "calendar of the time" was the Roman calendar, and since the Roman calendar was known to be out of whack with the astrological/astronomical signs (no distinction between those fields back then) while the Julian calendar was in reasonable agreement, it's highly probable the date is a Roman calendar date. Jc3s5h (talk) 16:10, 17 July 2019 (UTC)
  • Jc3s5h: Could you use Help:Ranking when making such advance date editing? Otherwise people are likely to re-add the date and you are likely to feel they missed your point. --- Jura 13:47, 20 July 2019 (UTC)
    I have considered that. Considering the quality and explicitness of the source I found, I should probably make the statement preferred rank.
    If I were going to mark a value as depricated, I would like to have
    1. A source that says the wrong value is widespread but wrong, or
    2. A superficially reliable source that clearly gives the wrong value (for example, clearly says August was born 23 September 63 B.C. in the Julian calendar, together with a better source included via the statement disputed by (P1310) property which contradicts the wrong value. (I didn't know about statement disputed by (P1310), thanks for pointing me in that direction.
    I am not aware of a source for the Augustus birth date issue that satisfies the criteria I would like.
    Jc3s5h (talk) 16:32, 20 July 2019 (UTC)
    • The idea is to input values found in sources (or an interpretation of the source) and then rank them. By default, everything should get normal rank. If some are clearly wrong or require additional references, I'd deprecate these. Then it's an editorial choice which of the normally ranked ones should get preferred rank. Sometimes, there is a clear choice for that, sometimes not. --- Jura 17:05, 20 July 2019 (UTC)
      • If a calendar is whacky, how does that matter? The Swedish calendar was whacky, the transit from julian to gregorian calendar was whacky. We still add extra seconds from time to time, so our calendar is still whacky. The constant shifts between normal and daylight saving time is whacky. Even if it does not have the precision of a atomic clock (Q227467), the roman calendar was still the general accepted method to measure time. What is strange about that? Very recently, the definition of kilogram (Q11570) changed. Last time I checked (at it was some years ago) the definition of volt (Q25250) was very whacky. If I remember correctly, that was why mains electricity (Q387400) changed when we joined the EU. IP (Hej!) 17:30, 20 July 2019 (UTC)


At Frederick George Shaw (Q63985101) a bot keeps removing that his sport is fishing, any idea how it should be structured to remain? --RAN (talk) 23:29, 18 July 2019 (UTC)

@Vanbasten 23: --Tagishsimon (talk) 23:42, 18 July 2019 (UTC)
There is an item for fishing as a sport: recreational fishing (Q283534) --Bluemask (talk) 02:43, 19 July 2019 (UTC)
That solves it! --RAN (talk) 04:09, 19 July 2019 (
The bot "understands" that "fishing" is not a sport, but if you have to put recreational fishing (Q283534) I can configure it to change one thing for another. Thanks and sorry. --Vanbasten 23 (talk) 08:07, 19 July 2019 (UTC)
The bot might understand but I don't! Surely there should only be one recreational fishing item. All sports are recreational activities - we don't have separate items for (for example) professional netball as a sport as opposed to netball as a hobby. Shouldn't these two be merged? (It does make sense to keep the occupation of non-recreational fishing a separate item). Waggers (talk) 12:20, 19 July 2019 (UTC)
FYI, what the bot was removing was the entry for "fishing" generally that comprises both recreational and commercial fishing. See fishing (Q14373) The issue of there being two recreational fishing items is a different topic. Calliopejen1 (talk) 19:23, 19 July 2019 (UTC)

What is the difference between recreational fishing (Q283534) and recreational fishing (Q5493838) both have Finnish definitions and I am not using Chrome today so I do not have auto-translate available. --RAN (talk) 04:14, 19 July 2019 (UTC)

@Richard Arthur Norton (1958- ): After investigation, it looks to me like Finnish Wikipedia treats recreational fishing (Q5493838) as recreational fishing and recreational fishing (Q283534) as sports/competitive fishing, though I might be wrong. recreational fishing (Q5493838) says something to the effect of "done mostly to acquire experience". Circeus (talk) 15:52, 20 July 2019 (UTC)

Behavioral policy

Hello, I want to point out that draft of behavioral policy is out now, it would be great if people review this and raise their points before this moves forward. Amir (talk) 15:21, 20 July 2019 (UTC)

maintained by WikiProject (P6104) on focus list of Wikimedia project (P5008)

What is the difference between maintained by WikiProject (P6104) and on focus list of Wikimedia project (P5008)? Could they be merged? Eurohunter (talk) 16:22, 20 July 2019 (UTC)

Q49324903 and Q7641353 - duplicate?

Both these items seem to refer to the same place but have different GeoNames IDs. Are they still considered duplicates? Ixfd64 (talk) 17:17, 20 July 2019 (UTC)

Item associated with this person

At one time I thought there was a field for an image with a field name something like "item associated with this person". What was the proper name, or did I imagine it? For instance it might contain an image of an article of clothing they wore from a museum. --RAN (talk) 04:42, 20 July 2019 (UTC)

There's related image (P6802). - PKM (talk) 19:10, 20 July 2019 (UTC)

List of missing labels

According to thread WikiProject Random above I wanted to ask if there are lists of items with missing labels sorted by field (music, video games, films etc.). Eurohunter (talk) 15:51, 20 July 2019 (UTC)

Eurohunter: You can use Wikidata Terminator for that. Esteban16 (talk) 15:57, 20 July 2019 (UTC)
@Esteban16: It doesn't works (it stuck at "Getting item candidates") and there is no way to choose field of interest. Eurohunter (talk) 16:05, 20 July 2019 (UTC)
Eurohunter: The tool seems to has an issue. You can pick the property and value of your choice in any language and it'll display the results (e.g. instance of (P31) -> video game (Q7889)). Esteban16 (talk) 23:13, 20 July 2019 (UTC)
Not sure about a sorted list, but there is Special:EntitiesWithoutLabel. You can find these kind of lists at Wikidata:Database_reports. --SilentSpike (talk) 23:27, 20 July 2019 (UTC)

WikiProject Random

I quite enjoy adding statements and sources to random items - you come across some interesting stuff and it's rewarding to see an item become much more fleshed out.

With that in mind, I decided to make Wikidata:WikiProject_Random as I think it could be an interesting way to collaborate. Check it out and fire away on the discussion page if you think anything could be improved! --SilentSpike (talk) 23:12, 19 July 2019 (UTC)

@SilentSpike: How is this different in spirit from the Wikidata:Flashmob? Mahir256 (talk) 01:00, 20 July 2019 (UTC)
I was not aware of that project - probably because the name is not easy to find. Thanks for bringing it to my attention --SilentSpike (talk) 07:14, 20 July 2019 (UTC)
After reading the page I would say the difference with this project is:
  1. The goal is really trying to flesh things out with sources/statements/constraints rather than just labels and descriptions. Encouraging research on the random topics.
  2. It's not anonymous and ideally collaborative discussion would be more encouraged - making use of the talk page.
  3. The method is to randomly generate items once per week rather than summoning a group by an individual. This means there's consistently new things to check out each week and it's an unbiased selection process.
With all of that said, the spirit of the idea is similar. I'd been looking for a project like the flashmob so thanks again! --SilentSpike (talk) 07:26, 20 July 2019 (UTC)
  • Shall we use on focus list of Wikimedia project (P5008) for this? --- Jura 16:14, 20 July 2019 (UTC)
    • Is there an advantage to that approach? I would think that property is more to mark a long-term interest in items (which perhaps we should). --SilentSpike (talk) 19:48, 20 July 2019 (UTC)
      • We could query to see how they evolve, at least for L and Qs. --- Jura 08:45, 21 July 2019 (UTC)
  • Is Flashmob still active? - PKM (talk) 19:07, 20 July 2019 (UTC)

Statement display

Is there a gadget or user script which allows statements and their qualifiers to be visualized and edited in a table layout? Something like this?

Property Values Qualifier 1 Qualifier 2 Sources Edit
item item item; item 2 [show] Edit this at Wikidata
item item 2 [show] Edit this at Wikidata
item 8 items [show] 2 [show] Edit this at Wikidata

For certain properties this would probably make it a lot easier to add multiple statements without having to make a QuickStatements batch. (It takes a lot of time just to select qualifiers, especially in the British English locale because some properties' labels don't load properly.) I don't know if this sort of thing exists yet, though. Jc86035 (talk) 14:53, 20 July 2019 (UTC)

Have you looked at Tabernacle? Here's an example with qualifiers and references.
If you use a list of items or a query, you can drag-and-drop statements from one to another.- PKM (talk) 19:26, 20 July 2019 (UTC)
I haven't; thanks for linking to that. I'll try it out at some point. Jc86035 (talk) 16:35, 21 July 2019 (UTC)

Languages on labels list

How to edit (translate) names on languages on labels list? Some of them are not translated. ~ Eurohunter (talk) 09:54, 21 July 2019 (UTC)

I like using the WE-Framework gadget. Of course, why editing labels in different languages aren't enabled by default is beyond me. --Trade (talk) 14:32, 21 July 2019 (UTC)
@Trade: I mean something different. For example if you ad "fr" it tell it's "French" instead of name in your language - "Französische"/"francés"/"fransk"/"francuski" etc. (French is already translated but some languages not, it refers mostly to "rare languages"). Btw. WE-Framework looks intersting. Can I install it on ENwiki or at Wikidata? Eurohunter (talk) 15:01, 21 July 2019 (UTC)
I wish Wikidata would just use the name of items in english when it doesn't have a label in your native language instead of giving you a bunch of numbers. I'm pretty sure WE-Framework works on both ENwiki and Wikidata. I mostly use it as an replacement for QuickStaments since i haven't figured out how to use it. --Trade (talk) 15:06, 21 July 2019 (UTC)

Invalid archive links

I've just added a format constraint to archive URL (P1065), and it seems that there are a lot of concerning invalid URLs (at least, ones which don't match the regular expression). Examples:

GreenC's bot usually fixes these sorts of things on certain Wikipedias, but as far as I'm aware there isn't anyone fixing these right now. Hopefully these can be cleaned up before anyone actually tries to use our references. Jc86035 (talk) 14:54, 21 July 2019 (UTC)

  • Looks like the new regex doesn't allow the default archive used by frwiki [27]. --- Jura 16:08, 21 July 2019 (UTC)
    @Jura1: This is an error on my part (I forgot that it existed) and I'll fix it in a few minutes. Jc86035 (talk) 16:31, 21 July 2019 (UTC)
  • @Jc86035: A good source for archive URL formats is here. As can be seen, the formats are quite diverse, even within a single provider (e.g. NLA Australia and WebCite). This is not a comprehensive list but the archive services I am aware of in use on Enwiki, there are likely some more. Other languages probably have others (the frwiki "wikiwix" is included in this list). A single regex filter would be challenging to construct, a lot of real estate to cover. I can provide regex statements for each individually if interested. -- GreenC (talk) 16:17, 21 July 2019 (UTC)
    @GreenC: The regex I used is on the property page. It only works for seven services right now (Wayback Machine, Internet Archive files,, Webrecorder,, WebCite and, but most of the errors are nevertheless neither links to valid copies of the pages nor links to the archive services listed on the page you linked to. I think simple domain matching would be passable for the rarer services, at least for now; so far I've only done more complex matching for and Jc86035 (talk) 16:31, 21 July 2019 (UTC)
Agreed, a lot of garbage ends up the archive fields. On enwiki, my bot detects if the url and archiveurl are the same and removes the archiveurl if true. On Wikidata, if there is an archiveurl but no url, and the archiveurl is not a valid/known webarchive, it might reasonably be moved to the url field.. but would need to spend time in the data to make sure that is a decent heuristic. -- GreenC (talk) 16:59, 21 July 2019 (UTC)

Bots for linking identically named categories

Hi. Is there an existing bot for linking identically named category on Commons with its equivalent on Wikipedia? I noticed that none of the categories in c:Category:Townships in Pennsylvania by county are properly linked up. I am willing and capable of writing my own bot if necessary but I would rather use an existing one if possible. Magog the Ogre (talk) 15:43, 21 July 2019 (UTC)

I recall doing something like this back in the interwiki days, but I can't find the query. Probably got lost in one of the moves. You can just do a database query for the categories on the English Wikipedia and Wikimedia Commons that have the same name, but the Commons Category doesn't exist on the linked Wikidata item. Not all exact matches are the same by the way, some disambiguation might not have happened on either side so you have to apply some filtering. Multichill (talk) 17:08, 21 July 2019 (UTC)

Music festivals

Hi, anyone experience working on music festivals? I looked around and couldn't find any really good examples so I expanded Pinkpop 2019 (Q58329019). It contains things like where it was, when it was, who was playing, how much it costed and amount of people (once that data becomes available). Anything missing? Some points I ran into:

What do you think? Anything missing or should be improved? On User:Multichill/Pinkpop I created an overview of all editions. Multichill (talk) 13:42, 21 July 2019 (UTC)

I use the gadget Recoin which suggests a number of additional properties. The most important I'd say is: organizer (P664). - Kosboot (talk) 14:25, 21 July 2019 (UTC)
@Kosboot: Good point, I skipped it because it was this tiny organisation (Buro Pinkpop (Q65720246)) that mostly focuses on the festival. Added now. Also added presenter (P371). Multichill (talk) 14:56, 21 July 2019 (UTC)
Looks good. If one uses a value in part of the series (P179) probably it shouldn't be in instance of (P31) as well. --- Jura 14:43, 21 July 2019 (UTC)
@Jura1: I looked at that when I was expanding the item and decided to do both. High profile items like Eurovision Song Contest 2019 (Q9095390) and 2018 US Open (Q50391517) are also done like that. Bit of redundancy doesn't hurt I guess. Multichill (talk) 14:56, 21 July 2019 (UTC)
I have removed them and added part of the series (P179) if neede. Eurohunter (talk) 15:35, 21 July 2019 (UTC)
  • Oh, the tennis stuff is somewhat complicated ;)
    To determine if one wants to use music festival edition (Q41582469) or not, it might be interesting to see if the properties used on these instances are similar to the generic festival edition (Q27968043) or not. --- Jura 16:41, 21 July 2019 (UTC)
Other festivals, Tour de France, etc., it's all using instance of (P31).
@Eurohunter: please don't remove valid statements especially when conversation is ongoing and the removal is disputed. Multichill (talk) 16:52, 21 July 2019 (UTC)
@Multichill: If they are obviously incorrect they should be removed (it was known before this discussion). You should notice that they are not valid. Eurohunter (talk) 18:12, 21 July 2019 (UTC)
Huh? You're disputing the fact that the festival took place in June 2019? Multichill (talk) 18:47, 21 July 2019 (UTC)
No but there is already start time and end time. Eurohunter (talk) 19:11, 21 July 2019 (UTC)
Both time and space can be represented in multiple ways. Having one representation doesn't mean you should remove the other one. Multichill (talk) 19:46, 21 July 2019 (UTC)
You should add references otherwise data is unreliable. Eurohunter (talk) 15:35, 21 July 2019 (UTC)

50 million triples deleted last week?

I just looked at this visualization of the number of triples and was wondering what that large-scale deletion (or merging?) was about. Any pointers? Do we have tools that provide answers to such questions? Thanks, --Daniel Mietchen (talk) 20:42, 21 July 2019 (UTC)

There's a ticket. Gene merging, maybe, but it was slightly more than a month's worth of triple-growth, so maybe not. --Tagishsimon (talk) 20:57, 21 July 2019 (UTC)

Zewditu Menilek and Zewditu I - duplicate?

Looks like Zewditu Menilek (Q56825534) and Zewditu I (Q148524) may refer to the same person. Slight difference in birthdate, so I'm not 100% sure. Can someone take a look? thanks! BennyOnTheLoose (talk) 10:55, 26 July 2019 (UTC)

Yes, it looks like the same person. I merged the items.--Ymblanter (talk) 19:52, 26 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 10:24, 27 July 2019 (UTC)

Retrieving a QID from a DOI query?

Is there a template to return a wikidata QID from a given DOI?

  • e.g. {{DOI_to_QID|10.1101/102046}} → Q28529449

I'm looking to implement something in Wikipedia and WikiJournals to return various Wikidata parameters of a publication given only the DOI. Thanks in advance. T.Shafee(evo&evo) (talk) 04:00, 20 July 2019 (UTC)

@Evolution and evolvability: You can do it with a query (example), but I don’t think it’s possible in Wikitext. --TweetsFactsAndQueries (talk) 15:08, 20 July 2019 (UTC)
Isn't a Lua module, which the template would then call, workable? Circeus (talk) 15:56, 20 July 2019 (UTC)
You have to search for it, e.g. haswbstatement:P356=10.1371/JOURNAL.PONE.0029797. You can do that by bot or query, but I don't think it can be done by Lua. The best you can do is probably something like en:Template:Cite Q, which takes the Wikidata ID as the parameter and returns the rest of the article information. Thanks. Mike Peel (talk) 16:22, 20 July 2019 (UTC)
@Mike Peel, TweetsFactsAndQueries, TweetsFactsAndQueries, Circeus: Is that a fundamental limitation of the wikitext/lua language or just something that hasn't yet been implemented? It seems that part of the reason Cite Q isn't used is that looking up the QID is an extra hassle if you only know the doi/pmid. T.Shafee(evo&evo) (talk) 00:42, 21 July 2019 (UTC)
Mostly the latter. --Matěj Suchánek (talk) 10:47, 22 July 2019 (UTC)

Racing and race

Shouldn't be race separated from racing (Q878123)? Eurohunter (talk) 21:48, 21 July 2019 (UTC)

@Eurohunter: race (Q22938576) already exists, so sitelinks could be moved between the two items. Jc86035 (talk) 08:30, 22 July 2019 (UTC)

documentation ? - name for Wikidata Query Service visualization package

The Wikidata:Wikidata Query Service can output data from queries into various visualizations.

Do we have a documentation page for the WDQS visualizations? Do we call these the "Wikidata Query Service Visualizations"?

Answer here or at Wikidata_talk:SPARQL_query_service#Name_for_the_visualization_suite?. Thanks. Blue Rasberry (talk) 22:44, 21 July 2019 (UTC)

Wikidata:SPARQL query service/Wikidata Query Help/Result_Views. --Matěj Suchánek (talk) 10:37, 22 July 2019 (UTC)

 Resolved This is what I wanted. Blue Rasberry (talk) 13:12, 22 July 2019 (UTC)

Wikidata weekly summary #374

archive url and archive date

archive URL (P1065) and archive date (P2960) are Wikidata property to indicate a source (Q18608359). Is there a property constraint that, if the source have the one then must have the other too? And if yes, do you believe we have to add that constraint? Xaris333 (talk) 20:32, 16 July 2019 (UTC)

Or where can I suggest the creation of that type of constraint? Xaris333 (talk) 14:05, 18 July 2019 (UTC)

  • In sources, I think "retrieved date" is the one more likely to be used. --- Jura 13:48, 20 July 2019 (UTC)

Ok. But I think that has nothing to do with my question. Xaris333 (talk) 01:06, 21 July 2019 (UTC)

@Jura1: Where can I ask for such constraint to create? Xaris333 (talk) 18:11, 22 July 2019 (UTC)

I did so in the Phabricator. --Matěj Suchánek (talk) 13:00, 28 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 13:00, 28 July 2019 (UTC)

wbEntity config value to be dropped on July 24th

Hello all,

This message is important for people who still use the config value mw.config.get( 'wbEntity') in their tools or scripts. This value, deprecated for two years, will be completely dropped on July 24th.

The config value mw.config.get( 'wbEntity') has been deprecated in order to improve the page load time, especially on large Items. Currently, a significant proportion of the HTML on every entity page is generated by this value. Dropping wbEntity completely will make the first paint (time needed to load the page) faster, and will better utilize the server and client cache.

If your code is still using mw.config.get( 'wbEntity'), you can replace it by the hook wikibase.entityPage.entityLoaded (see an example here).

On Wednesday, July 24th, we will drop mw.config.get( 'wbEntity') from Wikidata’s codebase and calling the value will result in an error.

If you have any questions or need help, feel free to leave a comment under the related task.

Thanks for your understanding, Lea Lacroix (WMDE) (talk) 08:44, 22 July 2019 (UTC)

quick search for the about 50 user scripts that will break. Multichill (talk) 19:18, 22 July 2019 (UTC)

Wikidata Shape Expressions Inference tool is working again

Wikidata Shape Expressions Inference (documentation) is a tool I wrote (as part of my master’s thesis, Schema Inference on Wikidata (Q57385555)) that takes a SPARQL query returning a limited set of exemplary items and infers a ShEx (Q29377880) schema from them. I’ve never properly announced it here – the original announcement to WikiProject chat was linked in the weekly summary, and Léa mentioned it below the Shape Expressions on Wikidata announcement, but by that point it was already broken and then I had other things to do for a while. I’ve finally fixed it now, which seems as good a time as any to properly announce it here :)

As detailed in my thesis, the schemas can’t be used directly for validation (they’re so large that every validator I tried would crash, though at least one new implementation has been announced since then, which I haven’t tried yet), but hopefully they can still be useful, by highlighting problems in the source data (“why does the schema say that given name (Q202444) is a valid class for child (P40)?”) or as a starting point for more useful schemas that you manually extract from the automatically infererred ones.

Please take a look and let me know if it works for you! --Lucas Werkmeister (talk) 21:21, 22 July 2019 (UTC)

Strange French Wikiquote item

coercion (Q24248421) came up as part of WikiProject Random and I've been unable to figure out what exactly it represents from the page history. It seems that the French Wikiquote does not have a Wikipedia:Embassy (Q1197883) page, so I figure the best I can do is share here for any French speaking Wikidata users to take a look. --SilentSpike (talk) 09:43, 28 July 2019 (UTC)

The Wikidata title ("constraint") was an Englsh false friend. This page should have been connected to coercion (Q325980). I've merged them. Circeus (talk) 16:59, 28 July 2019 (UTC)
Perfect, thank you! --SilentSpike (talk) 20:44, 28 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --SilentSpike (talk) 20:44, 28 July 2019 (UTC)

Need help fixing template transclusion issue

It's my first time really working with templates/transclusion in any meaningful way and I've run into an unexpected issue. If you check out the properties section of Wikidata:WikiProject_Random/Generation the table has an extra column compared to Template:Property_table which is breaking the links. I can only assume it's something to do with a pipe character resolving from Template:P - which I'm not sure how to fix. Any help is appreciated. --SilentSpike (talk) 15:02, 28 July 2019 (UTC)

  • I've just figured out it breaks when there are newlines between the template parameters. Ideally I'd like to fix the template so that it handles those gracefully rather than require it be used inline. --SilentSpike (talk) 15:47, 28 July 2019 (UTC)
    • Seems I've managed to solve my own problem by using Template:trim to remove excess whitespace around the passed through parameters of Template:P. --SilentSpike (talk) 22:20, 28 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --SilentSpike (talk) 22:20, 28 July 2019 (UTC)

Torture type, Confession dates/locations and the Kartographer extension

Hi, further to our ongoing work taking information from the Survey of Scottish Witchcraft database, we are looking at how to model the types of torture endured by the accused witches in this period (1563 to 1736) and how this could be incorporated into Wikidata. The information on the torture type is described as "this field records the types of torture used in witchcraft investigations. Torture was the application of measured pain in order to coerce a person to provide information. None of these tortures seem to have been technically judicial torture." As previous Project Chat discussions revolved around whether the torture was judicial or extrajudicial I'm wondering where we could go from here to model this information? A new property perhaps? In addition we are looking at different methods of visualisation the geographical elements from the database now that we have added 3,100 places of residence to Wikidata for these accused witches along with 543 places of detention. Simon Cobb has suggested the Kartographer extension might be one way to go. Does anyone have experience of working with this extension or could maybe suggest where we could find out more about it and how to use it? Very best, Stinglehammer (talk) 13:59, 15 July 2019 (UTC)

  • We do have penalty (P1596) but it seems like that property isn't what you are looking for. It might be that we do need a new property. Before making a proposal it would be valuable to look at how this kind of information is modeled out there in the world to find a good solution that works beyond the needs of your particular project. It's worth investigating whether organizations such as Amnesty International have a standardized way to record this information. ChristianKl❫ 16:45, 15 July 2019 (UTC)
    • Ok, thanks. Have raised with Amnesty International as to whether they think 'manner of torture' would be an appropriate Property name given what they record. Will let you know their response.Stinglehammer (talk) 14:13, 16 July 2019 (UTC)
      • Hi ChristianKl, have had a reply from Amnesty as follows:

We would normally categorise torture in line with UN descriptions - how much detail we record and/or make public depends on the circumstances and individual cases. The facts are sometimes stated when we discuss the experiences of individuals such as Moses Akatugba who, while imprisoned in Nigeria, experienced torture such as having his finger nails pulled out by police. Here are a couple of overview pages including official torture thresholds/descriptions:[28][29]. We are calling for clearer definitions of torture [30]. Agree it’s important to highlight the level of suffering someone experiences due to torture but we may not always describe explicitly due to concerns about who can access the information, is it exploitative or triggering etc. This might not be what you were after but do give me a shout if I can help with anything else.

  • My sense is that we do need a new property to cover 'method of torture' or 'manner of torture' as it will be applicable in historic and modern instances and many of the values will be ones that already exist or be useful if they do need creating. Thoughts? Stinglehammer (talk) 11:47, 22 July 2019 (UTC)
Given that the linked document speaks of "Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment" it seems to me that we need a hypernym that not only encompasses torture but also "other treatment and punishment". I do agree that it would be good to have a property proposal. We can also have the discussion about what might be a good term (e.g. the hypernym) on that proposal page. ChristianKl❫ 13:08, 22 July 2019 (UTC)
@ChristianKl: Property proposal created here. We do also have rich information on date of confession and location of confession. Is it worth considering properties for place of confession and date of confession? Stinglehammer (talk) 10:45, 23 July 2019 (UTC)

Representation of Visa Policies: Attempt 2

I'm not sure what the normal etiquette is for this, but since my last attempt didn't draw any discussion, I'm going to re-post my request for an explicit visa policy (Q16725895) representation. You can find my more detailed description of the problem here:

Basically, I would like some feedback from more experienced Wikidata users on which of the three representations listed there would integrate best with the rest of Wikidata. It seems to clearly be one of the first two, but I don't know enough about Wikidata norms to say which. --Tga.D (talk) 22:03, 23 July 2019 (UTC)

For one, it's not a simple graph; it'll be a graph with multiple edges per node pair, since visa policies are typically conditioned on things like criminal history. The way these multiple (directed) edges are disambiguated on Wikidata is typically through the use of qualifying statements. For the first proposal: it's not considered good practice to create items for every "edge" as that would lead to items and would be unscalable. Similarly, since the edges are directed, the third option doesn't seem to make sense to me. The second option seems best but even it needs refinement in my opinion: I don't really see the need to link another country's visa policy when discussing visa requirements for travel in the other direction. Visa policies do not partition evenly into a small number of subtypes so we should not try to summarize every visa policy "edge" that way. Instead, we should use a new statement for each type of visa entry, and for every different type of requirement. Saudi Arabia, for example, has different requirements for tourist and Hajj visas, and within each type of visa, differing requirements on a per-country basis.--Jasper Deng (talk) 22:13, 23 July 2019 (UTC)

Meta question

Is this always so? You're new to Wikidata, concentrate on a project topic, find a problem, come here for clarification, and then see that there are much more pressing, elementary problems on which you should work? --SCIdude (talk) 06:46, 27 July 2019 (UTC)

  • To some extent yes I have had a similar experience. Alternatively you encounter a modelling problem and there isn't enough interest to have a discussion and solve it (or perhaps it's that not enough users feel confident contributing to the discussion). That said, I think you should feel free to work on whatever takes your interest. What problem did you encounter? --SilentSpike (talk) 08:03, 27 July 2019 (UTC)
    • I found that protein objects were given nonsensical aliases by several bots (e.g. this double title or the alias "Listeria gene" in the same object or the redundant 1), and I was not sure if trying to stop the bot or straight batch re-editing was the best strategy. --SCIdude (talk) 13:40, 27 July 2019 (UTC)
  • when I don't feel like sorting out questions myself, I tend to deposit them on Wikidata:Interwiki_conflicts or Wikidata:Bot_requests. If it's a problem with a bot edit, you could try the bot operator, some are responsive, others prefer to be blocked by an administrator first. --- Jura 13:53, 27 July 2019 (UTC)
There used to a problem with the gene aliases, see User_talk:MicrobeBot#Alias. Maybe we didn't cleanup all old ones. From the sample given, I'd couldn't tell myself if it's problematic. Description in Indonesian seems fine. --- Jura 14:00, 27 July 2019 (UTC)
I added a Bot request. --SCIdude (talk) 14:19, 27 July 2019 (UTC)
I'm surprised you don't see the double title as problem. --SCIdude (talk) 14:22, 27 July 2019 (UTC)
Not sure why I was focused on the alias .. yes, it's odd. Definitely something for the bot operator to look into. Even if someone else eventually fixes it, we don't want the bot to add more of it. --- Jura 14:24, 27 July 2019 (UTC)
OK I added a question on the bot talk page. Probably I'll look into documenting all the mess too. One code bug, and people being busy for hours. This is not effective, nor sustainable. --SCIdude (talk) 15:21, 27 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. --SCIdude (talk) 07:05, 29 July 2019 (UTC)

phenylbiguanide duplicate?

phenylbiguanide (Q25100298) and phenylbiguanide (Q27071250) look like duplicates, but could someone familiar with this verify before merging? Seen at Wikidata:Database_reports/Constraint_violations/P117#"Unique_value"_violations. 18:59, 27 July 2019 (UTC)

The external identifiers on the two items are the same, so I went ahead and merged them. Mahir256 (talk) 23:08, 27 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Matěj Suchánek (talk) 07:29, 29 July 2019 (UTC)

Relevance Criteria

Heyho, Newbie from eastern germany here. Are there any guidelines regarding possible relevance criteria for being item of Wikidata? I also noticed that there are lots of duplicate items, are there possibilities for (speedy) deletions? Greets, Arjoopy (talk) 20:53, 28 July 2019 (UTC)

@Arjoopy: Wikidata:Notability? And duplicate items should be merged. Nomen ad hoc (talk) 22:01, 28 July 2019 (UTC).
Thx for your quick response, I'll try to get into it. Arjoopy (talk) 22:11, 28 July 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. - Premeditated (talk) 22:27, 29 July 2019 (UTC)

Duplicated values for the same property

Can anyone explain why it is possible to have duplicated values for the same property? [31] [32] -- Meisam (talk) 14:47, 21 July 2019 (UTC)

@Meisam: Sometimes this can be appropriate; for example, a person may hold an office twice. Although the qualifiers should be different in such cases, because the Wikidata website submits complex changes as multiple edits, I don't think it would be feasible to prevent this without that behaviour (and similar behaviour in third-party editing software) being changed. Jc86035 (talk) 14:58, 21 July 2019 (UTC)
Because nothing is preventing you from doing that. We validate after the item is saved. In this case I would expect a single value constraint (Q19474404), but Eurohunter completely removed that constraint. Not sure why because it had only 4 violations so these could easily be listed as exceptions or it could be turned into a suggestion constraint (Q62026391). I would restore it. Multichill (talk) 15:03, 21 July 2019 (UTC)
  • There is a bot that cleans out some of the unneeded statements. --- Jura 15:58, 21 July 2019 (UTC)
Thank you all for the explanations. You are right! With the current interface (complex edits being split into multiple smaller ones) it is not possible to automatically prevent this. The bots have to do the clean-up for now. -- Meisam (talk) 19:10, 21 July 2019 (UTC)

Looks like I have done a mistake here. But it should be cleaned by bots (like KrBot) who looks after duplicated values. I have fixed it and i'm working on a blacklist query to not add same ID on multiple sites. - Premeditated (talk) 08:46, 24 July 2019 (UTC)

Modifications via URL, batches not running

Hi, all: I posted some batches via URL. I received the correct JSON message with OK and the corresponding batch number.
Numbers are 16099...16105. The last one refers to wikidata, the others refer to wikidata-test.
None of them run. Why? Thx. Bargioni (talk) 15:10, 22 July 2019 (UTC)

It sounds to me like you are trying to be a bot without using the bot API or getting a bot approval. Is that the case? ChristianKl❫ 15:14, 22 July 2019 (UTC)
@ChristianKl: I assume he's referring to . Multichill (talk) 19:14, 22 July 2019 (UTC)
@ChristianKl: You are right. put me in a wrong direction: Use the token to submit a batch to QuickStatements from your own bot (please use POST): ./api.php?action=import&submit=1&foo&bar
Please give me the doc for writing a bot for atomic modifications. My idea is to keep items in sync with PUSC author id.
Automated bot tasked should only done by bot accounts that are approved via: provides a variety of tools for programmers of bots. The Python API would be an example: ChristianKl❫ 14:10, 25 July 2019 (UTC)

property related to fellow student (Q202402)

Hi! I could find the properties student of (P1066) , student notable student(s) of an individual (P802) as well as doctoral student doctoral student(s) of a professor (P185)..
Is there a property related to fellow student (Q202402)? Thanks in advance! Best regards
no bias — קיין אומוויסנדיק פּרעפֿערענצן — keyn umvisndik preferentsn talk contribs 19:13, 24 July 2019 (UTC)

As far as I know the property doesn't exist. You can still use significant person (P3342) with a qualifier. This is an effective approach to document important relationships in a person's life, which can be quite diverse (friend, confessor, psychoanalyst, colleague, business partner...). Alexander Doria (talk) 21:22, 24 July 2019 (UTC)
Thanks Alexander for the answer. What I wanted to do is adding the following relations:
Alex Jacobowitz (Q338336) has studied marimba in New York with notable colleagues Gordon Stout (Q1890630), John Beck and Leigh Howard Stevens (Q16185606). The information is from w:de: . Could you please make / add the relation as a further sample for similar Wikidata linking. Please see also Reasonator for Alex Jacobowitz. Thank you very much in advance! Greetings from Munich, Germany
no bias — קיין אומוויסנדיק פּרעפֿערענצן — keyn umvisndik preferentsn talk contribs 08:46, 25 July 2019 (UTC)
✓ Done. I wanted to add the marimba as an instrument as well but significant person (P3342) does not support that (it was anyway a bit of a hack). Alexander Doria (talk) 10:48, 25 July 2019 (UTC)

Per-statement protection

Several other Wikimedia projects are (validly) concerned about the reliability of information on this project. There is much information on here that should not be changed, such as people's birthdates, and yet the only way to protect such statements is to protect the whole item. For some highly-visible items, this should not be the case. I'm hoping I'm not the only one in favor of allowing protection at this level of granularity.--Jasper Deng (talk) 00:38, 12 July 2019 (UTC)

The issue just seems to be that the development team does not consider this a priority (I know I've informally raised this to them at least once). --Rschen7754 00:42, 12 July 2019 (UTC)
Birthdates are not a good example. There are still many errors, partly caused by the fact that the method of inserting julian/gregorian dates changed on the run, many others just because the sources don't indicate which calendar they use. The possibility to correct a date, to add another value (with a reliable source, of course) to add a source or to change rank should stay open.--Shlomo (talk) 05:53, 12 July 2019 (UTC)
As for the concerns about reliability: Protecting an item or a statement doesn't improve reliability, it just increases stability. If you protect a bad-sourced or misleading statement, it would more probably stay bad-sourced or misleading, because you'd reduce the set of people who are able and willing to do something to improve it. The reliability problem relates more to where the data come from. A nonnegligible (to put it mildly) part of data comes via robotic additing from unreliable and/or unindicated sources. Any perceptible progress in this issue would have much more impact on other WM projects perception of Wikidata than making editing more difficult for newcomers and occasional editors.--Shlomo (talk) 05:53, 12 July 2019 (UTC)
@Shlomo: Your criticism of my birthdates example is simply a red herring (i.e. irrelevant); the point being made still stands. Do not confuse this with blanket semi-protection, which there was somewhat of a consensus for (see the recent RfC), but which I opposed. Presumably, the per-statement protection would be applied after multiple instances of the correct value (as deemed by consensus and sourcing, not just any value) of that statement being changed to the wrong value.--Jasper Deng (talk) 06:45, 12 July 2019 (UTC)
This is a great idea which will help indeed to solve many issues, however, I am afraid, the bottleneck are the developers. We can consider trying to make it to the next year Community wishlist, but this would require a serious lobbying work on the projects.--Ymblanter (talk) 07:01, 12 July 2019 (UTC)
See phab. --Jklamo (talk) 08:44, 12 July 2019 (UTC)
  • It would be good to see a usecase for this. The birthday sample is somewhat suboptimal: either there are references for various dates and they all should be added or there additional references for the same date and they could also be added. Maybe a filter disallowing changes to the date would be the better solution. --- Jura 11:56, 13 July 2019 (UTC)
The example that popped to my mind is the claim that hydrogen (Q556) has atomic number (P1086) 1, which is unambiguously provably to be absolutely right and which is never going to change. (And similarly for the atomic numbers of the other ~117 elements, of course.)
There are a lot of external ID numbers which I think fall into somewhat the same category. {Wikipedia has some mechanism for tagging those as "verified", although I've never dug into the details.) —Scs (talk) 23:54, 13 July 2019 (UTC)
@Jura1: You are missing the entire point of my example, which is to protect the birthdate statement after we have referenced it and solidified it to the correct, reliable, value. Creating a myriad of filters would not be scalable and would bog down everyone's editing.--Jasper Deng (talk) 07:26, 15 July 2019 (UTC)
Provide a better sample then. Even if a statement has a reference, it can use better references. --- Jura 14:51, 18 July 2019 (UTC)
  • It would be useful to have the ability to give per-statement protection but it costs development resources. Given that statement-signing is part of the yearly roadmap for Wikidata I however don't think that it's best to push for per-statement protection right now. Let's see how statement-signing works first and afterwards we might discuss about the best way to do protection. Maybe we end up with some way where signed statements by specific sources are semi-protected. ChristianKl❫ 07:52, 15 July 2019 (UTC)
    • No amount of statement signing would obviate the clear use for per-statement protection at Q15925869, where I have had to protect the whole item just because some spambots are targeting one particular statement.--Jasper Deng (talk) 09:08, 16 July 2019 (UTC)

Somewhat related, it would be better if we can do per label/desc/alias language-protection: See Sea of Japan (Q27092)'s history for example. — regards, Revi 14:24, 19 July 2019 (UTC)

Classify sources --> protect statements or flag when mismatch in Wikidata

I think Wikidata as a world wide project with > 300 languages needs some kind of source "trust ranking" I can see the following scenarios that is away of a "loose protection" or at least warning scenario

  1. We use Federated search to check with a "trusted" source
    1. a mismatch is flagged in Wikidata
    2. a mismatch is seen in Wikipedia infoboxes
    3. a mismatch can be documented as OK

See more thoughts on this subject link

- Salgo60 (talk) 10:56, 26 July 2019 (UTC)

Discrepancy between gradient properties

I notice that average gradient (P2198) has a constraint to be between values of 0 and 1 (enforcing a normalised form). Meanwhile maximum gradient (P2595) doesn't have this constraint and allows for units of percent (Q11229) or per mille (Q181011).

It seems to me that both properties should follow the second approach since a gradient can be specified in multiple ways. Perhaps dimensionless unit (Q1618549) could be added as another allowed value for explicit use of dimensionless normalised form (or is it better to use no value?). --SilentSpike (talk) 13:13, 17 July 2019 (UTC)

  • @Smalyshev (WMF): could we add a conversion on query server for both as we do for SI units? --- Jura 20:53, 17 July 2019 (UTC)
  • Would we need to define Q199 = 1 Q199 for every (Wikidata) unitless quantity? --- Jura 22:48, 19 July 2019 (UTC)
  • Unsure if relevant, but worth noting radian (Q33680) is also a dimensionless unit which technically gradient could be specified in. --SilentSpike (talk) 17:12, 26 July 2019 (UTC)

Given name (or surname) populator?

I created Jiaju (Q65925324). Now I wanna add it to all special:search/家駒. Is there a handy tool/script?--Roy17 (talk) 13:54, 24 July 2019 (UTC)

Personally I'm not in favor of creating items for Chinese given names (there're millions of them with minimal additional information), but you can try this (make sure you have checked all entries before edit).--GZWDer (talk) 18:18, 24 July 2019 (UTC)
We have a lot more geonames material of more dubious quality. Circeus (talk) 00:39, 25 July 2019 (UTC)
@GZWDer: I still share the exact same thought with you. Given name should not be an item but a string. Anyway, other users disagree: Wikidata:Project_chat/Archive/2019/05#given_names. So be it. Many thanks for the link!--Roy17 (talk) 13:19, 25 July 2019 (UTC)

funny petscan result

The link above finds Mutsumi Ishii (Q11585042) even though it does not contain "家駒". Instead, its description ends in 家 and its alias starts with 駒.--Roy17 (talk) 13:32, 25 July 2019 (UTC)

  • You might want to filter more, e.g. ( including haswbstatement:P31=Q5 -haswbstatement:P735 in search, and applying some regex filter in the output tab (maybe ^.*家駒$) --- Jura 12:46, 26 July 2019 (UTC)

Unit using a redirect

Hi. While entering data for the new Wolfram Language unit code (P7007) I ran various consistency checks and allowed myself to fix a few unit-related data points. One fix was merging Psf (Q39468084) into Psf (Q11830636). What happens with values for the unit-part of a quantity that is a redirect? Is there some automatic process to replace them with the target of the redirect? So far this does not seem to have happened: The following query shows (I think) that this issue currently exists.

select * where {
  ?x wikibase:quantityUnit / owl:sameAs [].

Try it!

Toni 001 (talk) 17:53, 25 July 2019 (UTC)

I fixed it, as there is no bot which resolves redirects in units to my knowledge. —MisterSynergy (talk) 18:12, 25 July 2019 (UTC)
Thanks. I would have done that myself, too, but wanted to know whether this is something to look out for in the future. Toni 001 (talk) 18:18, 25 July 2019 (UTC)
I believe my bot still does it but as usually it waits some time. --Matěj Suchánek (talk) 10:36, 26 July 2019 (UTC)
Well, I would not be surprised if it didn't work on units as that is a rather remote place to expect these problems. There was a similar request in May, two months after the redirect in question had been created [33]. --MisterSynergy (talk) 12:40, 26 July 2019 (UTC)
I didn't understand why a redirect is bad in this case. Are they a problem in general? I tried the query above and there is still one hit, but it's a special page so perhaps to be ignored in the query: [34] Jagulin (talk) 14:05, 26 July 2019 (UTC)
Most people writing queries don't allow for them. This probably applies to every sample query and automatic unit conversion on Wikidata Query Service.--- Jura 14:10, 26 July 2019 (UTC)
Yes, that was the motivation for my question: One could insert owl:sameAs in various places when writing queries just to follow redirects, but this might or might not have performance implications and also clutters the query. Toni 001 (talk) 17:14, 26 July 2019 (UTC)

Enquête auprès de la communauté

Hello les wikidatiens !

Depuis plusieurs mois, les bénévoles des groupes de travail stratégiques réfléchissent à comment améliorer le fonctionnement de nos communautés (online et offline) et du mouvement Wikimédia en général. Pour recueillir vos avis à ce sujet, ils ont conçu un sondage qu'ils aimeraient vous soumettre.

9 thématiques sont abordées (par exemple Santé communautaire, Produits et technologies, Diversité, Plaidoyer...). Vous pouvez choisir de répondre aux questions de l'une ou plusieurs de ces thématiques.

Pour répondre à l'enquête, c'est ici :

(Je vous encourage vivement à y participer, c'est le moment de proposer vos idées pour améliorer tout ce qui ne vous convient pas ! - Et N'hésitez pas à le diffuser autour de vous.)

DRanville (WMF) (talk) 08:25, 26 July 2019 (UTC)

@DRanville (WMF): Le bistro francophone de Wikidata est ici ;-) — Envlh (talk) 18:50, 26 July 2019 (UTC)

Community Survey

Hello dear inhabitants of Wikidata ;)

For months, volunteers from working groups have been thinking about how to improve how our communities work(online and offline) and how the Wikimedia Movement works in general. Now they want to know what you think, so they designed a survey they would like you to complete.

The survey has 9 thematic areas (like Community Health, Products and Technology, Diversity, Advocacy...). You can choose to answer the questions of one or several of these thematic areas.

The survey is accessible here in 7 languages :

Your participation is important: it is a great opportunity to suggest ideas and say how you would like our movement to evolve!

Please also share this survey around you.

DRanville (WMF) (talk) 08:25, 26 July 2019 (UTC)

A bot should add [en] labels based on commons cat names

E.g. Huang Baojian's Stone House (Q11177331), Category:Painters in Edo period (Q10110053)...--Roy17 (talk) 20:42, 25 July 2019 (UTC)

@Roy17: I have code to do that, but stopped it due to User_talk:Mike_Peel#English_labels_from_Commons_categories_not_in_english. I can resume it again if that's desired. Thanks. Mike Peel (talk) 04:57, 26 July 2019 (UTC)
Yes, please do. I don't see why contributors wouldn't just edit the label if the default one isn't as good as they'd like it to be. --- Jura 05:20, 26 July 2019 (UTC)
OK, I'm going to start running it again via Pi bot (talkcontribslogs), in batch mode for now but I'll aim to run it daily in the future. Any problems, please shout. Pinging @Hjart: for awareness. Thanks. Mike Peel (talk) 16:22, 26 July 2019 (UTC)
Ok, I'll just remind you that lots of Commons Categories are named by native names/descriptions (see e.g. commons:Category:Lighthouses in France by name), so that this will effectively often be futile. --Hjart (talk) 09:06, 27 July 2019 (UTC)

Links for units

Hi. I recently fixed a conversion to standard unit (P2442) for kilogram (Q11570): From "grain(s)" to "grain(s)". I spotted this issue using programmatic consistency checks, but it is somewhat hard to discover in the user interface. It might help if units were links or had a tooltip. Also, when opening the edit dialog it is not possible to see which unit item has been selected: Only its label is shown. Toni 001 (talk) 10:35, 26 July 2019 (UTC)

  • The unit interface could probably be improved.
I think we can delete the conversion to standard unit (P2442) statements on kilogram (Q11570). Most other properties have checks on units implemented. --- Jura 13:28, 26 July 2019 (UTC)
In general, the Wikidata interface still leaves a lot to be desired. I would sum it up as "it's too slow and doesn't surface the right things unless specifically asked to". Jc86035 (talk) 10:56, 27 July 2019 (UTC)

Are these two edits normal?

special:diff/169878974 special:diff/169875877 each merely added a sitelink, but the change was +7k bytes.--Roy17 (talk) 17:32, 26 July 2019 (UTC)

Yes, these are normal. See Help:FAQ#Editing, #13. --Matěj Suchánek (talk) 10:21, 27 July 2019 (UTC)

Q65004088 and Q4987532

Q65004088 and Q4987532 seem to be about the same footballer. I created Q65004088 a few weeks ago aware of the existence of Q4987532. Now that I found it, I decided to merge both, but the problem is that the footballer's name is different in the two places. In Q65004088 (that I created) she is called "Barbara Lorsheyd". In Q4987532 she is called "Barbara Lorsheijd". Several sources call her "Barbara Lorsheyd" including her profile at the UEFA website, her profile at the World Football and even an article in Dutch. However, at least two sources: 1 and 2 calls her "Barbara Lorsheijd". What to do now? SirEdimon (talk) 08:51, 27 July 2019 (UTC)

As an example the name of this islet can be spelled in 2 slightly different ways. Just decide on which of your footballers names is the primary and enter the other name as an alias.--Hjart (talk) 12:37, 27 July 2019 (UTC)

Changes to Wikia/Fandom URL structure for non-English wikis

At some point, Fandom changed their URL structure for non-English wikis from $lang.$ to $$lang. The interwiki fixer at still works, and I updated the formatter URL for Fandom wiki ID (P4073) to use it (this also fixes any links to subdomains), but it's not clear if this will work in the future. Should the values for Fandom wiki ID (P4073) and Fandom article ID (P6262) be updated in any way? Jc86035 (talk) 10:46, 27 July 2019 (UTC)


Hello. We are not adding that a person has another person as and grandfather, if there is already father (P22) in the item? Because we are not allowed to use paternal grandfather (Q19682162) with type of kinship (P1039) for relative (P1038). See Q65942054#P1038 Xaris333 (talk) 04:33, 26 July 2019 (UTC)

Then why I am getting a constraint? Xaris333 (talk) 17:32, 26 July 2019 (UTC)

  • You read the proposal and the constraint matches that. So why should we still be adding it? --- Jura 17:42, 26 July 2019 (UTC)

How I suppose to add the information? And why the opposite is allowed? Q58456663#P1038 Xaris333 (talk) 19:21, 26 July 2019 (UTC)

@Xaris333:. The general rule is that we give simple relationships (father, child, etc) using the dedicated properties, and if we can trace a line through those on different items, we don't need relative (P1038). In this case, because Keti Clerides (Q65942054) has father (P22) Glafcos Clerides (Q317648), which has father (P22) Ioannis Clerides (Q58456663), we don't need to link Keti Clerides (Q65942054) and Ioannis Clerides (Q58456663) directly - by adding the parent/child links to all of the items in the chain, we have all he information we need.
relative (P1038) is to be used where we can't work out a middle item - for example, maybe we know someone was someone else's brother-in-law, but we don't know anything about the sister - or where there's a direct connection between the two items, but we don't have a suitable property (like "adopted sibling").
I don't know why grandchild isn't in the constraint, but it might be that someone just forgot to add it. Andrew Gray (talk) 21:53, 26 July 2019 (UTC)
Fully agree w/ Andrew, Nomen ad hoc (talk) 21:30, 28 July 2019 (UTC).

Display of "different from"

Has anyone an idea how to improve the display of different from (P1889)?

  1. Gottfried W. Locher
  2. Gottfried Wilhelm Locher
  3. Gottfried Locher
  4. Gottfried Wilhelm Locher
  5. Gottfried Wilhelm Locher

I need at least one more information (like year of birth or occupation) unless I want to click through all possible namesakes. --Kolja21 (talk) 00:02, 27 July 2019 (UTC)

  • Do we have an optional tool on here that lets you see description & a.k.a. by hovering? - Jmabel (talk) 04:02, 27 July 2019 (UTC)
    • I don't think we have one, but that would be a great idea - like the WP popups. Andrew Gray (talk) 07:48, 27 July 2019 (UTC)
  • "PopupsFix: Makes the Popups extension also work with items and properties." in the preferences should work. --Matěj Suchánek (talk) 10:23, 27 July 2019 (UTC)
I might be an idiot (I usually am) but ... I can see how to enable the fix in preferences, but I can't find how to turn on popups proper. mw:Page_Previews suggests it should be on Preferences > Appearance, but I don't have a popups option there. Andrew Gray (talk) 11:47, 28 July 2019 (UTC)
This time you were right. When I made that suggestion, I didn't realize that...
  • For some reason, I haven't had it enabled. So I did it now. When I realized it doesn't work itself I started wondering why. It seems the gadget only loads when you have Page Prewies enabled... which nobody can because we don't have it installed (it must have been uninstalled but I don't now know when).
  • Interestingly, it's enabled for all users by default. That's not good.
  • There is also "Descriptions: Show the description of items and properties when hovering them.". I tried to enable this one. It doesn't work, for the same reason as the other one.
Another source for Phabricator tasks. --Matěj Suchánek (talk) 12:50, 28 July 2019 (UTC)
@Matěj Suchánek: Looks like I did exactly the same as you :-). Found out it the fix enabled for me, got confused, tried turning on descriptions, found that didn't work either ... Andrew Gray (talk) 12:59, 28 July 2019 (UTC)
To test, I had activated both *Navigation popups" and "PopupsFix". It works partially: some error message indicates the use of deprecated code. --- Jura 13:06, 28 July 2019 (UTC)

SE SE Section 30 Mine (in Montana, USA)

Thanks to Wikidata:WikiProject Random, I ended up on SE SE Section 30 Mine (Q49447247). I wasn't really successful trying to figure out the naming convention. I added a map query on Talk:Q49447247, but maybe it's just the coordinates that are wrong. Any suggestions? ----- Jura 20:04, 28 July 2019 (UTC)

@Jura1: I think it's a w:Section (United States land surveying) reference format - the numbered section was the main division of land, a quarter of the section would be "SE Section 30", and a quarter of that quarter - the standard unit for selling land - would be SE SE Section 30 (or NW SE, etc). I haven't seen many places named like this so I guess they were feeling very unimaginative and there wasn't anything else locally to name it after! (Compare the tiny village of Section Thirty (Q7444673)...)
The problem is that Section 30 isn't quite enough to identify its location. The sections are numbered inside larger areas, so there will be a few different Section 30s within Silver Bow County (and that's even assuming the county label is correct). Poking around the Montana Cadastral data suggests that Section 30 of area 2S 8W is about in the right place for those coordinates, a little way east of Melrose. Nothing seems to be there any more, though. Andrew Gray (talk) 20:53, 28 July 2019 (UTC)

How to display names for the capitals of the Ottoman Empire?

Hi, guys! Due to complexities in the naming issues regarding the Ottoman capital and differences between Turkish and French/English, I started a discussion at Talk:Q12560. In a nutshell, the name Istanbul only became applied to the whole city after 1923, with the name applied in English and French after 1930. Before then Stamboul (French/English) or Istanbul (Turkish) referred to specifically the old city while Constantinople/Konstantiniyye referred to the entire city. What complicates it is that in Turkey today it is improper to use "Constantinople" to refer to the Ottoman capital even though historically that is what was done.

I would like to explore whether a custom display of the city names is possible, and making it different by language (displaying "Edirne" only in Turkish but "Adrianople (Edirne)" in French and English) is possible.

Thanks, WhisperToMe (talk) 09:34, 24 July 2019 (UTC)

@WhisperToMe: If this is an issue with what text is being displayed on the Commons category, and the infobox header should be different to the Wikidata label, I would suggest allowing the infobox header for each language to be customized in the template on Commons, with usage being tracked so that they can be easily be reviewed if someone wants to investigate the discrepancies. Possible parameter formats could be |header-fr=, |fr= or |header=fr:...;.... Jc86035 (talk) 10:08, 24 July 2019 (UTC)
Re: the header that sounds good to me! For contents inside (for example the entries in the heading "capitals") is there a way to make them display differently on the Commons? WhisperToMe (talk) 10:27, 24 July 2019 (UTC)
@WhisperToMe: It's possible in theory, but it could be difficult to code all of it while still making the template usable (you would have to make it possible to provide local values for each parameter in every language). It's also dependent on whether anyone's willing to do it. Jc86035 (talk) 04:27, 25 July 2019 (UTC)
Thanks for the answer! I wonder where I could/should make that request? Would it be in Lua? WhisperToMe (talk) 04:33, 25 July 2019 (UTC)
@WhisperToMe: Well, c:Template talk:Wikidata Infobox seems reasonably active, so I imagine you would ask there first? Jc86035 (talk) 08:21, 25 July 2019 (UTC)
@Jc86035: Ok! I asked at Commons:Template_talk:Wikidata_Infobox#Customizing_displays_of_contents_of_Wikidata_infoboxes_used_in_the_Commons and the response from User:Mike Peel was that language-specific stuff should be done on Wikidata, not on the Commons WhisperToMe (talk) 17:33, 25 July 2019 (UTC)
Yup, this seems to be a data problem, not a template problem. Thanks. Mike Peel (talk) 17:53, 25 July 2019 (UTC)
To elaborate a bit, @Jc86035: what you're suggesting is basically to reproduce Wikidata in template parameter code. That requires >200 template parameters per field, one for each language supported. Given that the infobox has circa 500 fields (= # of Wikidata items used), that quickly becomes >100,000 template parameters. It should be far simpler to model it better on Wikidata. Thanks. Mike Peel (talk) 05:10, 26 July 2019 (UTC)
(To make it even worse) You have a whole item in itself about one of the names of the city in Q2008228. IP (Hej!) 10:11, 24 July 2019 (UTC)
And all the enyclopedia entries are for "Miklagård"? I don't get this wikidata entry... WhisperToMe (talk) 10:27, 24 July 2019 (UTC)
Thought that Miklagård was referring to a fantasized Constantinople (a city in the viking tales) ? Just asking. Bouzinac (talk) 11:48, 24 July 2019 (UTC)
That some of them visited the city is beyond doubt. You find rune-graffiti inside Hagia Sofia. IP (Hej!) 12:11, 24 July 2019 (UTC)
@WhisperToMe:If it means anything I'd like to give own 2 cents, I think you're looking too deep into this. I mean while we're at it why not change Kraków (Q31487) capital of Polish–Lithuanian Commonwealth (Q172107) to Cracow or Kyiv (Q1899) capital of the Kyivska Rus' (Q1108445) to Kyiv as that was the historical spelling in English?
I think it's best practice in all cases to go with the modern, recognizable name for a given language (consulting something like ngram if it ever changes) and keep the rest as aliases. Or better yet maybe we should have a new property for historic names appear on the associated page instead of just an alias where it can be properly cited and everything? Themadprogramer (talk) 07:26, 29 July 2019 (UTC)

Q9868699 (Established in 338) and its siblings

One day I had the sudden urge to click random item to find something to edit. I happened to discover that there were a ton of category pages from Newari Wikipedia that follow the pattern below that had item entries designated for them here on Wikidata:

पुचः:ने॰सं॰ ३८८य् पलिस्था

Loosely translated as (based on the Bengali label, which was the only other translation available):

Category:Established in the year 388

Looking into the history, all of these Wikidata entries seem to have been generated by a bot, as even this one for Category:Established in 338 (Nepal Sambat) (Q9868699) is actually empty, more like a default template, rather than a non-existent page. It's probably not a good decision to enforce rigorous standardization on the Wikipedias but that being said we ought to make a choice for how they (or similar auto-generated pages from Wikipedias in other languages) should be handled:

  1. Make sure all of these pages are marked as instances of Wikimedia category (Q4167836) and label them for at least a couple more languages since Newari has around a million or so speakers it would be a good idea to help non-speakers understand what these are.
  2. Link similar year-based categorization items generated from other wikis
  3. Delete all items which correspond to an empty page
  4. Delete all such items and communicate with the Newari community to make something more compatible (kind of as a last result, not really a preferable option)

Personally I think the 2nd would be the option so I'd love it if anyone could tell us if their Wiki's have anything similar.

Themadprogramer (talk) 06:48, 29 July 2019 (UTC)

I'm new here but I think you should make a Wikidata:Bot request. --SCIdude (talk) 07:03, 29 July 2019 (UTC)
Just saying : are year calendars in newari same than in Gregorian calendar (Q12138)? Bouzinac (talk) 13:31, 29 July 2019 (UTC)
  • @Themadprogramer, Jura1, Bouzinac, SCIdude: If there's one thing I regret doing the most on Wikidata, it's early on creating items for every such other empty category on newwiki. (My honest view is that newwiki is to Indian topics what cebwiki is to world geography at large.) The year to which the category titling this section refers is not a Gregorian calendar year but instead one in the Nepal Sambat (Q1268741). (There are similar empty category items making references to years in the Vikram Samvat (Q1370848) and the Buddhist calendar (Q370752). There is no such thing as a "Newari Wikipedia community"; the only member of said community lives in Los Angeles, not Kathmandu, and I have not been able to get a hold of him myself to ask about these categories. I suspect that you'll have a hard time doing this too; therefore, in the absence of such a point of contact with whom to discuss these items, the third or fourth of your given options (preferably the third) appears to be the best, both on newwiki and on Wikidata. Mahir256 (talk) 13:31, 29 July 2019 (UTC)
    • As it's a category item, I added "(Nepal Sambat)" to its label. Forget about merging then. --- Jura 13:37, 29 July 2019 (UTC)

Wikidata weekly summary #375

Inverse of P138

Hello. Do we have the inverse of named after (P138)? For example if a street take the name of a politician, we can have "Street item -> named after (P138) -> Politician item". Do we have a property to show "Politician -> his/her name given to -> Street item"? Something like that. Xaris333 (talk) 14:37, 29 July 2019 (UTC)

I don't think such property ever existed. Nomen ad hoc (talk) 14:55, 29 July 2019 (UTC).
Just an useless inverse property with will lead to heavy items which will become impossible to download (some persons have dozen or even hundred streets/places with their name). The only reason of the multiplication of this kind of inverse properties seems that people don't want to use lua templates with proper extraction function. Or is there any other reason ? Snipre (talk) 15:43, 29 July 2019 (UTC)
Can you link such a LUA template or is it just something you made up now? --- Jura 15:48, 29 July 2019 (UTC)
@Jura1: Why do we have to develop lua template if people are not interested to use them ? Most WP contributor want to use wiki template and add some extraction commands in that code. Snipre (talk) 19:10, 29 July 2019 (UTC)
But if you are really interested, you can have a look at Module:Wikidata_list, the data extraction function exists, there is some work to handle the output in template code. Snipre (talk) 19:30, 29 July 2019 (UTC)
I don't think this template does what Xaris was looking for. "people don't want to use lua templates with proper extraction function" might suggest that such a template actually exist. --- Jura 19:37, 29 July 2019 (UTC)
Looks like you edited your comment while I was replying to it [35], replacing LUA with with Listeria ;) --- Jura 21:22, 29 July 2019 (UTC)
I don't think Listeria involves any LUA whatsoever. What did you lead you to believe that? --- Jura 21:27, 29 July 2019 (UTC)
I understood that Xaris333 wanted to perform a query and as I suppose he knows about Wikidata Query Service, the only reason to ask how doing a query is to find the way to present the results of it in table or infobox. The template I proposed indicates how to perform a query from WP and to generate a table with the results. But this template was banned from some WPs. Snipre (talk) 20:51, 29 July 2019 (UTC)
Use this query (streets named after Wilson Churchill):
SELECT ?street ?streetLabel WHERE {
  ?street wdt:P31 wd:Q79007 ;
          wdt:P138 wd:Q8016.
   SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
Try it!
Snipre (talk) 20:44, 29 July 2019 (UTC)


Now that every scientific publication is in Wikidata via ORCID, and editors are feverishly trying to disambiguate authors and aggregate their publications, is there any plan to upload every patent and do the same thing? Patents have extra data to aid in disambiguation. --RAN (talk) 19:31, 29 July 2019 (UTC)

"Every scientific publication"? Seriously...? Nomen ad hoc (talk) 19:35, 29 July 2019 (UTC).
@Richard Arthur Norton (1958- ): See Wikidata:WikiProject Patents. ArthurPSmith (talk) 20:07, 29 July 2019 (UTC)

And what about every book? --Scs (talk) 01:28, 30 July 2019 (UTC)

Wikidata read-only period on July 30 at roughly 05:00 AM UTC

This was already mentioned in Wikidata weekly summary #372, but I'd just like to highlight that there's a read-only period for Wikidata on July 30. We need to replace the primary database master: the old one is old and out of warranty, and we need a new one of continue the wb_terms redesign.

This means that the wiki can be read but not edited. The read-only window is 05:00AM UTC to 05:30 AM UTC, but if everything goes well the actual period of no editing will be shorter. When editing on Wikidata, it will be apparent that the changes are not working; you can read phab:T227063#5323866 to follow the discussion about how to best solve the problem when someone moves a page on another Wikimedia wiki.

If you've got any questions, please ping me. If I can't answer, I'll try to find someone who can. /Johan (WMF) (talk) 09:05, 15 July 2019 (UTC)

2019-07-30 05:00~05:30 UTC in your timezone — regards, Revi 12:03, 15 July 2019 (UTC)
This should stay till the readonly window is complete. — regards, Revi 05:30, 30 July 2019 (UTC)

How best to link the subject of an obituary at Wikisource to Wikidata;_Henry_Clay_Nelson and Henry Clay Nelson (Q65965909). If no other way I will just add as "described at url". In the past when I tried making a category for a person in Wikisource it was deleted. I was told only authors get categories. --RAN (talk) 01:34, 29 July 2019 (UTC)

Create a Wikidata item for the obituary copy at Wikisource. The person is the subject of the obituary, and the person is described by that source. --EncycloPetey (talk) 01:36, 29 July 2019 (UTC)
Yup. You'll want main subject (P921). Circeus (talk) 04:07, 30 July 2019 (UTC)

Notifying participants when creating properties

I am worried about our convention to ping all participants of a property proposal when the property is created. When creating many properties it generates a lot of notifications for people who are active in this sphere. People might hold back from property proposals because they do not want to be bothered by these pings. For instance NMaia recently supported a proposal and explicitly asked not to be pinged. Would it not make more sense to just ping the proposer? Isn't the weekly summary sufficient to stay updated about this process?

For now I am sticking to the existing convention but I have set up a mechanism for people to opt out of my property creation pings: User:Pintoch/property_creation/ignored_pings. − Pintoch (talk) 12:14, 24 July 2019 (UTC)

Yeah, I'd rather have this opt-in than opt-out. Maybe the MediaWiki thing that preloads the property proposal could have an additional text instructing participants to explicitly state if they want to be pinged. NMaia (talk) 12:49, 24 July 2019 (UTC)
Explicitly stating that on each proposal you comment on does not seem viable to me, it is going to be quite bureaucratic and heavy, and add to the cognitive burden of property creators (or the complexity of scripts to parse that information from proposal pages).
Isn't this what watchlists are for? If someone is interested in keeping track of the status of a proposal, cannot we expect them to just add it to their watchlist? − Pintoch (talk) 13:31, 24 July 2019 (UTC)
I don't mind the pings, I would definitely opt in. It's more noticeable than a watchlist item, and I find it useful to be aware of. I comment on a lot of property proposals and I can't say I find the notifications burdensome. ArthurPSmith (talk) 17:07, 24 July 2019 (UTC)
I value them a lot. My watchlist is very busy, but new properties really matter to me. I will find them anyway eventually, but I appreciate the (easy to dismiss) ping if I was interested enough to do the hard work of review the proposal. --99of9 (talk) 00:57, 25 July 2019 (UTC)
I like the pings. - PKM (talk) 03:24, 25 July 2019 (UTC)
I don't think that people hold back from voting because of pings. I am voting in proposals because I am interested in the outcome of the proposal (in some cases it can take months). I am not sure about the motivation of participants who are not interested in the outcome of a proposal.--Jklamo (talk) 08:04, 25 July 2019 (UTC)
Maybe make a template like: "Symbol support vote.svg Support and notify me". Send notification/ping to them and the property proposal user. - Premeditated (talk) 13:35, 25 July 2019 (UTC)
I see property creation as a collaborative process between the people who take part in a property discussion. The act of casting a vote comes with a certain amount of wanting to take responsibility for taking part in a property proposal. Removing the pings seems to encourage people to engage less deeply with the property proposal on which they are commenting which I don't see as going in the right direction. Especially, people who have the property creator right should frequently review the closing of proposal. That review process helps us having common standards when it comes to how proposals are closed. ChristianKl❫ 14:02, 25 July 2019 (UTC)

Ok, it looks like there is a pretty overwhelming consensus to keep these pings by default. Let us keep this convention then. Thank you all for your feedback! − Pintoch (talk) 16:24, 25 July 2019 (UTC)

@Premeditated: FWIW, I made this: {{Don't ping}}, which produces Font Awesome 5 solid bell-slash.svg Don't ping me. NMaia (talk) 13:04, 30 July 2019 (UTC)

Sample for everything

I'm sure we can find better samples. --- Jura 17:47, 24 July 2019 (UTC)

I don't see a problem here. I just completed missing examples (3 required). Most of examples come from property proposal because why not. Eurohunter (talk) 18:18, 24 July 2019 (UTC)
  • There is no requirement to have three on the property itself. Ideally, they don't all illustrate the same. --- Jura 18:49, 24 July 2019 (UTC)
  • @Jura1: While I wouldn't have gone through the work of adding all those examples myself I don't see a problem with it. It has even has advantages when people want to see a showcase item that shows how a bunch of properties are used. ChristianKl❫ 13:16, 30 July 2019 (UTC)

cleanup week?

Is there something like a cleanup week where bots/batches are prevented from creating/adding and fixing imports is encouraged (allowing fixing bots)? --SCIdude (talk) 15:36, 27 July 2019 (UTC)

Nope, cleanup is a 24x7 thing and never stops. Multichill (talk) 19:24, 27 July 2019 (UTC)
And careless imports are also a 24x7 thing which never stops, see above… More seriously, if you find issues with a bot, you should report it to get the bot fixed. If the bot owner is not cooperative, we can undo its actions or block it if necessary. − Pintoch (talk) 10:30, 28 July 2019 (UTC)
Lol! User:QuickStatementsBot has a hell of issues, but none of them is fixed. Steak (talk) 12:20, 28 July 2019 (UTC)
I regularly do cleanups. I have started using bots that run 24x7 that alert me about issues, but also do a lot of manual cleaning. For example, for autocreation of items based on Wikipedia pages. For chemistry, I found Wikipedia a major source of (chemical) problems in Wikidata. Tools like User:QuickStatementsBot help me solve these problems. --Egon Willighagen (talk) 07:31, 30 July 2019 (UTC)
re.chemistry, I wouldn't trust PubChem data, either... --SCIdude (talk) 07:53, 30 July 2019 (UTC)

To throw a stone..

With amazement I read a diatribe because the work on their specific subset of Wikidata is seen as difficult because of work on another subset. From my point of view I am quite happy when people work on subjects that have imho little merit like sports. However when they see it as a problem that others work on science, it follows that they fail to perceive that any and all people who are imported are linked to papers, to co-authors. When they did they would appreciate that the necessary enrichment of the associated items has been a process that is well under way.

Arguably sports is a niche of data that has as much right to be on Wikidata as any other part so I wont go there. I will point out that science and references is the key of why Wikipedia can be trusted. It has been indicated that all references need to find their way into Wikidata. So this nonsense of calling the data associated with this of no value is factually rather empty.

Wikidata is a project where people with notable data may expand Wikidata. There is a point to this. There is little point to many areas of interest including sports. So what is it that the stone throwers want to achieve? The deletion of data that cannot be argued to have intrinsic value as the notions of notability are to be reassessed?

Thanks, GerardM (talk) 11:18, 28 July 2019 (UTC)

Wikidata is a general purpose free knowledge base that got completely hijacked by the people using it as a scientific article dumping ground. Almost half of our items are an instance of scholarly article (Q13442814). This completely disrupted the balance. The integration of all these items with the rest of the knowledge base seems to be minimal. It's just a data island on it's own not making the knowledge base as a whole better. So yes, I and other people are critical of this. To restore balance, it would probably be best to fork it in a separate Wikibase instance and just link to it when we're in need of a reference. Multichill (talk) 13:15, 28 July 2019 (UTC)
"Dumping ground" is an accusation that is based on perception. My reality is that I seek specific subjects and scientists that are of interest, have a Wikipedia article, are in the news. Why? Because for one "Scholia" templates are increasingly found in Wikipedia articles. First of all what is included is notable by the current definitins, it serves a purpose. We know what papers are free to read and are seeking the best method to link the papers directly from a Scholia. So the objective of it all is very much that Wikipedia readers are enabled to read the sources themselves. With such a background, I the epithet "dumping ground" is problematic.
The notion of balance is wrong as well because balance in what way. What I have indicated before is that we are not using our technology to be smart about disambiguation. We do not use automated descriptions. This has two negative effects: the labels are hardly meaningful and mostly absent. When the notion of balance is introduced, lets kill off sports at the same time. It does not lead people to practice sport, its value is debatable. Thanks, GerardM (talk) 14:46, 28 July 2019 (UTC)
I'd just like to point out that the effort to import scholarly articles has also brought in many, many articles about clothing, dress history, and textiles that have lead to the creation of items about curators and dress historians through the author disambiguation tool. And some of these articles, in turn, have been used to reference other items in Wikidata. - PKM (talk) 20:21, 28 July 2019 (UTC)
A beautiful example of virtuous circle. Nomen ad hoc (talk) 21:59, 28 July 2019 (UTC).
How many items on Wikidata have at least one reference (via P248) to a scientific article? And how many scientific articles are used as references (via P248) on items? Exactly how much are we not taking advantage of the glut of scientific article items on Wikidata? Mahir256 (talk) 23:29, 28 July 2019 (UTC)
If you're going to talk about hijackers, then let them confront this accusation: User:Daniel Mietchen, User:Harej, to name a few. Mahir256 (talk) 23:29, 28 July 2019 (UTC)
With all due respect, this accusation says more about you. Hijacking of what? Wikidata is actually used and useful. Thanks to Scholia we now have an instrument that enables us to objectively assess the notability of scholars. The value of Wikidata is in its use, the scientific data is used, useful and I agree that this is a "mortal sin". Thanks, GerardM (talk) 05:37, 29 July 2019 (UTC)
Scholia is a very useful tool, that even got to the point where they managed to get the Sloan foundation to give them a grant to pay for more development. It's good for getting new users engaged with Wikidata if Scholia would have a separate Wikibase installation it would make it less likely that Wikidata profits from the new users that get engaged via Scholia. I'm not sure what this "balance" you are talking is about and why we should consider it to be valuable. The fact that half of our items are scientific articles doesn't cost us much. It would be a problem if half of vandalism fighting would deal with the items about scientific articles but that's far for the reality. ChristianKl❫ 12:05, 29 July 2019 (UTC)
@ChristianKl: in my opinion, the fact that Scholia got a grant is really irrelevant to this conversation - that does not give the Scholia project (or Wikicite in general) any special right on Wikidata. This attitude has been an issue since the beginning of Wikicite: having grants makes grantees feel entitled to bypass community rules. For instance cites work (P2860) was created during the Wikicite conference less than 3 hours after being proposed. Similarly, to import millions of publication items one could have expected to go through Wikidata:Requests_for_permissions/Bot - I do not remember seeing that (but I might be wrong).
The fact that half our items are scientific articles does cost us a lot, both technically and socially. We have issues with the Wikidata Query Service which is put under a lot of strain (by Wikicite edits but probably also by Wikicite queries), we have issues with terms storage in Wikibase (in good part because of the immense duplication of machine-generated descriptions on publication items), publication items and ORCID items flood the search results, making it harder to find other items (reminder: WMDE had to manually tweak the search profile to reduce the prominence of scientific articles in search). We have a statement from WMDE about the unsuitability of Wikibase for publications with hundreds (thousands?) of authors stored as individual statements. At the same time, the Wikicite corpus is still extremely patchy, so we would need a lot more items to make it usable. Is that not enough evidence? − Pintoch (talk) 12:57, 29 July 2019 (UTC)
I do agree that doublication of machine generated descriptions is a problem but I don't think it's limited to scientific articles. Having all those items might mean that it's more pressing to solve the issue by a clever description template system, but it's a problem we would also have without WikiCite activities.
Handling publications with hundreds of authors feels to me like a technical problem that could be solved by only showing the user 5 authors by default in the interface and only showing more when the user queries for more. It sounds to me like it's a question of development effort. Scholia having their own grant money could be helpful here if that ability to have papers with multiple authors is considered important for them.
I do agree, that it's unforuntate that WikiCite didn't went through the bot approval process. We don't have a formal policy that requires that. When thinking about the issue I advocated for batches being easy to undo. Batches that are easy to undo, are qualitatively different then bot edits that are by their nature often not be able to be undone via a single click. This allows for people to be bold with adding data but it still allows the community to find the consensus to undo edits that it doesn't like. If you think stopping WikiCite creates this many items is important, an RfC to find consensus would be a useful step forward.ChristianKl
Yeah, who cares about bot approval, we have QuickStatementsBot to bypass it entirely! Bot approval is only here to discourage newcomers from importing stuff. I wonder why we are not doing the same for properties by the way, I should create a self-service property creation bot and only leave Wikidata:Property proposal to the clueless newbies. I wonder who came up with this idea of "consensus" anyway… − Pintoch (talk) 15:58, 29 July 2019 (UTC)
"Seek consensus before editing" is not the general Wiki-way. "Be bold" is. We have approval processes for tasks where undoing the actions of someone who practices "be bold" is problematic. Solving the problems that come from badly modeled properties isn't as easy as pressing a button to undo a QuickStatement batch. I'm not opposed to creating a policy that actually requires approval for large QuickStatement batches but we would need to design a process that works in a timely fashion for that. ChristianKl❫ 16:21, 29 July 2019 (UTC)
Look, we are not talking about a few big QS batches - this import is split into hundreds (if not thousands) of relatively small batches, so it is clear that such a policy would not prevent anything. I don't think we need more policies here - we just need the basic decency of requesting community feedback before doubling the size of Wikidata with publications. That discussion could have happened anywhere on wiki. I am a big fan of the "be bold" policy, but this is another scale of boldness. Is it not common sense? − Pintoch (talk) 16:36, 29 July 2019 (UTC)
@Daniel Mietchen, Egon Willighagen, Fnielsen, GZWDer: Let's put this another way: if you can show me that there is a clear consensus both from the community and from WMDE that some set of publications can be imported (say Crossref), at a given rate (say 60 edits/sec), then I am happy to do this entire import for you. This will solve Scholia's coverage issues directly. If you are not able to show that this consensus exists, then try with a smaller set of publications or a slower rate. If this has already been discussed somewhere on this wiki, then apologies for the noise, just point me to that consensus. If you are not able to find such a consensus for any set of publications, then we should block SourceMD. It is just not okay to do such a large import in "who's-gonna-stop-me" mode. − Pintoch (talk) 17:38, 29 July 2019 (UTC)
I do consider your demand here reasonable given the millions of items involved. But I don't think the same should go for actual batches of 50,000 items that are also created by side-stepping the bot policy. ChristianKl❫ 06:59, 30 July 2019 (UTC)
I have created Wikidata:Requests for permissions/Bot/LargeDatasetBot to see how community feels in such an import.--GZWDer (talk) 22:06, 29 July 2019 (UTC)
I very strongly opposes separate Wikibase instance for this. One purpose of Wikidata is getting concepts linked (including other items and papers as source, papers and other works they cited, papers and their authors, etc.) making a separate Wikibase instance does not make it simplier. Instead "scholarly article" is a good use case of Wikidata.--GZWDer (talk) 15:33, 29 July 2019 (UTC)
In this comment I try to reply to comments by others. First, Scholia != Wikicite != Wikidata != me. Second, please accept that Wikicite and Wikidata are communities and that I cannot be held repsonsible for actions done by another community member. I will not comment on mass import on articles, as I have not done that, neither have I been involved in the process that led to that. Now, to explain why I support the idea, despite the growing pains, is simple. Let me first phrase to starting points, assumptions if you like: First, just like Wikipedia, we want Wikidata to be a resource where the provenance of facts is known. Second, I hope we can all except that Wikidata needs to be able to support Wikipedia and that Wikipedia is very rich. Third, can should accept that 60M entities is nothing compared to what a resource like Wikidata needs to support. In that context I like to remind people that we have 100M chemical compounds on which there is data. If we seek balance, why not just add 100M chemicals. Last starting point: we all hate the growing pains, we all hate the problems. So, to some specific replies. 1. there is consensus that provenance is important. 2. scientific literature is important provenance. 3. a scholarly article exists to support facts (this is complex, because Wikipedia actually has a lot of literature cited just because some person with a Wikipedia page wrote it; does that make Wikipedia bad?). 4. Many of the Wikidata community have been working on linking topics to literature, via References, via "main subject", etc. Scholia tries to visualize those links. And this is way beyond Wikicite/Scholia, but involves several other Wikidata projects 5. Scholia does not import literature. 6. I'm not a QuickStatements expert, but as far as I know it is rate limited, Magnus_Manske (talkcontribslogs) has been more than responsive, and has seen repeated approval (and SourceMD != QuickStatements). (Besides his ongoing Rust work!) 7. The Scholia project has actively engaged to overcome growing pains. 8. Literature is provenance and should be a central part of Wikidata, just like it is in Wikipedia. No one has yet convinced me that putting that on a separate platform will actually solve any problems. Federated queries are not known to speed up SPARQL queries, for example. 9. I have been involved in a few discussions, and there is no consensus what actually the cause of the growing pains is. Some say the size of the edits, others say the tool. 10. Links between literature and scholarly articles is hard; using them as reference even harder, but that is caused by copyright law. If you want to know how many Wikidata items are linked to articles, you cannot exclude 'main subject' and related properties. 11. Finally, I don't have the solution to the current problem. But if provenance is not an option for Wikidata, if a general-purpose (with sport) is not an option, then the community which just happens to have made most use of it (with a of benefit to the community, scholarly (many projects around that; Scholia and Wikicite are only two of many) and non-scholarly (like Gerard), should not be scapegoated. My 2 cents --Egon Willighagen (talk) 07:07, 30 July 2019 (UTC)

How to indicate parody?

I would like to indicate that one work is a parody of another work. I thought that using inspired by (P941) with the qualifier has quality (P1552) parody (Q170539) would be a valid approach since parody is a form of literary/artistic influence, but apparently not. Is there a better way or would it be necessary to create a new property? Alexander Doria (talk) 08:36, 29 July 2019 (UTC)

has quality (P1552) is the wrong qualifier. subject has role (P2868) would be more fitting for this usecase. ChristianKl❫ 11:42, 29 July 2019 (UTC)
@Alexander Doria: My proposition: add genre (P136) = parody (Q16669008) with a reference claiming that. Theoretically this should be enough to infer what you plan to show. If you want to reinforce the link between both works, you can add as qualifier for genre (P136) = parody (Q16669008), of (P642) = Bellerophon (Q1227693), but I don't really like this way of modeling. Snipre (talk) 14:34, 29 July 2019 (UTC)
I don't like constructions with "of". The word "of" is very vague and not easy to translate into other languages like Chinese. "subject has role" sounds more technical but it doesn't rely on particularities of the English language. ChristianKl❫ 15:18, 29 July 2019 (UTC)
I agree with using inspired by (P941)/subject has role (P2868). A work can be a parody of one work and inspired by others without parodizing them. Just having the a statement genre (P136) = parody (Q16669008) and some inspired by (P941)-statement does not make it clear what work (or other entity) a work is a parody of. But I would also support an own property, as this parody-relation is a quite frequent and distinctive relation. - Valentina.Anitnelav (talk) 15:12, 29 July 2019 (UTC)

Thanks a lot for all this feedback! @Snipre: Using inspired by (P941)/genre (P136) was my first approach but wasn't any more successful. In my opinion theses qualifiers referred to the statement and not the object but I agree it creates some ambiguity. For the time being, I'll use @ChristianKl:'s solution (inspired by (P941)/subject has role (P2868)). I also think, like @Valentina.Anitnelav: that in the long run, there may be enough parody relationships to justify a specific property (especially if Wikidata fiction coverage is extended beyond published books: there are millions of parodies in magazines and newspapers). Alexander Doria (talk) 18:32, 29 July 2019 (UTC)

The great thing about inspired by (P941)/subject has role (P2868) is that in case it gets used a lot it's easy to move the claims over to a newly created property. ChristianKl❫ 07:18, 30 July 2019 (UTC)

newly created item resistance group (Q65971239)

Hi! I created the item resistance group (Q65971239 R WLH) because most of the items linking now to it did not contain an instance of statement. I could not identify an appropriate article in any language project until now. If you are aware of one, please link it to this item.
There are many statements to be added at the various items. Articles about French resistance are using resistance movement (Q138796 R WLH) which is not the same as a group of people, Please verify what you can do and ping me. Thanks in advance
no bias — קיין אומוויסנדיקע פּרעפֿערענצן — keyn umvisndike preferentsn talk contribs 21:02, 29 July 2019 (UTC)

French resistance as a whole (or more broadly resistance during World War II (Q2094191)) is clearly a resistance movement (Q138796), but specific french resistance ggroups shoudl probably be moved to your new item, which I have taken the liberty to rename to a more specific name. Circeus (talk)
It's bad to rename an item in only one language to a more specific name but not in the others that have labels. Why do you think it's important to rename it in one language and risk people misusing it? What's the issue with having a more general item? ChristianKl❫ 06:45, 30 July 2019 (UTC)
I did not "think it's important to rename it in [only] one language and risk people misusing it", and I'd appreciate you not making blanket accusations of the sort. If anything, the original name was the one likely to cause misuse. I just happen, like 99% of wikidata users, to not display all possible languages when editing names, and when I changed the name, it wasn't showing that there even was a name entered in German. I've put a quick'n dirty edit to it because I haven't really done with German much since my 2 years of it in Cegep 17 years ago. Circeus (talk) 14:49, 30 July 2019 (UTC)
@Circeus: By narrowing the scope of an existing item, you caused the names of the item in different languages to be out of sync. That's a reason why scopes normally shouldn't be narrowed if you don't think doing so is important. It's not something that should be done in a dirty way. If you think it's important enough to do, then put in the work. An item having no label is better then it having a wrong label.
I don't think the fact that I assume that you acted in good faith (you had a good reason for doing something that creates problems) to be a 'blanket accusation'.
What kind of misuse do you expect to happen when there's an item called "resistance group" that analogous to resistance movement (Q138796) in name and scope?  – The preceding unsigned comment was added by ChristianKl (talk • contribs).

Abuse filter

Any abuse filter for that? --Metrónomo (talk) 11:30, 30 July 2019 (UTC)

I don't see how a abuse filter would help here. What exactly would you want to filter? ChristianKl❫ 13:18, 30 July 2019 (UTC)
Abuse filter is quite slow. I'll suggest to wait for a new ORES-feature that allows training the prediction system on the fly. Sjoerd de Bruin (talk) 14:58, 30 July 2019 (UTC)

From the nonsense of data mass imports

Screenshot 2019-07-27 Suchergebnisse für „Tao Chen“ - Wikidata.png

I just tried to find out if Wikidata already has an entry for a certain Chinese athlete named "Tao Chen" (see image). It was not easy to find out, because there are a lot of entries for this or similar name. Who hides behind most entries is almost impossible to find out, as there are imports of ORCID and CBDB. Without further details. This data is now on Wikidata and lazy, because it is usable for nothing. The project Wikidata is currently at a crossroads. Continue with an unbridled, unchecked mass import of meaningless data, or finally, focus on creating high quality records. -- Marcus Cyron (talk) 14:27, 27 July 2019 (UTC)

  • So you are trying to say these people should bear a different name, just because you can't tell them apart? --- Jura 15:00, 27 July 2019 (UTC)
Wow! I'm impressed, what you read out of my words! OO -- Marcus Cyron (talk) 23:35, 27 July 2019 (UTC)
I think you might just as well have written that we have plenty of items about taxons and this makes you unable to determine which football player is your athlete. --- Jura 12:46, 28 July 2019 (UTC)
I also find it sometimes hard to find an wikidata object for a person, because someone with the same name is included in a lot of scientific papers and some persons are listed not a the begin of the search result but somewhere inbetween. From my point of view it would be helpfull to have different sort options (e.g. list all items (not just some) that match that name first, than all papers which only include that name) and/or better filter options (e.g. only search for persons, exclude papers, ...). --M2k~dewiki (talk) 15:10, 27 July 2019 (UTC)
@Daniel Mietchen, Fnielsen: this thread is yet another indication that the indiscriminate import of ORCID profiles and scientific articles does more harm than good. It would be great if that effort could be limited to a well-defined scope which would limit the volume of the imports to reasonable levels. Such a scope would allow you to aim for completeness within that scope. Without completeness, Scholia's statistics are highly skewed. − Pintoch (talk) 16:13, 27 July 2019 (UTC)
Can you explain how that is so? The users is looking for an athlete. --- Jura 16:16, 27 July 2019 (UTC)
Marcus explicitly mentions the ORCID import as one of the causes of the issue, by flooding the search results with namesake items. These items often do not have enough information to disambiguate them (ORCID profiles are often empty apart from the name of the person, being of little use for disambiguation). − Pintoch (talk) 18:39, 27 July 2019 (UTC)
Correct. This is my problem. The majority of the examples I take to fix issues were impossible to fix, because there were not enough informations to figure out, which person is meant. -- Marcus Cyron (talk) 23:37, 27 July 2019 (UTC)
What issue did you try to fix for your athlete in relation to ORCID and CBDB? --- Jura 23:42, 27 July 2019 (UTC)
@Pintoch: I am not sure why you drag in Scholia. Scholia is not a ORCID GUI. I don't understand your argument, btw: ORCID makes considerable effort to only have unique efforts (and part of the core design and purpose), and this involves many international publishers, organisations, community projects, etc. A different ORCID means a different person. Sure, they have errors too. Do you have support for a claim that the error rate in ORCID is unhealthy large? Purely anecdotal, I have more issues merging chemicals because of important from Wikipedia (or splits, for the matter) than I ever had for ORCID authors. I am not advocating mass imports, and have been very hesitant about that (I have an approved bot, but hesitant in putting it into action). I am not aware of any effort of indiscriminate import of ORCID profiles into Wikidata and I hope you can give me more info. I can only assume you refer to tools like SourceMD, but that does not make sense to me, as it cannot import data that is not in ORCID. Can you please add details? You cannot blame Scholia for everything you do not like; if there are actions you do not like, name them specifically. Thank you. --Egon Willighagen (talk) 07:28, 30 July 2019 (UTC)
@Marcus Cyron: You can search for "Tao Chen" and filter out the ORCID iD (P496) and CBDB ID (P497) junk and only get the humans with this search. Multichill (talk) 19:23, 27 July 2019 (UTC)
@Multichill: - thanks for this hint! :) This will help a bit, mainly when I work on sportspersons. For scientists this ofcourse can not fix the problem. But I'm really happy for having the half of my personal problems with this solved :). -- Marcus Cyron (talk) 23:39, 27 July 2019 (UTC)
In my opinion 1. ORCID profiles are useful for disambiguation as there're little duplicate and they provide a basis for future improvement. Author items without identifiers instead do not prevent duplicates from happening. 2. For checking the biases in Wikidata, we may track the completeness of data relative to a specific dataset/data dump. Before entries in these dataset/dump are complete Wikidata will always be biased. It will take years to complete the import of a huge dataset of bibliographic data (about 200 million).--GZWDer (talk) 21:29, 27 July 2019 (UTC)
@GZWDer: I do not think the alternative is to create author items without identifiers. The alternative is not to create these items at all! Or to take the time to curate them manually (so, only to create as many as we can afford to curate manually). − Pintoch (talk) 10:10, 28 July 2019 (UTC)
I said "author items without identifiers" is a bad workflow (and should be avoided). Even if we only curate the authors manually, we will eventually have more and more items with same names (labels). The "ORCID=..." is already enough to differentiate items for researchers and other items with the same name.--GZWDer (talk) 03:25, 29 July 2019 (UTC)
Marcus, the premise of your argument fails entirely. A search for Tao Chen brings back many ORCID and CBDB imports, all properly described as being ORCID or CBDB persons, such that the very few non ORCID and CBDB candidates can be identified and checked. It doesn't appear that the presence of hundreds of thousands of ORCID or CBDB imports impairs the search function in any meaningful way. To the extent a 'solution' to this supposed 'problem' is required, it is provided by items' description fields which can & should be used to disambiguate items in such a way as to facilitate the use of the search function: and as noted in your Tao Chen example, for CBDB and ORCID, this has been done without exception. Should there be import sets where descriptions are lacking, these can be added with minimal SPARQL and QuickStatements effort. --Tagishsimon (talk) 22:30, 27 July 2019 (UTC)
+1 --- Jura 22:39, 27 July 2019 (UTC)
It seems, you both did not want to understand, so you do everything to bring the disccussion into an other direction. Funny. OK, once more: in the moment you work on something and you look for maybe already existing here but you only find data junk, it's hard up to impossible to find something. But what I had learned is, that Wikidata is full of this data junk. What makes this project nearly unusable. It's so much …, when do you will bring these data together? The Wikidata clue is to bring all this together. And not to be a mirror of other databases. But OK, I'm out. Once more I need to learn, that every critic is unwanted. Mass before class. These are those moments I need to ask the universe, why editor as Multichill & Co to a so phantastic job on projects as Summ of all paintings, when at the end mass is wanted, not class. We would have had a one time opportunity to create something very big, very great, not to say phantastic, a collection of all world wide knowledge on one place in a multilingual project. This will only work, when it's possible to work with and within the project. -- Marcus Cyron (talk) 23:54, 27 July 2019 (UTC)
The question above still stands: What issue did you try to fix for your athlete in relation to ORCID and CBDB? --- Jura 00:31, 28 July 2019 (UTC)
@Marcus Cyron: it took me quite a while too to find out who this guy c:Category:Tao Chen was. He's 陈涛 (Baidu Encyclopedia) from Hangzhou. (His name should be Chen Tao, btw.) I think his item is not yet created.--Roy17 (talk) 23:29, 27 July 2019 (UTC)
@Roy17: - I'm working on it. I do it by hand (!) and check everything. I can't do mass edits with this. My work needs time, but you will know then who is who! But funny, that I hit the problem so much, that you follow my edits over the projects …. -- Marcus Cyron (talk) 23:54, 27 July 2019 (UTC)
@Tagishsimon, Jura1: Hang on, since when are descriptions from the ORCID imports satisfactory? Are you familiar with Help:Description at all? Adding ORCID ids in the descriptions is an ugly hack, descriptions are not supposed to be used for that at all. Assuming that everyone with an ORCID is a researcher is also a gross approximation. This has only become necessary because of the scale of the ORCID import and the lack of proper curation of these items. The description researcher ORCID: 0000-0002-3941-1603 is not satisfactory at all, a satisfactory description for Tao Chen (Q55979803) would be something along the lines of chemistry researcher at the University of Surrey, UK. Yes it takes some effort to produce these descriptions: if the Wikicite project is not capable of doing that, then it should not import these half-baked items at all.
Moreover, athletes can have ORCID ids, so when looking for an athlete you cannot in principle discard all items with ORCID ids (take René Born (Q26932946) for instance, you can find plenty of others with SPARQL). It is just not acceptable at all to mass-import datasets carelessly and expect other users to adapt their workflows to deal with the inconvenience. − Pintoch (talk) 10:10, 28 July 2019 (UTC)
Its arrant bollocks to argue, as you appear to be doing, that if it's not possible to do an import job perfectly it should not be done at all. That's just not how wikidata works. No-one's arguing that researcher ORCID: 0000-0002-3941-1603 is optimal. It is a satisfactory method of disambiguating item X from item Y, and iirc Help:Description - if you've ben arsed to read it - makes clear the purpose of the description is disambiguation. Your suggested description chemistry researcher at the University of Surrey, UK fails in just the same way as researcher ORCID: 0000-0002-3941-1603 to convey that item X is the athlete we're after, because descriptions cannot & should not amount to a full biography. Wikidata is a shared resource which proceeds incrementally. You're just going to have to live with the reality that your workflow will necessarily from time to time be impacted by the 'inconvenience' of other users. Apologies that the world doesn't give a shit about your solipcism, but there we are. --Tagishsimon (talk) 10:42, 28 July 2019 (UTC)
@Tagishsimon: I am sorry if I offended you. Can you have a good look at your reply above and confirm me that you think this language is acceptable on Wikidata? I am not sure my testicles have anything to do with the discussion here. I think you should apologize. − Pintoch (talk) 11:34, 28 July 2019 (UTC)
The current descriptions "researcher ORCID: XXX" are not intended to be permanent, but is already enough to differentiate WikiCite items from others (If you think these description does not help for finding athletes, the full description like "chemistry researcher at the University of Surrey, UK" does not either). Once employer (P108) having added (more than 200000 items had) the description can at least be improved to "researcher from XXX University" where the term "researcher" can be further refined using field of work (P101).--GZWDer (talk) 03:34, 29 July 2019 (UTC)
@GZWDer: sure, in general it is fine to work on imports in incremental stages. The problem in this case is the scale - this import is really big, and it makes it significantly harder for others to improve Wikidata. This is similar to the Cebwiki import: mass-creating lots of duplicates for geographical entities was meant to be a work in progress too, but it is massively annoying for people trying to add locations on other items, by having to filter out "Cebwiki items". Wikidata is a collaborative project, so it is better if people can be mindful of others when doing their imports. We should not have to teach people how to ignore "Cebwiki items", "ORCID items" or "CBDB items": let us just not create items that are meant to be ignored by the rest of the community. It would also be great if Wikicite could stop using the number of items created as a metric of its success as a project. − Pintoch (talk) 12:23, 29 July 2019 (UTC)
I am curious at whether there is any rules against "too big file import" / "low quality import" ? I agree Cebwiki/SVwiki bot may be annoying, but they are stating often true (but poor quality) elements. And there should a button on a Qx element, that would say "Hey, look at this other Qx, they look to be very similar : please look at both of them and decide of a potential merge" Bouzinac (talk) 13:01, 29 July 2019 (UTC)
Whether an import is too big or whether its quality is too low is decided on an individual basis at Wikidata:Requests for permissions/Bot… but only when the authors actually use this space and respect the results. − Pintoch (talk) 13:21, 29 July 2019 (UTC)
  • I think it depends on the type of items and the number of statements that are being imported. The two imports initially mentioned clearly have some issues, but oddly those don't seem relevant to the question at hand (an athlete). The PubMed items that were also mentioned are probably relatively numerous and our tools are not quite up to it, but I think the items have relatively few issues and don't overlap with other content. As such, I think they are fairly unproblematic and give a good basis to improve relevant features. If we actually absorbed most of it, I think it's a great achievement for the project. The problem with "Cebwiki items" is that they tend to overlap with some content, initially included few statements and have some quality issues when they do. For some regions, I think steps were taken to clean them up, e.g. the US. Obviously, a direct import from GNIS might have been the better route. --- Jura 13:54, 29 July 2019 (UTC)
People searching for names in general should realize that matching by name is a highly complex task. Many names can be spelled many different ways - even ordinary English names with capitalization, middle names, initials etc. Chinese names probably should use the characters of the language, not the anglicized version of the name (especially with name parts reversed), but searching on those can be tricky too. Just because you've found no matches when searching by name doesn't mean an item doesn't already exist for that person; conversely you may find hundreds of matches on a common name; you should probably find some other way to filter your search, but it's work either way. This is a very general problem and complaining about the large number of people with the same name doesn't really help us address it. That's partly why identifiers like ORCID and ISNI are so important. ArthurPSmith (talk) 20:17, 29 July 2019 (UTC)
If I am allowed to add further complexity to the Tao-Chen-problem: ORCID may not always have a one-to-one relation to a real world person. I believe and are the same endocrinologist from Sichuan University West China Hospital, corresponding to Tao Chen (Q60032793) and Tao Chen (Q60032792). — Finn Årup Nielsen (fnielsen) (talk) 16:20, 30 July 2019 (UTC)

Apart from Multichill's neat query, "tao chen -orcid -scientific -researcher" seems also to give reasonable results.

On Wikidata footballer generally have the occupation association football player (Q937857) listed. If you seek for the item of a footballer named Tao Chen you can simply search "Tao Chen football" and you will see the two footballer with that name.
That said, sometimes you won't find the existing item and create a new one. Maybe, there are multiple variation of the name. That's also okay. Today, we have constraint reports when the single value constraint is violated for external ID and when you add decent external IDs, sooner or later we will merge the items. In the future we might have even better lists of candidates to be merged if someone gets around to writing machine learning code.
If we want Wikidata to grow, then we have to find ways to deal with having a lot of items of people with the same name. ChristianKl❫ 14:55, 31 July 2019 (UTC)


Edit 987,654,321 was done around the midnight (UTC). One bilion is near... --Matěj Suchánek (talk) 13:02, 28 July 2019 (UTC)

Special Wikidatalogo for 1 Billion Edits

Proposal 1


what do you think about changing the logo shown on the Wikidatawebsite for a few days, when Wikidata reaches 1 Billion Edits. I think that were a way to show readers that Wikidata has 1 Billion Edits. The Special Logo should not contain a name for the number because in German Billion is something else as in English. -- Hogü-456 (talk) 20:25, 28 July 2019 (UTC)

Any graphics to suggest? Nomen ad hoc (talk) 22:02, 28 July 2019 (UTC).
In SI terms Wikidata would have reached one gigaedit :) perhaps something based on that? --Lucas Werkmeister (talk) 14:12, 29 July 2019 (UTC)
Made some graphics suggestion in c:Category:Wikidata one billion edits. - Premeditated (talk) 22:02, 29 July 2019 (UTC)
Personally I like this (the one linked above as "Proposal 1")! ArthurPSmith (talk) 14:55, 30 July 2019 (UTC)
Same as for me. Nomen ad hoc (talk) 18:14, 30 July 2019 (UTC).
Me too, nice logo variant, maybe the start of a monthly "Wikidata Doodle"-program?? ;) Moebeus (talk) 00:30, 31 July 2019 (UTC)
I don't think we should use numbers of edits as a metric worth celebrating. Bot edits that add additional descriptions are not worth celebrating. If we change our logo it should be because of metrics such as monthly users with at least 1/5/100 edits that reach a milestone. ChristianKl❫ 07:02, 30 July 2019 (UTC)
As someone who will admit to having used QS very inefficiently, I do agree with Christian that this milestone in itself is not worth celebrating. Mahir256 (talk) 03:56, 31 July 2019 (UTC)

I'm not sure if this have been brought up before

Are trailers for movies, television shows or video games notable enough to warrant their own items? --Trade (talk) 20:40, 29 July 2019 (UTC)

I made a quick example here The Sims 2: Official Trailer (Q65972027). Are there any properties that you think i should or should not have used? I wish there was a way to indicate the soundtrack used but i cannot find such a property. --Trade (talk) 21:13, 29 July 2019 (UTC)
@Trade: Use advertises (P6291) to connect it to the movie. Thierry Caro (talk) 13:18, 31 July 2019 (UTC)
Was it ever revealed why they didn't created seperate items for film trailers? Given how popular some of these trailers are it just seems weird. I imagine it might be because we don't have third party databases to reference the information like we do with music videos and IMDB --Trade (talk) 21:50, 29 July 2019 (UTC)
At Wikidata_talk:WikiProject_Movies/Archive_1#Property_for_trailer_(Q622550)_? we discussed a possible property, but using existing ones worked out just as well. --- Jura 21:55, 29 July 2019 (UTC)

series ordinal

series ordinal (P1545). List of presidents according to chronological order: A B C D B E F G.

B was president two different periods. For the first period we can say that he/she is the 2nd president (P1545 -> 2).

  1. What about the second period? Are we going to say that he/she is the 5th president (P1545 -> 5) or again the 2nd (P1545 -> 2)?
  2. And what about president E? Is he/she the 5th or the 6th president?

Xaris333 (talk) 08:14, 30 July 2019 (UTC)

@Xaris333: it might vary by office / country, but for the US presidency, for example, Grover Cleveland (Q35171) is considered to be the 22nd and 24th president, and his successors are considered the 23rd and 25th president, respectively. But leaving that example aside – ultimately, that comes down to what the sources say, doesn’t it? --Lucas Werkmeister (talk) 09:55, 30 July 2019 (UTC)
Sometimes the one source says different number than the other. I have to ask here just to know I am not doing something wrong. Xaris333 (talk) 10:00, 30 July 2019 (UTC)
I have seen Wikipedias using this when numbering Swedish royals. It is then often a "homemade midden (Q1152199)". So add this only when you find good sources. Nobody know who was the first Swedish king. One list I have seen sets Charlemagne (Q3044) as first. Another puts Odin (Q43610) as second. IP (Hej!) 13:56, 30 July 2019 (UTC)
In my example, we know for sure who were the presidents and their order. Xaris333 (talk) 14:55, 30 July 2019 (UTC)
Yes Xaris333, stay with those. The problems often comes when people who only know about US, tries to extend the idea to other areas. IP (Hej!) 07:59, 31 July 2019 (UTC)

Wikidata:Database reports/Constraint violations/P301 - inverse violations and references to deleted objects

Hello, Wikidata:Database reports/Constraint violations/P301 contains a list of Inverse violations (current "violations count" is 28764). Is there a tool or bot to add the missing property for these 28764 entries? The list also contains or contained references to objects which have been deleted more than three years ago (for example [36][37][38] and more). Is there (in general) a tool or bot to identify (and maybe also remove) references to deleted objects? --M2k~dewiki (talk) 14:59, 31 July 2019 (UTC)

also see Wikidata:Forum#Nachtragung_von_Hauptkategorie_zum_Artikel_(P910)? (German) --M2k~dewiki (talk) 15:02, 31 July 2019 (UTC)
There is Wikidata:Database reports/Deleted Wikidata entities that are still linked which should in principle list use of deleted items, but the old ones which you have just removed did not appear there for some reason. Do you have more linked deleted items which are not yet removed? --MisterSynergy (talk) 15:14, 31 July 2019 (UTC)
Hello @MisterSynergy: if found about 30 objects with deleted references in Wikidata:Database_reports/Constraint_violations/P301#"Inverse"_violations with a "Violations count: 28764" but the report also says "Too many results. 23763 records skipped.", so there might be more similar cases, which are not show/skipped in this report. --M2k~dewiki (talk) 15:18, 31 July 2019 (UTC)
Okay, I found some more cases on other covi reports. Seems like there are some entries missing in the pagelinks table, thus the report mentioned earlier does not find them. —MisterSynergy (talk) 18:43, 31 July 2019 (UTC)
For adding, use SPARQL and QuickStatements. But keep in mind that the property is often misused and confused with eg. category combines topics (P971), so thoughtless import can cause problems (and annoy users). --Matěj Suchánek (talk) 19:52, 31 July 2019 (UTC)

Link to depictions on Commons

When I am looking at the rabbit (Q9394) item on Wikidata, it would be cool if I had a simple way to see all depictions of rabbits on Commons (API). Either as a paged gallery on the main panel of the Wikidata item, or as a link in the side bar, for instance below "What links here". Cheers! Syced (talk) 02:52, 29 July 2019 (UTC)

I was actually talking about this with a few of the guys on the Telegram group the other day. There seems to be a bit of controversy on whether or not image (P18) should allow multiple entries. While on one hand it is good to be able to have several images for a species like a rabbit, coming in different breeds, shapes and sizes; on the other it's a bit more desirable to have a single image to abstractly it overall.
I'm no expert but right now you should be able to get multiple images if you're good at querying, but adding such a feature as this is likely not a change to be welcomed by everyone. I for one however, would like to support it. At the very least having a tool to do this would be pretty sweet. Themadprogramer (talk) 07:00, 29 July 2019 (UTC)
Would be a nice gadget to build. Maybe something to work on in the upcoming Wikimania hackathon. Multichill (talk) 20:18, 29 July 2019 (UTC)
@Syced: I just created a tool to show the depictions. Hope it helps! --Stevenliuyi (talk) 20:43, 5 August 2019 (UTC)
Nice! I added the tool to my Wikidata sidebar, it works great. I hope it can one day become included by default for all users :-) Syced (talk) 02:06, 6 August 2019 (UTC)

Follow-up on RFC

@Ivanhercaz, Nicereddy, MarioFinale, Kristbaum, -jem-, Abián:

As you protected or supported protecting a large number of items, what's your plan for edit requests like on Talk:Q987? Will you monitor these or just hope someone will eventually find them? It seems to be me that implementation of your approach needs some work. Can you share how you plan to go about it?

If nothing is done, I suppose we just have to unprotect the pages. --- Jura 09:35, 26 July 2019 (UTC)

The fact that those users have supported semi-protecting, or have semi-protected, some Items doesn't make them responsible, or at least not more responsible than any other user or the development team, for implementing unrelated measures such as an improved interface or new mechanisms for tracking requests for change of semi-protected Items, in fact it's a matter of opinion whether or not such new mechanisms should be introduced. Having said that, personally I do think new features in the interface, such as a new link for "feedback", would be useful. I was hoping more users would participate in this thread but, since it was already archived, I think I'll open some Phabricator tasks about these possible new features anyway and I'll link them from here. Thanks for the reminder! --abián 10:23, 26 July 2019 (UTC)
Well, it's for us to ensure that WMF's goal of openness is upheld. It's clear that you feel that you aren't bound by protection policy, but I suppose you understand that edits need to be possible.
If one approach doesn't work, we need to implement an alternative. One thing is sure: developers wont be doing the edits, even if you open a phab task. --- Jura 10:38, 26 July 2019 (UTC)
A solution would be placing instructions on talk page of protected items in which user can add a category in edit request so that those request can be tracked and dealt. ‐‐1997kB (talk) 12:39, 26 July 2019 (UTC)
That would be similar as {{Edit request}} for fully protected pages. However, as far as I know we are not even able to determine which items should be protected according to the policy, and where protection needs to be removed again because the criteria are not met any longer. It seems very premature to me that we already apply this policy at this stage. --MisterSynergy (talk) 12:50, 26 July 2019 (UTC)
It only took a day for the page to be edited, no? That seems fine to me. Nicereddy (talk) 05:18, 31 July 2019 (UTC)
@Nicereddy: It was not only one protected I guess. And looks like it got handled because it was brought on noticeboard. Without a proper system to handle such requests, there's chances some will go unnoticed. ‐‐1997kB (talk) 06:37, 31 July 2019 (UTC)
This is the task I opened on Phabricator. What do you think? We can redirect those links to a new page Wikidata:Edit requests or similar, which may be edited by preloading (?action=edit&preload=...) a template. Any auto-confirmed user should be able to participate in resolving requests. --abián 09:26, 31 July 2019 (UTC)
Having a edit request page sounds like a good idea. At best the page should make it clear that the edit requests should include a link to the source for a claim and not just "change x to y".
Do we want to have a different page for the requests for each language or a common page? ChristianKl❫ 15:16, 31 July 2019 (UTC)
I'd say we can have a single page; even if we create different pages for different languages we can't guarantee that the language chosen by users and the page on which they write match, and requests in minority languages, even in languages other than English, might be ignored. --abián 21:27, 5 August 2019 (UTC)
(Since the administrators' noticeboard isn't the proper place for this thread, I'm moving this thread to the project chat.) --abián 09:34, 31 July 2019 (UTC)

User:Abián, the other part which is still missing is the monthly evaluation of pages which need to be newly protected, and where protection needs to be lifted because the criteria for protection are no longer met. We cannot implement the outcome of the RfC as long as we do not have numbers easily available.
I also have concerns about the some thousand protections which you applied last year before starting the RfC, and which are still in place. To my understanding, those protections have never been evaluated, although it should have been done according to the RfC. Can you please lift protection of those which are not used by 500 pages any longer, as the Wikidata:Page protection policy does not support protection any more? --MisterSynergy (talk) 11:29, 5 August 2019 (UTC)

@MisterSynergy: For now the Phabricator tasks for tracking the implementation of these regular checks and potential semi-protections are phab:T226883 and its subtasks, phab:T226882 and phab:T226885; I think I can take care of the latter. The Items that I semi-protected before the RfC were based on the data described in phab:T210664 and are just a subset (the most sensitive subset) of those that should be semi-protected right now according to the results of the RfC, so we probably don't have to worry about old semi-protections. Thanks for your interest! --abián 21:27, 5 August 2019 (UTC)