Shortcuts: WD:PC, WD:CHAT, WD:?

Wikidata:Project chat

From Wikidata
Jump to navigation Jump to search

Wikidata project chat
Place used to discuss any and all aspects of Wikidata: the project itself, policy and proposals, individual data items, technical issues, etc.
Please take a look at the frequently asked questions to see if your question has already been answered.
Please use {{Q}} or {{P}}, the first time you mention an item, or property, respectively.
Requests for deletions can be made here. Merging instructions can be found here.
IRC channel: #wikidata connect
Wikidata Telegram group
On this page, old discussions are archived after 7 days. An overview of all archives can be found at this page's archive index. The current archive is located at 2019/10.


Wiki data before 2016[edit]

I am trying to load the monthly page-views data between 2002-2019 with the following query:

Unfortunately, I am getting the data only between 2016-2019. Is there any way to get the earlier data? I understand that the earlier metrics can be different. As much uniformity in the data as available -- will work for me. Any help will be much appreciated.  – The preceding unsigned comment was added by (talk • contribs) at 01:44, 1 October 2019‎ (UTC).

Example of parliamentary group of the party[edit]

Hello. A new property constraint was added to parliamentary group (P4100). I don't disaggre. The problem is that I have used some months ago the property for many items using the party item and not the parliamentary group of the party. I just need a good example of a parliamentary group of the party (for national parliament) to create new items and correct the statements. Xaris333 (talk) 11:56, 5 October 2019 (UTC)

@Xaris333: I've updated the constraint, as this was giving an error on a lot of existing items. It should be okay to use a political party rather than a larger group as the value here, if that is what seems appropriate for your context. (It will vary a bit by country) Andrew Gray (talk) 18:33, 9 October 2019 (UTC)
@Andrew Gray: Thanks. But, what is better for wikidata? To add the party item or the parliamentary group of the party item? I am interesting about Cyprus parliament House of Representatives (Q1112381). Xaris333 (talk) 09:49, 15 October 2019 (UTC)
@Xaris333: As far as I can see from English WP - and I might be wrong! - Cyprus doesn't really have political groups (organised groups of parties), so I think it would make most sense here to use the parties, eg Progressive Party of Working People (Q212158) or Democratic Rally (Q644973).
An example of a country that does is Chile, where a lot of small parties are grouped together into larger groups in Parliament, and so we have Luciano Cruz-Coke (Q6696802) using parliamentary group (P4100):Chile Vamos (Q20859150), rather than Political Evolution (Q16565533), his actual party. See w:Chamber of Deputies of Chile for an example of how this breaks down. Andrew Gray (talk) 21:45, 15 October 2019 (UTC)
OK. I hope that we will not have problems in the future. Xaris333 (talk) 21:57, 15 October 2019 (UTC)

Can a bot add in siblings?[edit]

If someone has 11 children with entries in Wikidata someone has to add in the 10 siblings to each record of each child for 100 entries. This could better be handled by a bot that will give fewer errors. Does one exist already?  – The preceding unsigned comment was added by Richard Arthur Norton (1958- ) (talk • contribs).

I'm not sure what the point would be. Queries would be better off just checking the parent-child data directly. --Yair rand (talk) 02:13, 7 October 2019 (UTC)
  • I think most people would be coming to Wikidata through Google or Siri or Alexa, and not running a query using SPARQL. Especially for entries not in English Wikipedia. --RAN (talk) 13:29, 10 October 2019 (UTC)
Poulpy (talkcontribslogs) used to run such a bot, I believe. Thierry Caro (talk) 04:16, 7 October 2019 (UTC)
A bot would be best, but in the short term you can use User:Matěj Suchánek/moveClaim.js to copy claims from one item to another. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:17, 7 October 2019 (UTC)
If no one has the script available, I would be willing to create and run it, if a request is made via WD:RBOT. I believe there will be more family relations that can be handled the same way (sibblings, parents/children, grand-parents, uncles, etc). I am working out something similar with ranges (follows (P155) and followed by (P156)) in a serie/chain. Edoderoo (talk) 12:43, 8 October 2019 (UTC)
A bot or single-click user-operated tool would be absolutely wonderful for tasks like this, especially for items like parent-child relationships or spouse (P26) or any statement with symmetric constraint (Q21510862): it would be great if the nagging notice of "Joe Blow should also have the symmetric statement spouse Jane Doe" had a button saying "click here to perform this". This would make huge gains in improving the completeness of data entered largely by humans freely volunteering their time. If I can perform an action in one click instead of 3 or 4, Wikidata is all the better. -Animalparty (talk) 16:37, 11 October 2019 (UTC)

Building that has changed uses[edit]

Are there any good models of how to deal with a building that has changed uses over time? This came up with Mercy Friary (Q47766776)/Museum of Fine Arts of Seville (Q2163496). I'm not sure I found a satisfactory solution -- it seems here like what I've done associates the physical building with Mercy Friary (Q47766776) and there is no item for the religious community that happened to reside in the building (whereas the museum that happens to reside in the building gets its own item). Is there a better way to do this? Calliopejen1 (talk) 05:33, 10 October 2019 (UTC)

So I give up: I've just looked at three prominent buildings that have this situation and the way it's handled ranges from poor (General Post Office (Q5532118)/Hotel Monaco (Q5911750), Romanian Peasant Museum (Q623662)) to "at least you might be able to work it out" (Dallas County Courthouse (Q5211300)). Anyone got any good examples? - Jmabel (talk) 06:12, 10 October 2019 (UTC)
Would Musée d'Orsay (Q23402) qualify? It uses P138 to refer to Gare d'Orsay (Q2698691).--Ymblanter (talk) 09:45, 10 October 2019 (UTC)
Louvre Palace (Q1075988)? Ayack (talk) 09:52, 10 October 2019 (UTC)
  • I think it's important to remember that a building is a building and a museum is not just a building but also an institution. That's why there should always be 2 items : one for the building itself and a second for the institution in the building. This is what I did for some museum like Les Champs Libres (Q3231356) (in this case, there was no choice as there is 3 institutions in the building). With that in mind Museum of Fine Arts of Seville (Q2163496) seems wrong as it mix things related to the building and things related to the instutition. As some property are used in both context with different meaning (for instance inception (P571) means construction date for a building and establishment/foundation date for an institution), it's even more importante to make a clear distinction. Cheers, VIGNERON (talk) 13:04, 10 October 2019 (UTC)
Interestingly, I was about to bring up the same question about a school building. I made two entries for the organizations housed there and treated it as a succession, using "replaces" and "replaced by" when I could not find a standardized example. --RAN (talk) 13:23, 10 October 2019 (UTC)
See also hospital (Q16917) and hospital (Q180370) where we have the organization that is housed in a building. The hospital organization can be housed in multiple buildings over their century long existence. --RAN (talk) 16:15, 10 October 2019 (UTC)

Hagia Sophia (Q12506) is also a good/bad example of handling this. Calliopejen1 (talk) 18:31, 10 October 2019 (UTC)

  • St. Helena Public Library (Q58126363) is an example I am happy with. - PKM (talk) 01:18, 11 October 2019 (UTC)
  • I'd consider making all buildings simply instance of building (Q41176) and using other properties, like use (P366) and occupant (P466) to record how they are used, with start and end dates as required. This would help separate items for buildings from items for organizations. Ghouston (talk) 03:59, 11 October 2019 (UTC)
    • We need to address items like prison (Q40357), which is subclass of both “facility” and “institution”. - - PKM (talk) 21:25, 17 October 2019 (UTC)
      • It's certainly not unheard of for buildings to change use from prisons to something else. But I think generally we shouldn't be using subclassing to express information that has a dedicated property. With prison (Q40357), I'd be included to make it an institutional item, although even if use (P366) is used, you need an item to set as the value. Ghouston (talk) 23:40, 17 October 2019 (UTC)

item change log filtering[edit]

Hi, this may be a newbie question but I didn't find the answer myself. I want to filter the change log for an item to list changes only to a particular statement. E.g. for Q43 I want to get an overview of changes to demonym (P1549) through item history. This could be known as a blame tool, but searching for such thing on WD didn't get me any useful hit.

The ordinary history view can filter for tag, but that isn't very useful in this case. It would be more useful to filter based on change summary, I think. I could also see a use to filter out label/description changes through the same means. Is there a tool for it, or a way to query change log?

I had a look through Wikidata:Tools with no luck. If you have the answer, I'd be happy for a Ping! Jagulin (talk) 12:16, 10 October 2019 (UTC)

Please answer here, I'm interested too! --SCIdude (talk) 13:25, 10 October 2019 (UTC)
User:Ricordisamoa/WikidataTrust.js. This is not a scalable tool (for Q34 about 2.6GB need to be downloaded).--GZWDer (talk) 15:38, 10 October 2019 (UTC)
Thanks. In that case maybe this should be a suggestion for improvements to the history view. Anyone know of such already, in Phabricator? I also think it should be possible to do a query for the change log. I will have a look at the API. Jagulin (talk) 21:58, 15 October 2019 (UTC)

Location of an image of a person from the web NOT stored at Wiki Commons[edit]

I thought at one time we had a property that stored the url of an image from the web of a person or thing that is NOT from Wikimedia Commons. Google displays fair use images in Knowledge Graph (KG). Secondly, if we had a property storing the url of an image, is there anything legally that prevents us from displaying it under the doctrine of fair use. English Wikipedia allows one fairuse image per entry, except for living people, at one point they were all purged for living people. We could disable the "save as" function, so it cannot be downloaded directly from here. Twice this week I came across Google (KG) displaying the wrong image of a person. On some Google (KG) boxes I get an edit button where I can suggest a change, other searches, I get no edit box. Does anyone else here get the edit function, or is it a trusted user function some people get, like I had at Google Maps? --RAN (talk) 13:45, 10 October 2019 (UTC)

  • The commitment that all data here is CC-0 would presumably preclude including images on a fair-use basis. Remember also that fair use is a concept specific to U.S. copyright law. - Jmabel (talk) 16:34, 10 October 2019 (UTC)
    • Exactly. There data here is CC-0; That doesn't stop us linking to all manner of legitmately available copyright material, including numerous commercial websites, encyclopedias, directories, etc. So there is no objection to linking to indicate where a in-copyright or even an agency image may be found. What use downstream reusers choose to make of such links is for them to assess, based on the copyright laws local and applicable to their situation. Most EU countries, for example, include an exception allowing copying for personal private research; as well aw further exceptions going beyond that which may be applicable in particular contexts. Jheald (talk) 21:05, 10 October 2019 (UTC)
      • @Jheald: Linking from Wikidata is fine, but his question appears to be about displaying the image. "… is there anything legally that prevents us from displaying it under the doctrine of fair use … Twice this week I came across Google (KG) displaying the wrong image of a person." - Jmabel (talk) 02:51, 11 October 2019 (UTC)
        • If this was going to be done, it would probably be advisable to make a copy of the image in Wikidata, as is done on Wikipedia, to avoid dead links and since a lot of sites won't permit directly embedding an image. However, a proposal to allow fair use material, at Wikidata:Requests for comment/Non-free content, doesn't seem to have generated a lot of enthusiasm. Ghouston (talk) 03:38, 11 October 2019 (UTC)
  • I suspect you're thinking of Commons compatible image available at URL (P4765) and non-free artwork image URL (P6500). I'm not convinced that they are a good idea, though - they seem like maintenance burdens (e.g., when URLs change) that would be time better spent maintaining images on commons... Thanks. Mike Peel (talk) 03:57, 11 October 2019 (UTC)

"Duplicate references" has not been working for 14 days[edit]

Does anyone have any idea when the problem will be solved? HarryNº2 (talk) 23:46, 10 October 2019 (UTC)

There’s a bug opened in Phabricator, but as far as I can tell there’s no ETA and no one committed to fixing it, alas. Really crippling my work. - PKM (talk) 01:30, 11 October 2019 (UTC) ·
Thank you for your answer. At the moment I do not feel like creating new articles or adding missing documents. I can only hope that someone will quickly solve the problem. HarryNº2 (talk) 21:37, 11 October 2019 (UTC)
@HarryNº2: Fixed! - PKM (talk) 19:42, 15 October 2019 (UTC)

Semantic Scholar author ID[edit]

In the last iterations of multiple items with the same ORCID identifier, i find that many errors co-occur with errors in the Semantic Scholar author ID. I do remove these identifiers but what prevents re-occurence? Thanks, GerardM (talk) 08:43, 11 October 2019 (UTC)

@Magnus Manske:.--GZWDer (talk) 10:39, 11 October 2019 (UTC)
Why do you remove the identifiers? The optimal solution is often to merge the items. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:25, 11 October 2019 (UTC)
I remove them because I am not getting side lined from fixing ORCID problems. Thanks, GerardM (talk) 15:53, 12 October 2019 (UTC)
Then just leave things alone. Otherwise, you are damaging Wikidata; and making it harder for other people to improve it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:18, 12 October 2019 (UTC)
There are multiple possible cases here. In some cases merging is the way to go. In others where an identifier is wrong, the solution is to deprecate the wrong identifier. If you don't deprecate will likely be refilled sooner or later by a bot. ChristianKl❫ 07:32, 14 October 2019 (UTC)

unknown values for people who have long-since died[edit]

Today, via a post on Reddit, it was highlighted that the google knowledge graph results for the ancient Roman senator Lucius Pinarius (Q382127) declares him to be 2085 years old. [Google result:]. Obviously, this is the result of a Google querying Wikidata, finding a birth-date and no death-date, and making an age-calculation up to the present day. The way for Google to solve this would be to add a bit of logic that ignores the age calculation if there is no death date listed (ping: Denny).

But the way for Wikidata to solve this, and potentially many other cases like it... I suggest to:
Add an "date of death (P570) -> Unknown Value" statement to all the instance of (P31) -> human (Q5) who HAVE a date of birth (P569) before 1900 AND have no date of death (P570) statement.
[Note: I have already done this for Lucius Pinarius (Q382127) so the Google graph will update eventually]

This would apply to:

Is this a good idea?

[Credit to Fuzheado for making the sparql query!] Thanks, Wittylama (talk) 11:09, 11 October 2019 (UTC)

  • No. Why not add actual values instead? 11:13, 11 October 2019 (UTC)--- Jura
If the actual death-date were known, then yes of course that should be added. But in the absence of the specific information for each person we could at least say what we know: that they ARE dead - we just don't know when. Wittylama (talk) 11:29, 11 October 2019 (UTC)
It's also obvious that they have a place of birth, a place of death, a mother, a father, a place of burial we might not have included yet. Would you check sitelinks and listed references first? --- Jura 11:33, 11 October 2019 (UTC)
I suppose we could be adding 'unknown value' to those other things too - technically its true, as you say. But I agree with your implication - that doing that would be overkill. My thinking is that because we have actively said there's a start-date (birth), we should make some acknowledgement of the end-date (death) - for this MOST human of facts [as they say, "...death and taxes!"].
The second part of your question - what kind of checks could/should be done before mass-adding such a statement - gets to the heart of the matter. I certainly wouldn't be manually researching each person! And this is why I broke the research into those time-groupings, because we might feel differently about adding this statement for people born 2000 BCE as opposed to born in 1899. I suspect adding the 'unknown value' statement could also serve as a case of m:Cunningham's law - by overtly saying we don't know, might encourage some people to find and include a referenced fact in some cases. Wittylama (talk) 11:44, 11 October 2019 (UTC)
That really is true for many other things as well. Currently, in most approaches, we stop looking for it once someone determined it's unknown. So we just end up closing the door for incremental additions. --- Jura 11:52, 11 October 2019 (UTC)
It would be helpful if the GUI rendered this as some value (the meaning it is given in the underlying software, and the usual meaning of the blank node we use to represent it in RDF), rather than unknown. This special value does not necessarily mean that it is known that nobody anywhere knows the value. It means that we know that the statement has some value, but that (for some reason) we are not currently able to state it. That can include there not presently being an item for the value, hence common idioms such as spouse (P26) = some value, stated as (P1932) = "Elizabeth Smith". It would be helpful if the software did not represent "some value" as unknown, because this can often be misleading as to what we really mean by "some value". Jheald (talk) 12:28, 12 October 2019 (UTC)

Re. "unknown value" it has always been somewhat unclear how to use it:

  • Only if available sources concordantly and explicitly state that this information is generally not known to the public?
  • Or if you as a Wikidata editor have done some thorough research and did not find any public source which reports this information?

IMO, "date of death: unknown value" is still better than no "date of death" statement at all, as the latter somehow implies that the person might still be living. However, I would recommend not to batch-add such statements to items, as the "thorough research" for possibly available information cannot be automated. —MisterSynergy (talk) 12:07, 11 October 2019 (UTC)

In the past, I used century precision for a date of death around the date of birth sometimes. There is pros and cons to do stuffs like that, it may help some queries for example (if, say, we want to sort accourding to a date of birth/death, it’s better to use a fuzzy date than « unknown value » in some cases.)author  TomT0m / talk page 12:46, 11 October 2019 (UTC)
If the associated Wikipedia articles mention it as "unknown", I think adding a century precision DOD is a good idea. However, we currently do lack a process to attempt to add more precision to dates that don't have day-precision. --- Jura 16:16, 11 October 2019 (UTC)

If this were a real knowledge database the human item would include a max age, and the program would infer the incorrectness of its 2000 year age deduction from it. --SCIdude (talk) 16:48, 11 October 2019 (UTC)

  • Symbol support vote.svg Support date_of_death=unknown_value for people over 110 years old. We have a list of supercentenarians, which is everyone over 110 years old. We can flag the known supercentenarians so that they do not get date_of_death=unknown_value. We can find most dates of death for USA people who died between 1900 and 2015 at Familysearch which has most state death records. It also has federal death records from about 1970 to the present. --RAN (talk) 20:36, 11 October 2019 (UTC)
  • I used 120 years on Robert Bruce Douglas (Q56604483), with some qualifiers. Ghouston (talk) 09:16, 12 October 2019 (UTC)
    • Date of birth or date of death should never ever be “unknown”. It should be some date with low precision. Multichill (talk) 12:13, 12 October 2019 (UTC)
Hmm, the same would be true for the place of birth, because setting it to planet earth is always true. At the end it boils down to the question to which granularity something is still "unknown". For example, if somebody was born in Germany, the exact place of birth might still be "unknown". Going even further, for somebody born in Berlin, the exact district can be "unknown". Steak (talk) 12:39, 12 October 2019 (UTC)
  • If we do this - I'm not sure it's a good idea - I think we need to make sure we can distinguish between "actually unknown value" and "we simply don't have one yet, so we've put unknown in for the time being". Perhaps a sourcing-circumstances qualifier to say "automatically applied due to a long-ago birthdate", or something? Andrew Gray (talk) 13:01, 12 October 2019 (UTC)
    • I mentioned this proposal to Multichill, and his point was allied to your concern Andrew... he noted that a birth/death date should never ever be “unknown". Rather, it should be a date with low precision. That is, to avoid the question you raise (and also Jura1's earlier point about the potential for shutting down incremental improvements), it's better to give the most accurate knowable precision rather than to declare that it's not known. If you think that's a good point - and it sounds quite accurate to me - the question is, can that be turned into a bot editing task that does something like adding a death date of "no later than" [120 years after birthdate]? Wittylama (talk) 22:03, 12 October 2019 (UTC)
      • Yes and no. I think that decade or century inaccurate-precision dates are okay, but millennium-precision ones (barring deep historical dates like early Egyptian figures!) should probably be avoided - they will just confuse queries that aren't set up for complex date filtering.
      • For "no earlier than"/"no later than" dates, I presume these would be "unknown" w/ a latest date (P1326) or similar qualifier? If so, I think that would work okay. But we would still need some kind of "automatically inferred from..." sourcing qualifier. Otherwise it will become an incredible mess in the future, as we are left with trying to figure out if it's automatically calculated, or if some researcher actually went off and determined it was no later than that date for (some other reason). Andrew Gray (talk) 22:26, 12 October 2019 (UTC)
    • I'm not sure if there's any real difference between "actually unknown value" and "we simply don't have one yet, so we've put unknown in for the time being". What does "actually unknown value" mean? There's always the possibility that the information is recorded somewhere, on a piece of paper in an archive or a buried stone tablet. I'm also not sure how you use a "low precision date" when the date in question is near a millennial boundary, e.g., if we aren't sure if somebody died in 1999 or 2000 (or 2000 and 2001, depending on how it's defined). Ghouston (talk) 00:55, 13 October 2019 (UTC)
      • @Ghouston: I would say that "actually unknown" is where the external sources have said "we have looked and we don't know". For example, people who outright disappeared, or historical figures who stop appearing in the records after their period of notability. Yes, it's possible that a new date might be found sometime, but for the time being we can treat it as a "known unknown".
      • That's different from someone whose date of death is presumably a matter of historical record, but where we haven't imported it yet for one reason or another. EnWP models this with categories for "year of death unknown" and "year of death missing", which is a reasonably good solution. That doesn't work for us (because missing, no P570, can also equal "still alive") but I do think we need to distinguish these two cases.
      • For the case where someone died in 1999-2001 but we're not sure when, then year=2000 + sourcing circumstances "circa" would seem more useful than millennium-precision dates. Andrew Gray (talk) 10:47, 13 October 2019 (UTC)
        • 1999/2000 perhaps isn't a great example, but the problem is still likely to occur, e.g., with somebody born in 1900, with the year increasing as time goes by. Also, a date like "2nd millennium" seems pretty strange for somebody from the 19th century, while "unknown value" is accurate as long as we haven't been able to discover a date. Whether somebody else has also looked into it, and written "I haven't been able to discover a date" on a website somewhere, doesn't seem to make much difference to me. Ghouston (talk) 05:52, 14 October 2019 (UTC)
        • Since the currently oldest recognized living person was born in 1903, it would seem safe to declare anybody born in 1902 or earlier to be dead. However, I guess the most likely error with this approach would be if people have incorrect birth dates. Ghouston (talk) 06:21, 14 October 2019 (UTC)
  • It's up to the data-user to decide whether he wants to count a person where we don't know anything about their death as alive or dead. I see no reason why we should make a massive amount of unsourced entries in our own database with "unknown value".
There's some value on being able to list all people who died in the 3th century and for that reason dates with century precision can sometimes be useful. ChristianKl❫ 18:57, 14 October 2019 (UTC)

pq: reason for preferred rank[edit]

We have reason for deprecation (P2241) as a qualifier enabling us to specify why a statement has been given the deprecated rank. We have no similar property, afaics, to specify why a statement has been given the preferred rank.

Anyone else think that should be rectified? --Tagishsimon (talk) 13:55, 11 October 2019 (UTC)

Symbol support vote.svg Support Seems useful to me because it explicitly marks the motivation for why a statement is preferred, which helps future editors decide whether it's non-destructive to change the preferred statement. --SilentSpike (talk) 14:58, 11 October 2019 (UTC)
sure. even if the value would be "editorial choice". --- Jura 16:17, 11 October 2019 (UTC)
Yup, that's where I was coming from w.r.t. coordinates. --Tagishsimon (talk) 02:12, 12 October 2019 (UTC)
Symbol support vote.svg Support Makes sense, and maybe it would nudge the football guys to rethink how they're using deprecation on names to make their wikipedia infoboxes look pretty ;-) Moebeus (talk) 02:09, 12 October 2019 (UTC)

See Wikidata:Property_proposal/Generic#reason_for_preferred_rank. Feel free to improve/correct the property proposal - property proposals not my speciality. --Tagishsimon (talk) 03:13, 12 October 2019 (UTC)

Question re: portraits modeled after different people[edit]

This question is re: Jerónimo Pérez (Q70298355). The current image is a painting theoretically of the friar, but it was painted after the friar's death and likely actually modeled off of a different person living in the friary (perhaps a Juan de Herrera). Is it appropriate to keep this picture on the page of the friar? I assume yes in the same way that we have images of other historical persons that don't actually represent them (e.g. Jesus Christ (Q302)). But is there any way (qualifier?) to indicate that this painting is not a literal likeness of Jerónimo Pérez (Q70298355)? Calliopejen1 (talk) 18:35, 11 October 2019 (UTC)

@Calliopejen1: related image (P6802) exists, though I’m not sure if it’s more appropriate in this case. --Lucas Werkmeister (talk) 11:35, 14 October 2019 (UTC)

Wikidata turker[edit]

Hello, I am a regular french Wikipédia editor and each time I move a page in Wikipedia (french version), the change of the title's page is registered in the corresponding Wikidata's item (e.g. : "Startup Studio" to "Startup studio", 21:52, 30 August 2019). The frWiki sitelink of the item's page is updated. As far as I know, this is a feature of the Wikibase. But I do not want to be credited for the automatic change in Wikidata. I disagree to release my contributions under the Creative Commons CC0 License and I simply do not want to contribute to the Wikidata project. Is there a Wikipedia account's option to stop to be forced to contribute to Wikidata while contributing to Wikipedia or to stop to be credited for an automatic Wikidata's update made by the system ? --ContributorQ (talk) 18:40, 11 October 2019 (UTC)

No, there is no such option. If you edit Wikimedia projects and do stuff that affects other projects as well, as changes of interwikilinks due to page moves do, then you do not have the option to opt-out of these edits. You would otherwise leave outdated mess for interwiki users (i.e. all other Wikipedia projects). Please also mind that organizing interwikilinks usually does not create a copyright-able work anyway, so these sort of edits are in the public domain (or very close to it) anyways. —MisterSynergy (talk) 18:53, 11 October 2019 (UTC)
Thanks for your answer. So, unfortunately, such an option does not exist. But my point is that I do not want to be credited for an automatic change made by the system itself. It is like a bot using my own signature/account. I chose to join the Wikipedia project and I do not want to take part in the Wikidata project. I do think that to be forced to be credited on the Wikidata project while working only for Wikipedia is a problem.
Note: when one directly makes an interwikilink's change in a Wikidata's page, the system displays the message "By clicking "publish", you agree to the terms of use, and you irrevocably agree to release your contribution under the Creative Commons CC0 License." --ContributorQ (talk) 13:02, 13 October 2019 (UTC)
Okay I see your position, but I really cannot follow you regarding why you consider this process to be problematic. Wikipedia interwikilinks are managed at Wikidata since almost 7 years, so it is sometimes inevitable to make contact with this project as a Wikipedia editor as well. There is really no disadvantage for you if you accumulate some (automatic) edits in this project; also, you do not release your contribution under a license that is less restrictive than necessary as interwiki relations are not copyright-able as a matter of principle; furthermore, you agree to the ToS with each and every edit in Wikipedia anyways. So what's the thing here…? —MisterSynergy (talk) 13:37, 13 October 2019 (UTC)
@ContributorQ: I don't recall seeing that message on Wikidata, but it was clear: Don't press publish if you disagree. User contributions are optional. Now you know that some frWp changes cause a change in Wikidata. In your case, maybe you can ask a bot to do the change for you or ask on talk page for someone else to do the move. Aside from that, I would agree that from license point of view there is nothing to worry about. You can consider the changes to be credited to a bot and your name is there just for reference to the originating change on frWp. Is that ok? You can point to this discussion if someone accused you of being active on wikidata. Jagulin (talk) 06:20, 15 October 2019 (UTC)
Do you imagine that there is any jurisdiction under which you can copyright the act of moving "Startup Studio" to "Startup studio"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:43, 13 October 2019 (UTC)
@Pigsonthewing: I don't see where this question is leading. What if I say "yes, I can envision that"? I don't think the question was about copyright, but contribution and credits. Are you saying that contribution and credit only covers copyrighted material? The ToS normally mention the CC-BY-SA rather than CC0. If the change history had a tag saying "automated change, no copyright claimed" it would have been more clear which parts that are CC0 and maybe the user would accept that more easily. Jagulin (talk) 06:20, 15 October 2019 (UTC)
The OP includes the text "I disagree to release my contributions under the Creative Commons CC0". If that's not about copyright, I'm a banana. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:00, 15 October 2019 (UTC)
I understand how you peel, but regardless of copyright there is something attributed to the username (as if there is something to to waive) and "contributed". Another quote is "I do not want to take part in the Wikidata project", showing that the focus for the OP was no avoid having their name in the logs, regardless of license. Jagulin (talk) 21:51, 15 October 2019 (UTC)

Why are unpatrolled changes rendered immediately at Wikipedia?[edit]

Hello. In this edit to Q592 (gay), 2620:22:4000:1202:1fa1:f76b:c060:2f54 (talkcontribslogs) vandalized the English description. The diff page shows a bracketed hyperlink, "Mark as patrolled", which implies to me that this edit has not yet been patrolled, so, presumably, it's a low-confidence change, or not ready for prime time, or something of that nature; is that right? Nevertheless, the Wikipedia page w:Gay was showing the vandalized description as the "short description" at the top of the page. Can unpatrolled edits be blocked from being ported over to descriptions, infoboxes, or anywhere they might appear in sister projects, until they are patrolled?

I'm uncertain if this is a Wikidata or a Wikipedia issue; if the latter, please let me know, and I'll move this discussion. Thanks, (ping, please) Mathglot (talk) 19:31, 11 October 2019 (UTC)

Mathglot: Hi. We should contact the development team to know whether such restriction can be applied or not. If it is possible, pros and cons should be carefully analized. And this is a Wikidata matter, indeed. Esteban16 (talk) 00:07, 12 October 2019 (UTC)
This feature is called pending changes. I doubt that there are enough patrollers on Wikidata to check everything, though. Ghouston (talk) 09:01, 12 October 2019 (UTC)

Oh, didn't know that extension could be used in this case. But like said above, it would require a tremendous amount of help. --Esteban16 (talk) 13:19, 12 October 2019 (UTC)

@Esteban16, Ghouston:, thanks for your comments. Yes, I'm familiar with pcr on Wikipedia, and not infrequently review changes myself. Since I'm not nearly as familiar with Wikidata as either of you, would one of you mind taking the lead on this, as far as taking it to the dev team? If you'd prefer I do it, I'm willing; but it sounds like you know much more about the situation than I do. I totally believe you that there aren't enough patrollers on Wikidata, but that would mean that mostly autopatrolled changes would get ported to Wikipedia. I don't know the ratio of (edits by active, non-autopatrolled Wikidata editors / edits by active wikidata editors); if it is not huge, maybe having unpatrolled changes blocked from porting to Wikipedia is not too large a price to pay, even if some percentage of unpatrolled edits are good ones. What do you think?
Also, there must be some cross-project interest in this discussion from the Wikipedia side as well; I'll see if I can come up with the right project or noticeboard, and alert them to this discussion. Mathglot (talk) 00:59, 13 October 2019 (UTC)
Patrolling edits on Wikidata can be difficult, since changes to labels and descriptions can be in a multitude of languages. If one edit isn't patrolled, then subsequent edits would also be backlogged as unpatrolled. It would also break site links when pages on Wikipedia are renamed, since the related Wikidata edits are often by users who aren't autopatrolled. Some bad edits on Wikidata are probably found when they are displayed on other wikis, and blocking that process would presumably make Wikidata worse, not better (it's likely that unpartrolled edits would just sit around, until somebody notices the backlog and patrols them en masse, just so that their own edits can go through). Ghouston (talk) 01:11, 13 October 2019 (UTC)
So, it would seem like the focus should be on blocking unpatrolled edits from appearing on Sister projects then; do you agree? Mathglot (talk) 01:13, 13 October 2019 (UTC)
No, since sister projects would then tend to display old data, which would not necessarily be any more correct than new data. It may be outdated, for example. Ghouston (talk) 01:31, 13 October 2019 (UTC)
Why do you say that old data wouldn't be more correct? Don't we rely more on autopatrolled users, than newer ones? Otherwise, what is the point of even having this distinction among users? As a Wikipedia user, I would rather see formerly correct data that is now possibly old, than up-to-the-minute vandalism. What am I missing? Mathglot (talk) 01:40, 13 October 2019 (UTC)
An item would stop updating once it has been edited by a new user, as I understand it. They may have just changed a sitelink because they renamed a page on Wikipedia, so it's a perfectly good edit. All subsequent edits would also be ignored, since you'd have to take the last version with no unpatrolled edits. You'd be throwing out a lot of perfectly good edits just to avoid one perfectly good unpatrolled edit (or vandalism, maybe in some language you don't even care about, as the case may be). Ghouston (talk) 01:48, 13 October 2019 (UTC)
Thanks for all your feedback. We need more eyeballs on this; I've notified WP:VPT and relevant WikiProjects at Wikipedia. Mathglot (talk) 02:26, 13 October 2019 (UTC)
Surely the simple solution is not to import Wikidata descriptions into a language Wikipedia? Keeping each short description on the local Wiki allows local Wikipedians to curate them there. There's nothing lost by not using the Wikidata description as the version in a given language isn't used in more than one language Wikipedia. --RexxS (talk) 02:41, 13 October 2019 (UTC)
At first glance, I could support that approach, but I know little about this topic so far. As something that would affect the interaction between multiple sister projects, this deserves airing at a venue where it can elicit thorough discussion by numerous concerned editors from multiple projects, and I'm beginning to wonder if this is the right venue. Perhaps at Meta? Mathglot (talk) 18:10, 13 October 2019 (UTC)
Here it is probably still better than Meta in terms of engagement. However, different projects use vastly different strategies of interacting with Wikidata, and I am not sure there is one solution which would be good for everybody. Then discussing it separately at every project might be an even better strategy, though right now any discussion at the English Wikipedia concerning Wikidata has a good chance to be closed as "Never ever any interaction possible" and will likely not be productive.--Ymblanter (talk) 18:21, 13 October 2019 (UTC)
@Ymblanter:, can you disambiguate your quoted remark: does that mean, 1) Discussions between wp and wd are impossible, 2) The data on the two projects should not interact/be imported, 3) Something else? Thanks, Mathglot (talk) 18:49, 13 October 2019 (UTC)
I meant "We do not want any data from Wikidata, and ideally we do not even want to hear about Wikidata anymore". Note that this is not my personal opinion but my experience of participation in many discussions over there.--Ymblanter (talk) 18:51, 13 October 2019 (UTC)
@Ymblanter:, Okay, thanks. It sounds like there's a lot of history here that I'm unaware of. Can you perhaps drop me a link or two here or at my Talk page, which will help me come up to speed on past discussions about this? It also explains RexxS's somewhat exasperated response at VPT, which I now understand better; RexxS: I apologize if I barged into a touchy subject without enough background. Just trying to do the right thing at WP with respect to shortdesc. Sounds like I have a lot of reading to do. Thanks, Mathglot (talk) 19:06, 13 October 2019 (UTC)
This gives a good flavor how these discussions typically run. There was something more contentious in the beginning of 2018, but I can not easily find it now. Arguments [ here[ also give an impression.--Ymblanter (talk) 19:19, 13 October 2019 (UTC)
(edit conflict) Thanks, Ymblanter. Mathglot (talk) 20:59, 13 October 2019 (UTC)
Bear in mind that Commons also uses the descriptions from Wikidata (via the infobox) - the use of descriptions isn't something exclusive to Wikipedias. Thanks. Mike Peel (talk) 08:27, 14 October 2019 (UTC)
As far as I understand the decision against not showing unpatrolled edits is to allow Wikipedia users who might not be autoconfirmed on Wikidata fix errors in Wikipedia infoboxes that display Wikidata content and display the new information immediately. ChristianKl❫ 20:51, 13 October 2019 (UTC)
If they were autoconf on Wikipedia, couldn't they simply override it there, by adding param=value in the Infobox, rather than coming here? Mathglot (talk) 21:01, 13 October 2019 (UTC)
@Lydia Pintscher (WMDE):, I think you argued last time for the status quo. ChristianKl❫ 07:02, 14 October 2019 (UTC)
Yeah. Mathglot: They could but that is really not what we should be striving for as a movement. We need to work together and not everyone doing their own thing. If we continue down the road of everyone doing their own thing instead of fixing something that's wrong for everyone then we will continue to massively disadvantage medium and smaller languages and wikis. That should not be ok for the movement because it goes against our vision. --Lydia Pintscher (WMDE) (talk) 07:53, 14 October 2019 (UTC)
So, what is the right approach here, then? Mathglot (talk) 22:47, 14 October 2019 (UTC)

Award rationals[edit]

Is "award rationale" meant to be a direct quote from the award giver or is it meant to be as terse as possible? I went for the quote at Francis Sherman Currey (Q5482358). --RAN (talk) 20:09, 11 October 2019 (UTC)

I've assumed that it's supposed to be a quote, although I'm not sure how long a rationale may be before it potentially leads to copyright problems. Ghouston (talk) 09:05, 12 October 2019 (UTC)
Does it make sense to have them when there it can also be the reference ?? GerardM (talk) 16:22, 12 October 2019 (UTC)

Auto-notifying users when their username is added to a wikidata item[edit]

Moved from Wikidata:Project_chat/Archive/2019/10: T.Shafee(evo&evo) (talk) 00:16, 12 October 2019 (UTC)

Apologies for having let this thread time out into the archive. I think that it bears a bit more discussion in case there is an automatable solution. I've been thinking about this from entering Wikimedian in Residence info into wikidata (see table in progress), where the WiRs either associate their real name with their account, or sometimes are known only by their username. Non-consentual outing of users seems like something that users should be informed of by some mechanisms @Jura: Is there a mechanism to email users through the address associated with their account? Perhaps the notifications tab? T.Shafee(evo&evo) (talk) 00:16, 12 October 2019 (UTC)

@Evolution and evolvability: I question the notability of these people. Besides that you'll make some people very uncomfortable by violating their privacy or at least making it very easy for others to violate privacy. Multichill (talk) 10:29, 12 October 2019 (UTC)
I see no use for this feature to exist. Unless there is vandalism, you know you are associated with a real world issue, why do you need to know if you get named on it's online item.
And as to Multichill's notability concern, I take a different stance. I think the "questionable notablity people" would be the ones adding their own names to items, they wouldn't be concerned with being notified because they already know (they would want to know when it gets removed). However, there are instances (far and few between), where people do manage their own "press". Not every one is Lady Gaga with a team of publicists. For example, in my family, I have a published educator, published doctor, and a recorded musician. If one of their "works" get added, I would love to help populate information. I search. I follow by adding to my watchlist. I use "what links here". I think "what links here" serves the same purpose without the notification. Quakewoody (talk)
Many of the people on the page headed "Questionable notability Wikimedians" meet our notability criteria easily (disclosure: I am one of them) - and a number of them have been kept by consensus, after deletion proposals. They include winners of significant awards, and authors of academic papers and books cited on Wikipedia and in academia. Labelling people as such is a violation of our BLP policy. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:25, 12 October 2019 (UTC)
I guess I'm coming from the point of view of hoping that such users would be able to then add additional information to the item when notified of its existance. For wikimedians in residence, it' be useful to have info on sex, nationality, profession etc, but that info is often difficult to come by. However the subject may be better placed to add it. E.g. This query is still sparsely populated. T.Shafee(evo&evo) (talk) 01:34, 13 October 2019 (UTC)

Wikidata requirements[edit]

I just reinstalled my browser and Wikidata doesn't works correctly. There is nothing under statements and If I want to edit labels it redirects me to page with description "You may have reached this special page because the item you tried to edit wasn't fully loaded for label/description edits to work there." Is it problem related to browser cookies, frameworks, Java or Flash Player? Eurohunter (talk) 14:20, 12 October 2019 (UTC)

That special page is shown to users who have an outdated browser. Any browser older than the ones specified as "modern" on mw:Compatibility#Browser support matrix are considered outdated.--Snaevar (talk) 14:56, 12 October 2019 (UTC)
When reporting browser-related issues, please state your browser, its version, and your operating system. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:21, 12 October 2019 (UTC)

Merged but not attached[edit]

I noticed that recently I merged a couple of items, but the transferred site links did not seem to be attached to wikidata anymore. For example Q9827687 I merged into Category:Social economy (Q7214799), but now oc:Categoria:Economia sociala seems not attached to wikidata anymore. Or Q10767336 I merged into Category:Ditrema (Q9610495) but zh:Category:海鮒屬 does not seem to be attached. Is this to be repaired in some way? Can something be done manually? Does it just need patience? Lymantria (talk) 15:46, 12 October 2019 (UTC)

Just add a sitelink to the WD item to re-attach it. --SCIdude (talk) 15:54, 12 October 2019 (UTC)
Oh, it the site links are listed at the wikidata item, but when you visit them, there is no sign of a wikidata item. See for instance Lymantria (talk) 16:00, 12 October 2019 (UTC)
I removed the sitelink, then added it again. Now it's there. No idea about the cause. --SCIdude (talk) 06:58, 13 October 2019 (UTC)
I've been finding a number of cases like this recently - items are merged, and although it looks fine on Wikidata, the information from (interwikis or content fetched through the sitelink) it no longer shows up on Commons/enwp. Any edit (e.g., adding 'test' to the labels, or removing a property) seems to clear it up. It doesn't seem to be any particular user, and I think the oldest case I found yesterday was from the 26th September (Category:İlkadım District (Q32709947)). Some other examples were Category:Polish books (Q10116410), Category:November-class submarines (Q6207422), and Category:2013 papal conclave (Q6215728) Thanks. Mike Peel (talk) 07:21, 13 October 2019 (UTC)

Deprecated ORCID identifiers[edit]

I asked attention of ORCID for our duplicate ORCID identifiers.. Many of them have been deprecated. this is an example. Is it possible to identify deprecated ORCID identifiers and remove them? Thanks, GerardM (talk) 15:55, 12 October 2019 (UTC)

Yes, it is possible. No, it should not be done. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:19, 12 October 2019 (UTC)
For deprecated identifiers they should be set as deprecated in Wikidata.--GZWDer (talk) 16:50, 12 October 2019 (UTC)
Please make sure that the functionality that reports of errors takes them into account and, PLEASE what is the point, these indicators can be found at source. Thanks, GerardM (talk) 13:26, 13 October 2019 (UTC)

Adopt a user program in Wikidata[edit]


I am interested in Wikidata and I have looked to some filters and there I saw that the number of editors with a little amount of edits is big. I have thinked about how it was for me to start editing in Wikidata and for me it was at the begin not so easy. I think it were great if there is a program like the Mentorenprogramm in the German Wikipedia or the Program Adopt-a-user in the English Wikipedia to help people start Editing Wikidata. What do you think about starting a Project like the mentioned ones here in Wikidata. -- Hogü-456 (talk) 19:40, 12 October 2019 (UTC)

Many edits is probably just a sign that someone knows how to write program code. Many might not want this. --SCIdude (talk) 03:55, 13 October 2019 (UTC)
That depends on what you mean with many edits. A user that makes 10,000s of edits is likely writing some code. We however have a lot more users that make 5 edits/month then we have users who make 100/month. ChristianKl❫ 07:59, 15 October 2019 (UTC)
I like the idea of a user mentorship program, if not (per SCIdude) for the specific reason of increasing the quantity of a user's edits. Mahir256 (talk) 04:17, 13 October 2019 (UTC)
This is a good idea. Are you interested in participating in a project like that @Mahir256: and help people by specific questions in getting started with Wikidata. -- Hogü-456 (talk) 14:07, 13 October 2019 (UTC)
I'm not familiar with the mentoring done, would you actively push the users to come back and contribute or be available only if they ask? I got a welcome note with starter links, I think that was nice and interpreted that as having someone to turn to if needed. On the other hand I think there are chats for those asking for help. Proactively improving the existing documentation and tutorials is a good way to lower the threshold, so reviewing beginner FAQ could be a project start. In many ways I think most users come here with the assumption that WD supports their language wiki needs rather than having a goal of itself. Clarifying that balance I think is important to foster quality contributions, but I also think that WD currently doesn't have the answer. Having the "mentoring project" at the Wp rather than in WD, I think would be a good way of showing how WD improve Wp, and I assume a lot of that mentoring is already managed that way. Jagulin (talk) 22:24, 15 October 2019 (UTC)

Inappropriate (Wikisource) badges added to Wikipedia articles[edit]


Lately, a lot of wikisource badges have been added to wikipedia articles ; those are inappropriate because they just measure the correction level of a text, not a decision of the community.

Very often, these addings have been made at the same time as the link was added, thus making it impossible to undo the edit without removing the link at the same time, and making it very long to remove the badge from each link, and impossible to add an explanation for the contributor to prevent further error.

The concerned badges are :

This has been first discussed by the fr community, and many wrong badges have been removed, especially on frwp, but it is not a typically fr problem, and should be solved on the general scale.

[edit] : there seems to be a real misunderstanding of what badges stand for, as on Tallinn (Q1770), the hawp link was added 4 (!!) badges, 2 of them being wikisource badges… (which I removed, without touching the other badges, not being able to judge of the relevance). - same here on lesbian (Q6649), and 6 !!!! on Samira Moussa

At least, Special:SetSiteLink page should contain a message about which links to add and why, instead of blindly invite users to add them...

Would it be possible to put constraints for badge additions, so that badges that were made for wikisource could not be added to wikipedia links (and the other way round too). - Thanks for your help on this. --Hsarrazin (talk) 09:37, 13 October 2019 (UTC)

No, I do not think this is possible. I do not see how a check could be run whether it is a correct badge or not. On the other hand, only autoconfirmed users may add badges, and I guess there are not so many users of the French Wikisource who are autoconfirmed at Wikidata, it might be easier to leave all of them a personal message explaining what went wrong.--Ymblanter (talk) 18:15, 13 October 2019 (UTC)
  • I fixed all remaining cases of Wikisource-only badges on non-Wikisource sitelinks.
  • Mind that there is another Wikisource-only badge which you have not listed above: digital document (Q28064618)
  • On a side note: there is a badge type good list (Q51759403) for ptwiki and maybe eswiki, but frwikivoyage does use it quite a bit as well (query). Is there such a page status defined in frwikivoyage at all? fr-N user Hsarrazin could maybe help me here :-)
  • There is some more stuff related badges available on a page in my user space: User:MisterSynergy/misc/sitelink badges

MisterSynergy (talk) 08:26, 14 October 2019 (UTC)

Thanks for finishing the cleanup MisterSynergy, - my personal final query for detecting them all is now - thanks for your queries on badges Face-smile.svg
I was not aware of digital document (Q28064618) and would not know how to use it... do you know who created it, in what context ? I think it was perhaps enwikisource ?
the problem is not of Frwikisource users wrongly adding badges (I'm probably the only one doing so, using a tool specifically designed to retrieve it from frwikisource)
the problem is wikisource text correction badges (indicators) being used on wikipedia links (or list, categories, and I even templates), by wikipedia users who do not understand those badges. I've tried to leave a message on each of those users personal page, but it is not very easy when they speak russian, arabic, chinese, etc...
I thought perhaps a constraint on those badges, excluding wikipedia links, or an abuse-filter could detect those wrong additions (the specific values of these badges) used in "wikipedia group" links ? --Hsarrazin (talk) 08:50, 14 October 2019 (UTC)
  • Yes, digital document (Q28064618) seems to be somewhat enwikisource-specific per this query. It was defined per request at phab:T153186.
  • User:Abián created another phabricator task at phab:T235374 yesterday to request that only applicable badges should be displayed in the UI. If that was implemented, it would already be quite helpful.
  • I wrote a small script yesterday to batch-remove wrong Wikisource-only badges from non-Wikisource sitelinks, including a descriptive edit summary. I could run it occasionally if nobody wants to set up a continuous bot job.
MisterSynergy (talk) 09:06, 14 October 2019 (UTC)
I knew a script could do it, but I can't write one, so I must have cleaned up hundreds of them by hand yesterday, and let some messages on contributors pages Face-wink.svg --Hsarrazin (talk) 09:44, 14 October 2019 (UTC)
(ec.)@MisterSynergy: Wondering how to translate digital document (Q28064618) in German? It looks like the only existing badge that miss German translation. --Liuxinyu970226 (talk) 23:37, 14 October 2019 (UTC)
Thanks, now it has German label+description. —MisterSynergy (talk) 06:10, 15 October 2019 (UTC)

Badges for sitelinks to redirects[edit]

What is the process for creating new badges?

Per the community comments at Wikidata:Requests for comment/Allow the creation of links to redirects in Wikidata, it would be good for sitelinks to redirects to be identified, ideally distinguishing intentional redirects (those marked with Template:Wikidata redirect (Q16956589) on the Wikipedia) from redirects not so marked (which may be left-overs from merges, and may indicate duplicated items here that should be merged).

Would it be straightforward to create a pair of badges for these, ideally kept updated by a regularly-running bot? Jheald (talk) 09:25, 14 October 2019 (UTC)

this should be asked on --Hsarrazin (talk) 09:48, 14 October 2019 (UTC)
Done. Thanks. Phab:T235420 "Create wikidata badges to indicate when sitelinks point to Wikipedia redirect pages" now created. Jheald (talk) 11:26, 14 October 2019 (UTC)
@MisterSynergy: If/when such badges do get created, would it be easy for you to write a bot to keep them updated? Jheald (talk) 11:31, 14 October 2019 (UTC)
Lydia has responded on Phab that "Adding two badges is easy enough. We just need the name and the icon for it." [emphasis added]
I have therefore created two items, sitelink to redirect (Q70893996) and intentional sitelink to redirect (Q70894304) for these badges, where we can start to assemble translations.
Icons next to the links in the sidebar will be very very small, and therefore need to be quite simple. I would suggest that possible might be:
Symbol redirect arrow grey.svg   English
Symbol redirect arrow blue.svg   English
for redirects, and 'intentional' redirects (ie redirect pages marked with Template:Wikidata redirect (Q16956589)), respectively. Yes, the grey and the blue are almost indistinguishable, but to most Wikipedia readers the distinction will hardly matter.
In the links section of wikidata item pages we could use
Charlie Reid (musician)   Symbol redirect class.svg
Charlie Reid (musician)   Symbol redirect blue.svg
These badges would be straightforward to pick up with WDQS queries, in the same way that we can currently pick up featured articles etc.
Does this seem viable?
Other potential icons can be found in c:Category:Redirect_arrows, c:Category:Norro style 1 icons (arrows), etc., if people would prefer somthing like
Redirect arrow without text (cropped).svg   English
for the sidebar, or
Charlie Reid (musician)   Symbol redirect vote blue.svg
for the wikidata page (or something quite different).
The other redirect icon is also available off-the-peg in red,
Charlie Reid (musician)   Symbol redirect vote 4.svg
if people want something more attention-seeking, since redirects that are not "intended" can often indicate unresolved duplication here. I have thought to go for the less strident grey, as not all wikipedias offer the Template:Wikidata redirect (Q16956589) template; but what do people think? Jheald (talk) 16:41, 14 October 2019 (UTC)
Alternatively, perhaps we're better without the circular border. (The existing "rosette" icons don't have such a thing.) Would something more like
Charlie Reid (musician)  Dynamic-blue-right.svg
Charlie Reid (musician)  Gtk-redo-ltr.svg
look better?
Or one could use the same redirect icons as the sidebar,
Charlie Reid (musician)  Symbol redirect arrow grey.svg
Charlie Reid (musician)  Symbol redirect arrow blue.svg
This is perhaps more direct and explicit. Jheald (talk) 17:53, 14 October 2019 (UTC)
Also in red, if that would make redirect links not confirmed as intentional stand out better
Charlie Reid (musician)  Symbol redirect arrow red.svg
But perhaps that needs to be a little more muted. Jheald (talk) 22:06, 14 October 2019 (UTC)
I think you should just make a suggestion according to your personal preferences in the phab topic. If there is opposition at a later point, we can still update the icons. —MisterSynergy (talk) 22:11, 14 October 2019 (UTC)
My current preference is for the last two -- the red for redirect pages that don't have the Template:Wikidata redirect (Q16956589) template, the blue for those that do.
I'll sleep on it, and give the chance for anyone else to comment overnight, and then after that let's go for it. As you say, it's easy enough to update later, once it's clearer how it looks in the wild. Jheald (talk) 22:20, 14 October 2019 (UTC)
Such a bot looks doable at first glance, but putting all parts together would be quite a task. SQL queries are rather simple, the required SPARQL query is no problem either. Sorting things to figure out required edits based on the different query inputs is not overly complicated. What looks a bit nasty is the requirement that these queries have to be run for each and every Wikimedia project individually, which means that one probably does not want to do it daily.
Regarding the actual editing, I am not sure whether pywikibot has useful badge editing capabilities. The aforementioned script does not use any bot framework and makes direct API calls, which is okay for a one-off 500 edits batch, but for a continuous job that is not sufficient… —MisterSynergy (talk) 20:15, 14 October 2019 (UTC)
Looks like badge support in pywikibot went live in June this year (phab:T128202). If we ping @Lokal Profil: he may be able to tell us how complete the support is, and whether he thinks it would be up to the task. Jheald (talk) 20:38, 14 October 2019 (UTC)
Thanks for the hint, indeed there is support now. I tried it yesterday with PAWS, but there seemed to be no badge support yet in spite of some indications in the documentation [1]. Now I just found that PAWS has an outdated version of pywikibot (Pywikibot version: [http] r-pywikibot-core.git (71e36c1, g10980, 2019/04/30, 16:42:46, OUTDATED)). On Toollabs, there is apparently now badge support. I have not actually used it yet, but the badges are visible at least… —MisterSynergy (talk) 21:06, 14 October 2019 (UTC)
If it takes multiple days to have up to date icons I would advocate against red as a color to draw attention. Having blue/grey is enough to give the user the information they need and I see no need to draw a huge amount of attention. We also need to decide about what to do with Wikis that don't have the template. ChristianKl❫ 08:05, 15 October 2019 (UTC)
Just to add - I think the blue/grey plain arrows look great. Hopefully avoids any colourblindness issues that might come from blue/red or blue/green, and it's a nice muted colourscheme that fits with the rest of the UI. Andrew Gray (talk) 21:48, 15 October 2019 (UTC)

Quality of location properties[edit]

Hello. I've been testing stuff at a Wikidata sandbox, and stumbled upon a situation where country (P17) is set to a particular country, but location (P276) is set to a location not located within that country. Do we have a way of notifying the editor for such occurrences, and maybe also a way of searching such issues currently on other items? Note that this is not limited to the mentioned properties, but could also be in other property types such as located in the administrative territorial entity (P131). Cheers. Rehman 12:03, 13 October 2019 (UTC)

You may be looking for a query like this one: Cheers, Bovlb (talk) 22:10, 13 October 2019 (UTC)

SPARQL query map[edit]

Hi all,

I am thinking about embedding the result of a SPARQL map query in the frontpage of a website. It shouldn't have many visits but I am not sure about the limitations. I was wondering if this is it the right way to do it. Thanks in advance! --Gustavo.candela (talk) 08:03, 14 October 2019 (UTC)

If possible, please set up a regularly running script that retrieves and caches the query result on your server. It could run daily or hourly or so, depending on how dynamically the data set changes. Otherwise the request might quickly get blocked by the Query Servers. —MisterSynergy (talk) 09:54, 14 October 2019 (UTC)

Adding provinces to dutch municipalities[edit]

I would like to add provinces to all the dutch municipalities ( in Wikidata, the problem is I don't really know the best way to do it. The municipalities sometimes use but this creates some problems as there are other entities there (non-provinces or other municipalities from the past). This makes the query show double results which is inconvenient, an example: (4 results in the query). What would be the best way to look up the provinces only? Maybe somehow use in this situation?

Here is the query:

SELECT ?municipality ?municipalityLabel ?begindatum ?einddatum ?CBS_gemeentecode ?Amsterdamse_code ?gelegen_in_bestuurlijke_eenheid ?gelegen_in_bestuurlijke_eenheidLabel WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  ?municipality p:P31 ?statement.
  ?statement ps:P31 wd:Q2039348.
  { ?statement pq:P580 ?begindatum. }
  OPTIONAL { ?statement pq:P582 ?einddatum. }
  OPTIONAL { ?municipality wdt:P382 ?CBS_gemeentecode. }
  OPTIONAL { ?municipality wdt:P6434 ?Amsterdamse_code. }
  { ?municipality wdt:P131 ?gelegen_in_bestuurlijke_eenheid. } }

Try it!

PS, Should I delete old and solved questions when creating new ones? Antoni1626 (talk) 13:07, 14 October 2019 (UTC)

P131 is the correct way to go; that's what it's there for. There are two or three expedients to consider; but you'll need to decide on a policy going forwards. Using Scotland as a model, we appear to have moved towards redundant P131s at a couple of different levels of government - Local Authority and Civil Parish. Each P131 is qualified with object has role (P3831). It's thus possible and easy to query to get items by Parish or by Local Authority. (We'll probably implement a preferred rank for Local Authoriies eventually). Example at Shemore, Farmsteading, South East Range (Q17772983). Next, it is also possible to end-date old provinces, if that's an issue - example at Dunoon and Kilmun (Q68815389). Next, so far as queries are concerned, one can constrain the result set accoring to the P31 of a P131, so as long as all contemporary provinces have a common P31 there is no query problem. We're moving beyond the point at which we can rely on simple queries which do not test for values but rather make bad assumption. Next if there are P131s which are not administrative entities (e.g. they're really a location (P276), then they should be moved from P131 to P276, and/or removed. And finally, as already noted, statement rank may be used to provide a wdt: view of whatever the thruthy situation appears to be. Hope some of that helps. --Tagishsimon (talk) 13:38, 14 October 2019 (UTC)
Old questions shouldn't be deleted. They get archived after a while automatically and having them in the history can be useful for people who have similar question in the future. ChristianKl❫ 09:31, 15 October 2019 (UTC)

How would a query look that does the same as the query above and looks for located in the administrative territorial entity (P131) object has role (P3831) province of the Netherlands (Q134390) ? Thanks for the help! Antoni1626 (talk) 08:49, 15 October 2019 (UTC)

@Antoni1626: Something like this (last three lines changed, compared to your original query):
SELECT ?municipality ?municipalityLabel ?begindatum ?einddatum ?CBS_gemeentecode ?Amsterdamse_code ?gelegen_in_bestuurlijke_eenheid ?gelegen_in_bestuurlijke_eenheidLabel WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  ?municipality p:P31 ?statement.
  ?statement ps:P31 wd:Q2039348.
  { ?statement pq:P580 ?begindatum. }
  OPTIONAL { ?statement pq:P582 ?einddatum. }
  OPTIONAL { ?municipality wdt:P382 ?CBS_gemeentecode. }
  OPTIONAL { ?municipality wdt:P6434 ?Amsterdamse_code. }
  ?municipality p:P131 ?gelegen_statement .
  ?gelegen_statement ps:P131 ?gelegen_in_bestuurlijke_eenheid.
  ?gelegen_statement pq:P3831 wd:Q134390
Try it! -- Jheald (talk) 10:34, 15 October 2019 (UTC)
@Antoni1626: A different query to try to do the same thing could look like this:
SELECT ?municipality ?municipalityLabel ?begindatum ?einddatum ?CBS_gemeentecode ?Amsterdamse_code ?gelegen_in_bestuurlijke_eenheid ?gelegen_in_bestuurlijke_eenheidLabel WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  ?municipality p:P31 ?statement.
  ?statement ps:P31 wd:Q2039348.
  { ?statement pq:P580 ?begindatum. }
  OPTIONAL { ?statement pq:P582 ?einddatum. }
  OPTIONAL { ?municipality wdt:P382 ?CBS_gemeentecode. }
  OPTIONAL { ?municipality wdt:P6434 ?Amsterdamse_code. }
  ?municipality wdt:P131 ?gelegen_in_bestuurlijke_eenheid.
  ?gelegen_in_bestuurlijke_eenheid wdt:P31 wd:Q134390
Try it!
This query just checks that the final ?gelegen_in_bestuurlijke_eenheid is a province of the Netherlands (Q134390), rather than checking whether the located in the administrative territorial entity (P131) statement has a object has role (P3831) = province of the Netherlands (Q134390) qualifier.
Normally if the different levels of administrative areas nest well within each other, we recommend only to give the nearest (lowest) level of the hierarchy in the P131, and let higher levels be worked out from it.
The complication happens if there are different types of 'lowest' administrative area that a location may be in; or if the nesting from lowest to highest doesn't work neatly, so one can't tell the query to just go up the chain until it finds an item with instance of (P31) = province of the Netherlands (Q134390). It's really only then, as a last resort, that one should start adding extra P131 values to items, with object has role (P3831) qualifiers to distinguish what they all are. Jheald (talk) 11:03, 15 October 2019 (UTC)
@Antoni1626: happy to see you working on this. I think you need to improve it a bit. A municipality of the Netherlands (Q2039348) should always be located in a province. On items like Nieuwolda (Q2483664) this should have an end time. I think these are the ones to fix:
SELECT ?municipality ?municipalityLabel ?begindatum ?einddatum ?CBS_gemeentecode ?Amsterdamse_code  WHERE {
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  ?municipality p:P31 ?statement.
  ?statement ps:P31 wd:Q2039348.
  { ?statement pq:P582 ?einddatum . }
  OPTIONAL { ?municipality wdt:P382 ?CBS_gemeentecode. }
  OPTIONAL { ?municipality wdt:P6434 ?Amsterdamse_code. }
  MINUS { ?municipality wdt:P131/wdt:P31 wd:Q134390 } 
Try it! Multichill (talk) 17:30, 15 October 2019 (UTC)
@Multichill: Is ?municipality wdt:P131/wdt:P31 wd:Q134390 required (887 results to fix), or is ?municipality wdt:P131+/wdt:P31 wd:Q134390 sufficient (104 results to fix,, i.e. allowing a P131 chain to get there in more than one link ? Jheald (talk) 17:53, 15 October 2019 (UTC)
Yes, it is, municipalicities are always directly under the province. Multichill (talk) 17:55, 15 October 2019 (UTC)

Wikidata weekly summary #386[edit]

type mismatch with License property[edit]

Currently, the names of various copyright licenses are alternate names for copyright determination methods, for example PD-US-expired is alternate for published more than 95 years ago (Q47246828), which is an instance of copyright determination method (Q61005213) only. license (P275) has its possible values restricted to only instances of license (Q79719), copyright status (Q50424085), or as a service (Q25036597). What is the solution for this? Should there in fact be items for licenses separate from determination methods? Or should we add instance of license to published more than 95 years ago (Q47246828); or expand the possible values for license (P275) to include copyright determination method (Q61005213)?  – The preceding unsigned comment was added by Levana Taylor (talk • contribs) at 19:57, 14 October 2019‎ (UTC).

@Levana Taylor: I think you're mixing up license (P275) and copyright status (P6216). From the license property: "license under which this copyrighted work is released". The property should only be used on things that are copyrighted (at some point in time). I see you corrected yourself. Multichill (talk)
OK, yeah, forehead slap time. Thanks Levana Taylor (talk) 18:39, 15 October 2019 (UTC)

Proposals to change data types[edit]

I've refactored the above. Please use the talk pages of the respective properties to discuss this set of proposals, rather than flooding this page with duplicate and empty sections. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:39, 15 October 2019 (UTC)
Okay, I will continue there, I was working according to Help:Data_type#Changing_datatype which reads "To propose such a change, you may start a discussion at the talk page of the property or project chat." - apologies for the annoyance, will discuss on each property page instead then. Iwan.Aucamp (talk) 13:05, 15 October 2019 (UTC)
@Pigsonthewing: I removed the list and replaced it with links to specific properties below. Iwan.Aucamp (talk) 17:02, 15 October 2019 (UTC)

There are some properties which have Data type International Standard Identifier for Libraries (P791) but should in my view be External identifier, please find below links to the proposals to change their data types, please add your opinion.:

Iwan.Aucamp (talk) 17:02, 15 October 2019 (UTC)

@Iwan.Aucamp: some of these didn't get converted due to opposition. I recall some properties that looked like identifiers, but turned out not really to be one. A lot of single value constraint (Q19474404) and distinct values constraint (Q21502410) violation is usually an indication that something is wrong. Some just got missed and can get converted. Would probably be nice to have links to the original discussions. I think these are on these pages:
Multichill (talk) 17:16, 15 October 2019 (UTC)
@Multichill: Thanks, will link to original discussion. I was not aware that External identifier has a uniqueness constraint (it should maybe be added here: External identifier). I will go through though and remove proposals from properties where the constraint cannot be met. Iwan.Aucamp (talk) 17:21, 15 October 2019 (UTC)
@Iwan.Aucamp: Not a hard constraint, more like a soft one. If the uniqueness drops say below 90% that's a good indicator that something is maybe not an identifier. I wouldn't close any of these discussions now. Some people might not agree and should be given the chance to comment. Multichill (talk) 17:34, 15 October 2019 (UTC)

page is not showing on Google.[edit]

Dear sir,

I craeted a page on wikipedia, . Then I Created wikidata profile of this page, Q70884813. Problem is that....this page is not showing on google. What is the problem??

User talk:Virenderthind2019

  1. convenience link en:The Game of Votes (don't presume everyone is on a mobile device).
  2. We have no control of what Google does. - Jmabel (talk) 02:04, 15 October 2019 (UTC)
@Virenderthind2019: Articles in (only) English Wikipedia may only be indexed after they are patroled or at least 90 days old. See w:WP:NOINDEX.--GZWDer (talk) 14:00, 15 October 2019 (UTC)
Jmabel how about if i made an article months ago in ArWiki - have no entire in EnWiki but Wikidata item is complete -but still year of death not showing when you search the item on Google in English? Something is wrong for date of death, especially for new items. --Ruwaym (talk) 16:14, 15 October 2019 (UTC)
Again, we have no control of what Google does. - Jmabel (talk) 22:29, 15 October 2019 (UTC)

bulk load in wikidata[edit]

Dear friens, I have the following situation with a private instance of wikidata, the case is that we want to upload 1,5 millions of dataset, that is moreless 30 millions of triples. We are loading at a rate of 1 data set per second, but that is too slow, because we would be talking about 46 working days to load this dataset. We just created a bot, and modified the average time per edit, however nothing changes, and the processing time holds. I thank you in advance any recommendation.

Luis Ramos  – The preceding unsigned comment was added by (talk • contribs) at 11:23, 15 October 2019‎ (UTC). If your Wikibase instance is internal, you don't need to care the speed of editing (Wikidata itself care it because 1. the database is replicated, 2. changes must be dispatched to Wikibase clients and 3. there's a public query service that should be updated continously). You may try to load the dataset in parallel. Note the speed may still be limited by the performance of the server.--GZWDer (talk) 14:04, 15 October 2019 (UTC)
I don't pretend to understand anything you guys are talking about other than "speed". So, all I have to add is - I've found Wikidata (and other the sister wiki projects) has been going really slow. Take that into consideration. If the site itself is running slow, anything you are doing with it is going to run slow. Quakewoody (talk) 14:47, 15 October 2019 (UTC)
You have come on to wikidata, Quakewoody, a wikibase site which receives iirc ~750,000 edits per day to tell us that your wikibase instance is loading data slowly and to suggest that the cause is the wikibase code. Rather than, for instance, your slow server, or some deficiency in the method you're using to upload data, or a configuration parameter mis-set. You profess not to understand a mildly technical answer, above. I suspect you may have exhausted our ability to help you. I note which seems like a better starting point to access wikibase help, than this board. Good luck. --Tagishsimon (talk) 23:37, 15 October 2019 (UTC)
What I said was - the site has been running very slow. I didn't say it was a slow site. My "slow server" has seen a dramatic slowdown in the last few weeks on this site, not all other sites. Quakewoody (talk) 23:54, 15 October 2019 (UTC)

Dear friends, at first thanks for the recommendations, I do not pretend to create more than technical discussion for improvement of our work. My expertise area is knowledge representation and reasoning, and perhaps my question was very basic in the sense of server edition, performance and that stuff. Currently, I am working in parallel loading, and I also have read about all the effort is being done to improve bulk load in wikidata, but my interest was to know about any other possible experience with the issue I am facing, which could be caused by server, network, my code or other causes. Any other technical recommendation will be welcome.

Luis Ramos

Request for help with an item[edit]

Hi all,

I'm still a newbie to WD so I apologize if it's a stupid question, but is there a way to add two "inception" values to an item? I wanted to add a party (Q277879) to a politician's data page, but an error message came up, saying the party was dissolved in 1949 and the policitian was born later. The party was dissolved in 1949 but was refounded in 1989, and it considers itself the same party. Even page Q277879 shows this error next to the leader's name who was also born after 1949. How can I add the 1989 date without getting error messages?

Thanks in advance. Alensha (talk) 15:27, 15 October 2019 (UTC)

  • If the two incarnations of the party are to be considered a single thing, we need to treat the period it was inactive as a hiatus, not a dissolution and a new inception. - Jmabel (talk) 15:39, 15 October 2019 (UTC)
thanks! how can I add a hiatus? haven't found this property on List of properties/Time. Alensha (talk) 16:50, 15 October 2019 (UTC)
I think a 40 year hiatus is pretty questionable, is there any evidence that they are really the same party, such as having the same people involved, or is it basically a new party that's taken over an old name? Dormancy can be indicated using dormancy (Q55909176) like on United Australia Party (Q15130081). Ghouston (talk) 21:41, 15 October 2019 (UTC)
That party was also at one point claimed to be a continuation of United Australia Party (Q1728956), but Wikipedia unsurprisingly treats it as a new party. Ghouston (talk) 21:45, 15 October 2019 (UTC)
In Wikidata it's generally more useful to split an item into two in cases like this to make the meaning of the individual items more clear. ChristianKl❫ 09:35, 16 October 2019 (UTC)
In this case, there is some overlap of members between the original and refounded parties, and the English and Hungarian wikipedias both treat it in a single article. Enwiki says: "The link between the historical party and the present one is disputed, although prominent members of the original party, like László Varga, took part in its refoundation." Ghouston (talk) 02:29, 17 October 2019 (UTC)

Thank you, I used dormancy, like in the Australian party's item.

It is a good question whether it can be viewed as the continuation of the old party or not, but as long as Wikipedia doesn't have separate pages on them, we have to treat it as one party in WD... Alensha (talk) 13:08, 17 October 2019 (UTC)

military specialty[edit]

How do we handle a military specialty like "radioman" or "medic" or "rifleman" or "sharpshooter", they are not ranks but military specialties. They are not really occupations, since they only occur during military service, usually just a few years. --RAN (talk) 19:10, 15 October 2019 (UTC)

field of work (P101) maybe? I don't think that there's any rule that occupations need to last for any particular length of time. Ghouston (talk) 21:37, 15 October 2019 (UTC)
Why not position held (P39)?

Feedback wanted on Desktop Improvements project[edit]

06:53, 16 October 2019 (UTC)

Visualizations using Wikidata: a wordmap and an etymology radial graph - check them out[edit]

I have published two new visualizations based on Wikidata/Wiktionary. Any feedback would be really welcome.

The first visualization is a wordmap: it shows word translations on a map. Try it out and let me know what you think about it and if you have ideas on how to improve it. Link: Wikidata Wordmap.

The second visualization shows words that are etymologically connected (i.e. share the same ancestor) on a radial graph. While a representation using trees would be ideal, data actually belongs to a directed cyclic graph, which is difficult to visualize. After testing different representations I tested the visualizations on two words:

My idea is to have a tool to search any word, visualize the radial graph of etymologies using Wikidata lexemes, add an interaction to show word definitions and pronunciations. The data in the visualizations was extracted from the English Wiktionary using etytree. The RDF database generated by etytree (SPARQL endpoint) can potentially be used to populate Wikidata lexemes. Please let me know your feedback on the visualizations as well as on a possible integration of etytree data into Wikidata.

Epantaleo (talk) 08:18, 16 October 2019 (UTC)

Calendar issues ? Two date of births that may be the same, what to do ?[edit]

See Robert Hooke (Q46830) (at the time I write those lines). 2 dates of birth, 10 days apart, both mentioned on enwiki as the same date … in old and new style. Is it enough to mark the old style one with « Julian » calendar, even if the other sources does not seem to specify a calendar ? Which one should be marked as preferred ? author  TomT0m / talk page 13:18, 16 October 2019 (UTC)

A spooky calendar issue. Isaac Newton (Q935) is a more prominent example. ;) --Succu (talk) 19:34, 16 October 2019 (UTC) See en or de. --Succu (talk) 20:32, 16 October 2019 (UTC)

Declaring the flavor of something[edit]

How would I, for example, say that vanilla ice cream is vanilla flavored? Flavors do not seem to have entries, but flavorings do, like vanilla --MoonyTheDwarf (talk) 19:01, 16 October 2019 (UTC)

I ran into the same problem when i made items about energy drinks. --Trade (talk) 23:14, 16 October 2019 (UTC)
Possibly has quality (P1552), maybe with a qualifier of applies to part (P518) - flavor (Q4173974)? --Tagishsimon (talk) 02:11, 17 October 2019 (UTC)

Kilmalkedar Church[edit]

Hello! Should Kilmalkedar Church (Q55046509) and Kilmalkedar Church (Q30247213) be merged? There is nothing in the place of the first (according to satellite), so probably wrong coordinates, but there is also a different Wiki Loves Monuments ID. Wikisaurus (talk) 16:12, 17 October 2019 (UTC)

The Irish National Inventory of Architectural Heritage ID (P4088) identifiers seem to point to different buildings about 600m apart, and both have buildings shown on the GIS server it links to. One's a modern building (pictured by HEI with a roof), one a historic ruin (pictured by us without). Seem to be clearly distinct. Andrew Gray (talk) 18:53, 17 October 2019 (UTC)

Unidentified authors, again[edit]

Further to a discussion from last week: @Jura1: said that the practice, when you don't know anything about the author of a certain work except the initials they signed the work with, is to create an item for them, with the initials as label and something like "author of [work]" as description. Ergo, every work has an author item, even if it's a very uninformative one. I can see the advantage of that, since the placeholder author item can be updated or merged if the writer is ever identified. Jura1 didn't say what is done with unsigned works by unidentified authors. The extension of the logic of what I wrote above would be to create an item with the label "Author of [Work]" like the Pearl Poet but usually not nearly as memorable verse! :-) Thoughts? Levana Taylor (talk) 16:27, 17 October 2019 (UTC)

I don't know whose practice is to create authors with initials only... apart from Jura's
in fact, many authors, on the contrary have "unknown value" or anonymous (Q4233718) + stated as (P1932) (including *** sometimes), which allows to have the signature at it is stated on the article, and easily fetch all articles for which author is yet to be found.
there are currently 3583 author (P50) -> anonymous (Q4233718) (see ), not counting translator (P655) or other functions... - I do not know how to query the "unknown value" @VIGNERON: can you help on this, please ?
it does not seem a good idea to create hundreds, potentially tenths of thousands of so-called author items, with only initials... better to have info stored on the article item, then create (or link) the author item if you finally identify them...
the "placeholder" idea may seem practical on a small scale, but when you'll have hundreds of B. authors, how will you manage the possible wrong adding of one of those on another work ? even with homonyms (which are rarely in big numbers), there are already people who confuse the authors... - there is absolutely no info apart from initials and the fact that the person has written this specific text ? no IDs, nothing to help with completing these items... if 2 or more of them are merged, do you think it will be noticed ? --Hsarrazin (talk) 18:14, 17 October 2019 (UTC)
@Hsarrazin: For querying somevalue/novalue, see Wikidata:WikiProject sum of all paintings/Anonymous to fix.--GZWDer (talk) 18:24, 17 October 2019 (UTC)
thanks GZWDer but I don't understand the syntax. Could you modify to query items with author (P50) -> unknown value, please ? - that would be very helpful... --Hsarrazin (talk) 18:47, 17 October 2019 (UTC)
I think there is a misunderstanding. I think you should keep using "author name string" (and not an item for anonymous and the qualifier "stated as" with the pseudonym .. not sure where this came from?). You can, if you want, create items do so. This allows to link them to other publications of the same person identified by the initial. --- Jura 18:26, 17 October 2019 (UTC)
OK, I rephrase because I'm not sure of understanding what you mean... : author name string (P2093) should be used, and not author (P50) with anonymous (Q4233718) + stated as (P1932) ? - author name string (P2093) was not created for this... besides, it was created much later after we began to catalog anonymous works, and it was meant as a temporary dump... not as a permanent value
in fact, anonymous (Q4233718) links to templates used on various projects to state that the author is anonym - that's why it was used in the first place...
it would be possible to change to author name string (P2093), but this would mean changing thousands of items edited previously... for what gain ?
moreover, I do not think it is a good practice to use 2 different ways to state anonymous author/creator between books/articles and works of art... the logic is the same, the practice should be the same... --Hsarrazin (talk) 18:43, 17 October 2019 (UTC)
I think we don't use "pseudonym" and "anonym" in the same way. "B." did write other articles in the same publication. --- Jura 18:48, 17 October 2019 (UTC)
Disagree. The articles signed "B." are not all by the same person. The problem with putting too much stress on an initial like this is leading people into the mistake you just made: thinking it is an individual with an identity beyond that article, which would tempt them to add other articles to that record. Study of the signatures in Once a Week magazine shows that the same author often used different signatures, too. To be specific, "B." was used twice by Robert Buchanan, once by Mrs. Bairn, and twice by unknown persons whose works don't much resemble either of the other two. The initial "A." was used by 5 known people... Levana Taylor (talk) 19:01, 17 October 2019
thanks Levana Taylor - exactly my point ! Moreover, "Once a Week" is not the only publication involved... and I bet many more articles are just signed "B", and for most of those, there are no payment registers to help indentify them Face-wink.svg --Hsarrazin (talk) 19:48, 17 October 2019 (UTC)
@Hsarrazin: here is your query (very broad, you can add condition and filter if you need) @Jura1: how do you know that a text signed B. and an other one also signed by B. are by the same author B.? and yes, "B." is not really anonymous nor pseudonymous but it's a very similar the situation. Cdlt, VIGNERON (talk) 19:09, 17 October 2019 (UTC)
Face-smile.svg Thank you. a lot VIGNERON. Contrary to what I expected, it shows that this option is not used a lot with P50 : only 97 items...
as for "anonymous authors", in libraries, that's exactly how these authors are called by librarians (in French, at least) : "anonymes"... (because absolutely nothing proves that this letter has anything to do with their real name)
my sister happened to be first-hand witness of what could happen when they are poorly handled during retroconversion of national libray with thousands of them : 26 "anonyms", A, B, C, D… - floruit 12th-20th century (sometimes earlier)... a gigantic mess !!!
that's why I think it is not a good idea to create items for these authors with only initials ; store those and create the author item (if needed) when identified is no more difficult and less risky. --Hsarrazin (talk) 19:48, 17 October 2019 (UTC)
Apparently Levana knows precisely who the author is . --- Jura 19:52, 17 October 2019 (UTC)
not exactly : she knows that at least 4 different authors wrote using B. as signature... which is not necessarily "precise" :) --Hsarrazin (talk) 19:55, 17 October 2019 (UTC)
I still don't see why you think they are anonymous .. --- Jura 20:02, 17 October 2019 (UTC)

(unindent) Part of the reason for the confusion in this conversation may be different definitions of the word "anonymous." Some people (Jura?) use it in a very narrow sense, to mean someone who is deliberately concealing their identity. There actually are a couple of instances where the payment in the Once a Week account book is recorded as being made to "Anonymous, c/o [address]" -- there we have someone who doesn't want even the magazine's editor to know who they are! On the other hand, sometimes unsigned articles are unsigned as a matter of editorial policy: for instance, I recently read a history article discussing how a lot of magazines published unsigned book reviews to give the magazine the feeling of having a unified cultural-aesthetic outlook. That isn't anonymous in the narrow sense. Problem: we usually don't know what is the reason for an article being unsigned. So from a practical point of view we can't use the narrow definition of anonymity; we perhaps ought to not use that word. We can divide author identities into face identities (most basically, "author of X article," which may have a identifying designation like "B." -- you can call that designation a pseudonym, without stretching the meaning of that word too far) and personal identities (an individual human being). Sometimes we can connect one or several face identities with a personal identity, and sometimes you have a face identity designation which is used on several articles and you know they are all connected to the same personal identity so you can consider them to be the same face identity, designated by a single pseudonym. I think we can all agree that we ought to create items for personal identities and connect them to the articles they wrote, and make a note of what face identity each article has (either a designation or just "author of [article]"). But the question is, is it useful to create items for face identities not connected to personal identities?

There's two cases to consider when thinking about whether to create an item for a face identity. One is, when there's a distinctive face identity designation that groups together a bunch of writings that use that designation and you are sure they all have a common personal identity behind them, even if you can't say anything about the nature of that person (for example, we don't even know if the person using the pseudonym Louis Sand was a man or a woman or what, but there clearly was only one Louis Sand). It is definitely useful to have an item for this. Philosophically speaking, you could also create an item for the personal identity behind the pseudonym because you know there is such a person, but I think that a null-content item like that can be omitted because its existence can be simply inferred from the existence of a pseudonym. No reason to create an item with no additional information attached to it when its existence can be inferred in one step. And that I think is analogous to the situation of face identities that are so non-distinctive that they can't be connected to either personal identities or other face identities. The existence of a piece of writing implicates the existence of a face identity, either its signature or "author of", and if you know nothing further about that face identity, you can simply leave it to be inferred.

Which is all a very, very long-winded way of saying that I agree with Hsarrazin that creating author items for articles when you know nothing whatsoever about the author except their signature, if there even is one, is burdensome without being useful. Levana Taylor (talk) 22:15, 17 October 2019 (UTC)

Yet more thoughts. I'm sure I'm reinventing the wheel because I wasn't around for previous discussions, but I do want to get this down on paper, sorry. In what I said above, personal identities would be "instance of human," and face identity designations would be "instance of pseudonym." (I will have to go back and change some items I created for pseudonyms where I did put "instance of human," before I clarified my thinking this way.) I'm starting to think that the author property should only have values that are humans. That makes it easy to cover all cases: either you know a personal identity for the author, in which case you put in that item, or you don't, in which case you put in "unknown value." "stated as" could be mandatory (well, highly recommended), and would be either the signature or "no value" if there's no signature. So what if this is one of the articles that has a useful pseudonym but no personal identity? That could be an additional qualifier for "author": "pseudonymous author." and no, it wouldn't be redundant with "stated as." consider the case of D. Richmond. This person, whoever they were, published Book A signed D. Richmond, Book B unsigned, and Book C signed "the author of A and B." Voila, books grouped together by a common face identity, which face identity can only be referred to as "D. Richmond." So the authorship info for Book B is author:unknown value, pseudonymous author:D. Richmond, stated as:no value. Levana Taylor (talk) 22:45, 17 October 2019 (UTC)

I'd say there's no point in creating items that contain no information. If the item just says "author of book X", it contains nothing beyond what you can already find in the item for book X, and it probably violates the notability policy. However, if you can somehow establish that an author, whether completely anonymous or using a pseudonym, is the author of more than one work, then it would be worth creating the item make a link between the works. Ghouston (talk) 23:07, 17 October 2019 (UTC)
Agree with Ghouston. If there's not enough information to reliably match/not-match people to an item, then it shouldn't be created. Per WD:N there needs to be sufficient information, from serious sources, for an item to be justified. Jheald (talk) 23:14, 17 October 2019 (UTC)

somevalue + stated as[edit]

Further up the thread there is discussion about setting values to unknown somevalue with qualifier stated as (P1932) to be able to record a text value for a statement when there is no wish to create an item.

This is a standard mechanism used with a number of properties, which I am fully in favour of, though apparently User:Jura1 sees a problem with it (diff, thread)

However, it does seem that the infoboxes in some languages may need some updating to reflect it. For example, see fr:Albert Lynch now reporting values of "Inconnu" for "Père", "Mère", and "Conjoint" following edits to Albert Lynch (Q2407898). If somebody is familiar with the fr-wiki modules, it would be great if this could be sorted out. Jheald (talk) 23:31, 17 October 2019 (UTC)

anonymous (Q4233718) is constraint violating for author. If two works have the same item as author they are supposed to be authored by the same person. That assumption stops working once you add anonymous (Q4233718) as a value for author. ChristianKl❫ 12:17, 18 October 2019 (UTC)

SourceMD cruft?[edit]

2C (Q57483490) claims to be a scholarly article (Q13442814) but if the DOI is followed what is found looks like it might be a single image (figure 2C?) from an academic paper (gory image warning). There's no traceability in the item history, just anonymous bots. Clearly there's an issue if wikidata is being loaded with items representing themselves as academic papers but turning out to be jpgs. --Tagishsimon (talk) 23:57, 17 October 2019 (UTC)

Mysterious. I searched for the DOI and found which contains a reference "Crinnion, W. J. (2011). 2CSauna as a Valuable Clinical Tool for cardiovascular, autoimmune, toxicant- induced and other chronic health problems. Alternative Medicine Review, 16(3), 215-225. doi:10.15417/1881." Note the title starts with 2C. But the actual paper this represents (here the Google Scholar cluster) does not have the 2C.
My hypothesis is the author of the first PDF copied the ref from somewhere leaving the 2C as artifact (isn't "=2C" a space in some coding?). Second, some bot scraped the ref from the PDF and put it in some database... but the picture desn't fit. Better ideas? --SCIdude (talk) 13:08, 18 October 2019 (UTC)

Spectacled flowerpecker, Dicaeum dayakorum[edit]

en:Spectacled flowerpecker opens with the words :

The spectacled flowerpecker (Dicaeum dayakorum) is a species of bird belonging to the flowerpecker family, Dicaeidae.

cited to [2], published yesterday.

The linked Wikidata item, Spectacled Flowerpecker (Q3551001), is classed as a group of organisms known by one particular common name (Q55983715), with no taxonomic name. A search of Wikidata for "Dicaeum dayakorum" find no results. Should Q55983715 be remodelled, or should a new item be created. and if the latter, should the en, and other, interwiki links be moved? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:19, 18 October 2019 (UTC)

P.S. on further examination, the item was previously classed as an instance of Q16521, so the first part of my question is really, should that statement be restored? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:23, 18 October 2019 (UTC)
I would change it to a normal species item, since that's how it now seems to be described. Ghouston (talk) 10:59, 18 October 2019 (UTC)



Göran Andersson Nordberg (Q71400196) and Jöran Nordberg (Q737301)? --Magnus Manske (talk) 11:45, 18 October 2019 (UTC)

  • Yes but calls him Göran for some reason. --Wolverène (talk) 11:56, 18 October 2019 (UTC)
    • enwiki: "Göran or Jöran (both pronounced [²jœːran]) is the Swedish form of George, not to be confused with the Slavic Goran." --SCIdude (talk) 13:13, 18 October 2019 (UTC)

Interface suggestion[edit]

It'd be nice to have an "add statement" link at the top of pages as well as the bottom. Levana Taylor (talk) 13:12, 18 October 2019 (UTC)

Just use "a" on your keyboard. --SCIdude (talk) 13:14, 18 October 2019 (UTC)
Sorry, what do you mean by that? Levana Taylor (talk) 13:36, 18 October 2019 (UTC)
See KeyShortcuts in your preferences. I'm sorry, I'm using these so frequently that I thought they were on by default. --SCIdude (talk) 14:12, 18 October 2019 (UTC)

Unusually long lists in some items[edit]

@Jarekt: just pointed out some cases where commons:Template:Wikidata Infobox is timing out before rendering the infobox. I think these need cleanup/remodelling here, but I'm not entirely sure how. Is anyone interested in having a look?

Thanks. Mike Peel (talk) 14:11, 18 October 2019 (UTC)

I am removing all or most location (P276) values from ice cream cone (Q1156634). They make no sense to me. --Jarekt (talk) 14:26, 18 October 2019 (UTC)
I counted number of rail stations listed as owner of (P1830) of SNCF (Q13646). We have 3556 of them. That is crazy. I think we need to remove them. --Jarekt (talk) 14:41, 18 October 2019 (UTC)