This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
Request for advice/user input for a new Open Knowledge Network Project
Hello,
This is a request to get guidance from expert curators in the Wiki Data Community in support of a new project to create an Open Knowledge Network for Intelligent Textbooks. The project is funded by the National Science Foundation in USA. WE are looking to talk to a few curators of WikiData to understand things like (a) their motivation to work as a curator (b) success and challenges (c) reward structure for themselves and for others (d) general thoughts and strategies on starting a new open source project for curating structured knowledge for textbooks. We hope to do this in a phone conversation which we hope can be done within one hour, but in no longer than two hours.
• I've been doing some English to Dutch translating. The only item left on Wikidata:Wikipedia is "super language family" (in the tekst just before header: "1.3 What does this mean?"). However, there's no such thing in either wikipedia or wikidata. I edited it to "language family" on the main English language page without any problem but it doesn't show this when I click "Nederlands" on "Other languages".
• "super language family" is the only item left that needs to be translated when I click that button, but there's actually quite a lot of text still in English on that page! What am I overlooking? Dutchy45 (talk) 11:52, 1 December 2019 (UTC)
Use of <no value> to convey current status of an item statement
From a perspective of truthyness it feels unsatisfactory to leave the Heritage designation set to 'Category A listed building' with normal rank, albeit with an end time, with no other indicator of current status. It's a truthy fails because it's no longer true (!) and because wdt: queries shouldn't have to check P582 to ascertain (contemporary) truthiness. Deprecation of the Cat A status is wrong, since the value constrained by start and end time was true. <no value> with a start date is true. However it might be considered radical or wrong. Discuss.
Seems right to me, as long as the designation was formally rescinded. I know of things in Seattle that still have "landmark" status even though the actual structure no longer exists. - Jmabel (talk) 17:58, 1 December 2019 (UTC)
I want a page where like Wikipedia’s vandalism report page(and username report page), you report vandalism users (and also users that break the username rules). I also want a warning template for creating articles about them selfs which could help because Users want to make pages about them selfs which is not allowed, also can you make a page about the user warnings
(And how they look simler to Wikipedia). LoganTheWatermelon (talk) 19:05, 1 December 2019 (UTC)
Wikidata has no username rules that could be violated. We don't have anyone creating articles about people on Wikidata. We have items but those are qualitatively different and it's not forbidden for users to create items about themselves. Currently, it seems to me like reporting vandalizing users to the Admin noticebord to be banned works fine. ChristianKl ❪✉❫ 22:43, 1 December 2019 (UTC)
I think the ideal solution is either case would be Wikimedia-wide bans. To the extend that it makes sense to have a more formal system, that system shouldn't be located at Wikidata. ChristianKl ❪✉❫ 10:35, 2 December 2019 (UTC)
I have no problem with ChristianK1 and ChristianKlMustDie. UNLESS the sole purpose is to harass ChristianKl. I mean, look at all of the names that have 2, 69, and 666 at the end of them. They aren't trolls, they are editors with the same name using numbers to distinguish each other. Quakewoody (talk)
The closest we have to a specific policy is currently the help page Wikidata:Autobiography which says "In general on Wikidata it is allowed to edit your own data item. However, you should not create an item about yourself unless you are sure that it fulfills the notability criteria." ChristianKl ❪✉❫ 13:51, 2 December 2019 (UTC)
Improvements to Special:EntitiesWithoutLabel and Special:EntitiesWithoutDescription
Hello all,
While working on wb_terms table, we made some changes on the way Special:EntitiesWithoutLabel and Special:EntitiesWithoutDescription display their results. Currently, they are displaying a list of unordered items with labels. In the near future, these two pages be replaced by the search page using ElasticSearch: with the keyword -haslabel:en or -hasdescription:en, the list will return more information (QID, label, description, number of sitelinks and statements if relevant, date of last edit). The results will be sorted by relevance, using the current rules set up for Elastic search.
This change will be deployed on Thursday, December 5th. If you want to see how it will look like before that, it's already active on test.wikidata.org.
Following the first changes we made on summaries of edits that are made from the wbeditentity API (for example editing terms from the new mobile termbox), we're about to roll out the next step of this change.
For edits where terms in less than 50 languages are changed, the format of the edit summary will be Changed labels, descriptions and/or aliases in: fr, es, en, pl, zh, it, hu (so called shortened version)
For edits where terms in more than 50 languages are changed, the format of the edit summary will be Changed label, description and/or alias in N languages (so called fallback version, that we have at the moment)
This change will be deployed on Thursday, December 5th. You can see more details in this ticket. If you have any questions or issues, feel free to ping me.
Also pinging @Pintoch: because it has an impact on EditGroups.
Alt Labels displays item labels in other languages when there is no label in your default language. You may click a label to add it as the label for your language. This tool is very helpful for labels for people, books, articles, or films that do not need their labels to be translated.
Not necessarily. Copyright protection for short phrases is tricky. It is not simply the length but the level of unique creativity involved. A slogan like "Drink milk every day" may not cross the threshold of originality, but witty bumper sticker slogans like "I may not be totally perfect, but parts of me are excellent" are copyrightable.[1] And the European Union has a stricter threshold of originality than the U.S.,[2] so that milk slogan may be copyrightable in some jurisdictions. Lastly, it is debatable whether merely hosting copyrighted text on Wikidata satisfies the fair use doctrine, which generally invokes a "transformative" purpose, e.g. to comment upon, criticize, or parody a copyrighted work.[3]-Animalparty (talk) 18:47, 3 December 2019 (UTC)
The relevant policy is that material added to Wikidata should be CC0 or public domain, so first and last lines should generally only be added for public domain works. There's also a question raised at that RFC about whether long titles are an issue too. Ghouston (talk) 21:57, 3 December 2019 (UTC)
Thanks, all. This discussion is really helpful. I will delete the 1st and last lines since in this case they are neither CCO nor public domain. Thanks again! -- M2545 (talk) 22:07, 3 December 2019 (UTC)
On a related note, is there specific language regarding the use of freely-licensed but non PD images? There are surely hundreds of thousands of images in image (P18) that are CC-BY (and are we then violating the license by reusing without directly attributing the author?). -Animalparty (talk)
The image properties are just links, the images themselves are on Commons. The attribution is no different to how Wikipedia does it: click through. Ghouston (talk) 23:05, 3 December 2019 (UTC)
Submitting an Organization's Data Wikidata
Hello,
I would like to add a subsidiary of our organization to Wikidata. What are the next steps to getting this implemented? Also, our parent company already has a Wikidata page which needs to be updated. As an employee of the company can I make edits to the page?
@Fleetcor: Welcome to Wikidata! You can edit items about a place that you work, but if you are paid to edit or otherwise receive benefits from editing here you should list that relationship on your user page. To add a subsidiary use the "Create a new item" link (if you're sure it's not already listed) and then link the two items using the has subsidiary (P355) and parent organization (P749) properties. ArthurPSmith (talk) 21:28, 3 December 2019 (UTC)
How to put the fact the Queen sent her first email in 1976 in Wikidata?
Hi all
I found out Queen Elizabeth II sent her first email in 1976, which is pretty amazing. I can't work out how to put this in Wikidata. I assume she is the first head of state to send an email.
There's such a constraint on the talk page but the logic seems to be too complex for the current automatic constraints. ChristianKl ❪✉❫ 13:42, 4 December 2019 (UTC)
Thanks, inexperienced in WD, I was unsure whether to delete or look for a proper link. I’m taking two guesses. First, either of those courses would be acceptable. Second, with so many more items and fewer watchers than in Commons, vandals will get away with more. Jim.henderson (talk) 02:09, 5 December 2019 (UTC)
Self-referencing P17 for 373 objects - ok to remove?
For 373 objects there are self referencing claims for country (P17), i.e. the object's country (P17) is the object itself.
I am sorry but I don't understand if anyone is against my request to remove self referencing P17 claims. @SCIdude: agrees, but "extends" my request to something more and the comments from @Tagishsimon: and @Bouzinac: are, to my understanding, about that extension, not about the original request. Please forgive me if I misunderstood you.
@Jura1: Not sure who are you asking here, but if you are asking me, I am trying, in Lua code, to find out from a Wikidata object the value for place of birth (P19) and to that I want to add the country (P17) for the place of birth (P19), like this:
Of course I can check for the self referencing, Qnnn/P17 → Qnnn, in my Lua code, but as I think that self referencing property country (P17) is an error, I could skip that test if these self references were removed.
This coding issue is not a very big problem, but I raised the question here in the project chat because I wanted to know if self-referencing is a break of some basic rule or not. Maybe there is some better place for that question.
Generally, it's not useful, e.g. author (P50) can't refer to itself. However, it's not a problem as such. The country associated with a country is the country itself and the statements are actually quite helpful.
For your sample usecase: If you'd want to check in LUA for completeness as well (is it a country or is the country just missing), you need to check if the location value is a country if you would remove P17 from countries. If you query in SPARQL for the country of birth, it would get complicated and many location based statements would be harder to check for completeness. So, if you want to save a few bytes on country items, maybe there are other things more worth looking into. --- Jura05:26, 5 December 2019 (UTC)
Wikidata attracts attention and users from many institutions from the cultural sector – and the Wikidata Team wants to support their needs. However, while we did exchange and learn a lot from cultural institutions in the past, we did not have a written account based on research on cultural institutions’ needs which helps to spread knowledge and understanding for these users. This is why the UX Team of Wikimedia Germany did such research. We now can share our Research Report on “Wikidata Use in Cultural Institutions“.
From June-September 2019, our UX researcher Jan talked to 16 users who worked at different cultural (“GLAM”) institutions to find out about “How and why do people in cultural institutions use Wikidata?” and thus learn more about participants’ motivations, activities and problems.
In the report, we describe…
The advantages participants saw in sharing data
Problems of modeling
How people come to Wikidata via Wikimedia Commons
…and much more
The things we found out probably not new to some of you. Rather than uncovering surprising results, the report aims to describe workflows and motivations and to serve as a point of reference. This can be particularly useful for people who had no previous exposure Wikidata and/or GLAM institutions’ work and its context.
The research thus focused on describing motivations, activities and problems. While the data gathering was often like a conversation, the research methods were based on qualitative research methods and design ethnography – thus they supplement our other ways of learning from the community, like direct personal exchange or interacting on talk pages.
We hope that the report is useful and interesting for you. Many thanks to all the people who helped in creating the report, particularly the participants from GLAM institutions who showed and explained to us how they use Wikidata.
datasets about the most published subjects on a some scientific media
Hi, I'm Marcio, a data science student pursuing a computer science MSc degree in a Brazilian university.
Currently, my colleagues and I are finishing a work on public data from scientific databases like arXiv, IEEE Xplore and Springer Nature. Gathered data is about the most published subjects from 1970 up until these days.
Now, we are considering where to make our data available to the world. We already know Wikidata enforces public domain, and we are OK with that.
What we have to donate are 3 XLSX dataset files plus descriptions of each.
Wikidata doesn't host a series of datasets but it hosts a list of items. Items often appear in multiple different datasets. Without knowing more about your data it's hard to say whether or not it's possible to integrate it into Wikidata.
@ChristianKl: hello, thanks for replying back. Our rough data is available here. We can export it to a more suitable format, if necessary. I don't think Commons is appropriate for datasets. Anyway, let me know if you have any doubts. M. B., Jr. (talk) 23:37, 5 December 2019 (UTC)
Added "owner of", I don't know what may fit better than this. It's also unclear if we need this statement at all: Q64726 (Aenne Burda) hasn't been mentioned as the owner of Q2640282 (Burda Style), same with Hugh Hefner and his Playboy Magazine. Blogs are in fact semi-to-non-professional online magazines... --Wolverène (talk) 08:22, 6 December 2019 (UTC)
I've asked about this twice. The reason is that SMILES (both isomeric and canonical) properties are not external-id type byt regular string type, so the WD software does not urlencode it, the gadget do. See [3]. Wostr (talk) 14:31, 6 December 2019 (UTC)
Find Q items of WP pages that are not using a template
best practices for former institutions (mergers, name changes, etc.)
Is there an established protocol or guidelines for handling former institutions? Universities, hospitals, companies, museums, and other institutions often have different names over the course of their history: sometimes as a result of a simple name change of an existing organization, other times as the result of mergers or more complex changes, becoming a more distinct entity. I can see the logic in creating distinct items in some cases, and merely listing former names as aliases in others. Case in point, I created United States International University (Q77518402) which was formally merged with other colleges to Alliant International University (Q1066989): I figure it makes little sense to claim someone was an employee or alumnus of an institution that did not exist when they were affiliated with it. On the other hand, San Francisco State University (Q1256981) was formerly known as "San Francisco State College" and "San Francisco State Normal School", among other names, but was still largely the same institution (although it appears the physical location changed over the decades). Are there recommendations on how to handle cases like this? (ideally somewhere more accessible than the bowels of chat archives - overall, Wikidata guidelines are either quite arcane or non-existent). Should the subject named as (P1810) qualifier be used when a former name of interest is merely an alias of an item with a different name? Note: this is a separate issue from the perennial problem of separating institutions from the building that houses them. Cheers, -Animalparty (talk) 03:40, 7 December 2019 (UTC)
I don't know if guidelines are even possible, because any change can be represented either with one item with date qualifiers to describe the change or with two items. Sometimes the choice is inherited from Wikipedia, when it chooses to have a single article or separate articles. I think a change of name is for some reason considered more significant than other changes, and more likely to lead to separate items (and external sources like VIAF will often make a new item after a name change). The ambiguity particularly affects organisations and countries, while we are happy to represent a person with a single item regardless of name changes. If you wrote a guideline, and tried to apply it uniformly, you'd end up with all kinds of problems. Why do we have a single item for the USA, since its inception, ignoring the loss of the Philippenes, for example, while we represent the UK a new country after the loss of part of Ireland? Probably because the latter changed its name and the former didn't. Ghouston (talk) 06:16, 7 December 2019 (UTC)
Loss of the Philippines was more analogous to the loss of each British colony. Loss of most of Ireland was a change of what union of countries was the United Kingdom. But that is more or less beside the point.
In some cases, continuity is so clear this is best handled by attaching start and end dates to official name (Q11938905). It's only when there is more than that to express that there would be a reason to split items, or when we are forced to by the existence of multiple Wikipedia articles. - Jmabel (talk) 17:40, 7 December 2019 (UTC)
We have "member of military unit" and "member of sports team" and generic "member of" defined as "member of a club or organization". Should we have "member of military group"? Perhaps "group" isn't the optimal name. It would hold the smallest entry known for that person that we have. It might be a regiment or a division or some other subdivision of the military. --RAN (talk) 22:50, 6 December 2019 (UTC)
After I wrote my comment, someone edited 'We have "member of political party" and "member of sports team"…' to 'We have "member of military unit" and "member of sports team"…', which makes my comment look stupid instead of spot-on. This is rather inappropriate. - Jmabel (talk) 00:47, 9 December 2019 (UTC)
Yes. "member of military unit" works. Better than the generic member of (P463) we seem to fall back to at the moment.
Unfortunately I don't think formatter URL changes will automatically update - I think we've had problems with this before. They'll update when the item is next edited or purged, but otherwise will keep the old ones. Andrew Gray (talk) 14:52, 7 December 2019 (UTC)
It looks like @Jura1: has taken charge of the immediate situation at hand without answering my question. Am I to assume that is the correct way to handle such changes? It should be emphasized that many infoboxes and external link templates on various Wikipedias source directly from Wikidata statements, so a single borken or improperly formatted entry here may immediately affect thousands of other sites. -Animalparty (talk) 19:24, 7 December 2019 (UTC)
@Animalparty: It's complicated, but I think that Wikipedias will be okay. It's just the rendered version on the Wikidata pages that doesn't properly update. Unless the Wikipedia templates hardcode the URL and attach the Wikidata value to it, in which case they'd not be affected by the formatter URL here and need to be manually updated whatever happens. Andrew Gray (talk) 20:54, 8 December 2019 (UTC)
Commons-to-Wikidata?
With all of the great work done to make WD properties to the Commons Artwork template, has anyone got a tool or script to take a Commons file with a well-populated Artwork template and create a WD item from it? - PKM (talk) 20:59, 7 December 2019 (UTC)
User:Pasleim's HarvestTemplates is one tool one could use for templates on regular Wikipedia pages. For Commons files I imagine it would need to be adapted to pick up what is the Wikidata item for the artwork, since there would not be a sitelink to identify it. Jheald (talk) 21:24, 7 December 2019 (UTC)
Not so relevant, because you're talking about creating new items, rather than adding to existing ones.
A significant issue would be matching strings to items, though. Not sure to what extent that could be automated. Jheald (talk) 21:30, 7 December 2019 (UTC)
PKM: If a file on Commons has a well populated Artwork template, template, my modus operandi is to manually create a WD item, give it a single statement of instance of (P31) -> painting (Q3305213), then add the Q-id to the Wikidata field of the Artwork template. The rest (or at least most) of the template data should then auto-populate via Quickstatements (you need to click the Arrow that appears at the top of the Artwork template after adding the Q-ID). This is much more effective when the Artwork template includes templates rather than bare text (e.g. templates for Creator, Institution , Size, Technique, etc.). See Edit history of Dr. Silvester Gardiner (Q77479026) for example. Once the QID is added to the template, each subsequent change can be synced to Wikidata in the same fashion (sometimes it requires a cache purge / dummy edit to clear). -Animalparty (talk) 23:07, 7 December 2019 (UTC)
@PKM, Animalparty: I've looked into importing data from Commons to Wikidata and usually the data isn't very good and actually hinders me in my imports. I rather get the data from the source and link up the existing Commons image and add any remaining data. Let's take Dr. Silvester Gardiner (Q77479026) as an example. This painting is in the Seattle Art Museum (Q1816301). Previously they had a really crap collection website, but looks like they got a new one. I created Wikidata:WikiProject sum of all paintings/Collection/Seattle Art Museum to see what we have now. Fortunately only a couple of paintings and all seem to have a valid inventory number so I updated the notes. Often I run into collections where I spend more time cleaning up the existing paintings than doing the actual indexing. The actual indexing isn't a lot of effort. Order of magnitude of 2 hours of work depending on the quality of the website. After the indexing bot is done, several other bots might kick in (run every night):
Bot that tries to find the creator (P170) based on the description "painting by <some painter". see for example this edit
Add all sorts of other missing data some of which might be imported from Commons. The property statistics give a good overview of what to add
I try to optimize as much as possible on reducing the human effort. Bots can do all the stupid work, humans should be able to focus on the more challenging tasks.
@Multichill: thanks for those insights. I typically work with decorative art objects rather than paintings (at the momement I'm doing tapestries), so I end up adding artists (both designers and weavers/workshops) and materials all the time. The particular item which drove my question was a Met object that had not been imported to Wikidata, so the metadata was well-structured, and as always my plan is to add references from both the museum site and relevant books after the import. So much to do in this space. - PKM (talk) 20:54, 8 December 2019 (UTC)
Rename
Could someone rename "18 And 19, Abercromby Square (Q26638916)" to "19, Abercromby Square (Q26638916)" to match the correct name on Historic England and Commons? Normally you'd just stick a template on the page but there seems to be no way of doing that. I don't know why everything seems so difficult here. Rodhullandemu (talk) 20:45, 8 December 2019 (UTC)
Given the articles that are sitelinked, it appears that there may be a distinction specific to German and Lithuanian. - Jmabel (talk) 02:00, 3 December 2019 (UTC)
It seems that the last time we discussed it nobody cleaned it up. Given that I'm a native German speaker I tried to clean it up now. I'm not 100% certain that the Lithuanian is right but I think it should also now be at the correct place. ChristianKl ❪✉❫ 13:24, 3 December 2019 (UTC)
It seems that police officer (Q384593) is a mix of concepts. From my understanding of police officer (Q361593) it is every sworn police officer being a public servant or having a status of a public servant (so e.g. it is every police officer in Poland). But the Lithuanian article from police officer (Q361593) says it is not only police officers but also police civil employees. Probably there should be more items than these two. Wostr (talk) 16:16, 3 December 2019 (UTC)
To be a "Beamter" you have to swear to serve the state. Afterwards you can't be fired, but you also can't strike. "Beamte" are subject to a package of additional rules. I'm not sure to what extend Poland has such a status. ChristianKl ❪✉❫ 08:26, 4 December 2019 (UTC)
Very similar, but nevertheless I think it should be different item for the German (+Austrian?) term and different items for similar terms in different jurisdiction. Wostr (talk) 13:43, 4 December 2019 (UTC)
There should be one "police officer" item per jurisdiction where a law/legislation authorises specific rights and obligations to the jurisdiction specific "police officer" item. Why? Because each "police officer" item specific to a jurisdiction will have different attributes, particularly in the duties authorised to be caried out by the "police officer". The generic "police officer" profession type would still exist too, and this would be what the jurisdiction specific "police officer" items are subclasses of. Dhx1 (talk) 22:00, 3 December 2019 (UTC)
That might be the case but it's separate from the distinction between "de:Polizist" and "de:Polizeibeamter". Austria for example has a a similar legal system as Germany and also has the distinciton. ChristianKl ❪✉❫ 08:26, 4 December 2019 (UTC)
Changes to Wikidata Query Service UI’s example loading
Hello all,
We recently made some changes to the way the Wikidata Query Service UI loads the example queries. This change can impact people who are maintaining these queries, as well as people running their own Wikibase instance including the Query Service.
Before the change, our approach was to get the HTML of that wiki page from Parsoid (this includes some template metadata which the normal parser output doesn’t include), and from that extract the query parameter of all {{SPARQL}} and {{SPARQL2}} transclusions.
With our improved approach, we get the HTML of that wiki page from the parser, and from that extract the contents of all syntax highlighted blocks.
The improvements resulting from this change are the following:
The queries no longer have to be specified directly on the page using {{SPARQL}} or {{SPARQL2}}; they can be transcluded indirectly, e. g. using {{query page}}. You can see a comparison at User:TweetsFactsAndQueries/Queries-test-transclude and User:TweetsFactsAndQueries/Queries-test-copy. If we go with the solution of one query per page, we should be aware that we can fit less queries on the examples page before we hit some parser limits. Maybe now is a good time to sort and tidy the list of examples? :)
Queries that contained an unescaped pipe character (|) were previously cut off at that character in the query service UI, this should now be fixed and all queries should be displayed just like on the wikipage.
If the examples page hits some limit of the parser, then some examples will not be loaded, whereas with the previous approach they would still be loaded and shown on the query service UI even though they weren’t working correctly on the wiki page.
More use of transclusion should allow greater flexibility and variety of ways the examples can be presented and reused. Which is good. But will it have an impact on tools which index the examples, that help people find queries demonstrating the syntax they're looking for? (Also cc: @TweetsFactsAndQueries:). Jheald (talk) 16:12, 4 December 2019 (UTC)
Can we have a corporate version of the Ancestors tool that will show all the mergers and acquisitions that went into a current corporate entity? --RAN (talk) 22:25, 8 December 2019 (UTC)
I’d love that. My impression is that most of our corporate history doesn’t exist... this might help fix that. - PKM (talk) 19:21, 9 December 2019 (UTC)
Please do not assume what I'm asking. What I am asking is quite clear, and it is "Is there an item that means 'work in several volumes'," nothing more, nothing less. I am perfectly aware of the impractically complicated separation of work and editions, and it has nothing to do with what I am actually asking. Circeus (talk) 04:37, 9 December 2019 (UTC)
@Circeus: Please accept my apologies for misconstruing your comment. My direct answer is not that I know of. I agree with Jmabel that we may not need a dedicated item for such: is "multi-volume work" a distinct, notable class of works? Is it necessary to clarify relationships among items? Although I'm not I experienced with querying and data management theory, think it would needlessly complicate subclass/instance of hierarchies. An encyclopedia (Q5292) or book (Q571) may have 1 volumes or several: I think it's better to just state what it is, with number of volumes represented by number of parts of this work (P2635). -Animalparty (talk) 06:08, 9 December 2019 (UTC)
Just to note that what Circeus was asking was not clear, and them getting pissy with someone trying to answer the poorly posed question is deserving of derision & contempt. --Tagishsimon (talk) 11:18, 9 December 2019 (UTC)
@Circeus: One question is whether one actually considers the work to be a single book (Q571), with individual volumes perhaps first published in successive years, but still in essence parts of a single book that would be incomplete without them; or whether alternatively the work is more naturally part of a book series (Q277759), with the individual "volumes" able to stand as books in their own right (eg His Dark Materials (Q129240)).
Hello!
For a few times now it happened to me that I have uniwittingly added a P21 property to a random item I've edited or even only viewed.
So far I know of [4],
[5] and
[6].
Every time I have noticed it in my contribution list afterwards and I reverted it, but it troubles me, since it's strange and silly. An only "upgrade" I use to edit Wikidata is The brown tool. Could this happen because of some misclick? Thanks! --GeXeS (talk) 07:14, 9 December 2019 (UTC)
I used to use the Brown Tool and I made these kinds of accidental edits constantly. I switched to Wikidata Usefuls and the problem stopped. Gamaliel (talk) 14:38, 9 December 2019 (UTC)
@Gamaliel: OK, tried that one, thanks for the suggestion... But it doesn't seem to do anything (might be my Vivaldi browser, or also not). Let's say I need to edit Czech description of 200 items of some beetle genus, adding a label "druh hmyzu" (species of insect) to each of them. With The Brown Tool, I'd just go to every item and click the prepared clickbox on the Tool, which would automatically update the item. Wikidata Usefuls doesn't seem to work this way. Is there any documentation on it? --GeXeS (talk) 16:53, 9 December 2019 (UTC)
overpass: Embeds a map displaying features tagged with the current item in OpenStreetMap. Powered by overpass turbo. It helps check if the Wikidata object is mapped in OSM and also if it's correct.
Other Noteworthy Stuff
A WikiCite event will take place in May 2020 in Cologne. Call for contributions is now open (more info)
Can somebody finally settle for me what the intended difference in usage is between these two properties? I've never got it straight, and every time I look at it, I once again get confused.
Is it:
that sourcing circumstances (P1480) is for shortcomings/approximations/other qualifications about things not quite right about the value that are stated in the source ("sourced qualification"), whereas nature of statement (P5102) is for things about the statement independent of the source
or is it about different kinds of things that may be need further explanation (or caution) about the stated value?
Case in point: Suppose I have an identifier extracted from a source being applied by that source to two different incompatible entities (eg separated by a distance of 30km from each other). I'd like to flag that on the statement with a qualifier to say "possible misidentification" (cf misassociation (Q21097088), but in this case I don't know which one has been misassociated). Which property, P1480 or P5102, is more appropriate, and why? Jheald (talk) 22:41, 2 December 2019 (UTC)
Esch of these properties has a list of “one-of” value constraints which (unfortunately to my mind) overlap in a few cases (presumably, possibly, attribution). I tend to use “nature of statement” if the value I want is valid for that property, and otherwise use “sourcing circumstances” (usually for “circa” on dates). This is not terribly scientific. - PKM (talk) 01:14, 3 December 2019 (UTC)
Ping didn't work but I saw this while browsing PC for fun. I recommend sourcing circumstances (P1480) because it would appear in this case that you know the source is wrong, not that the sources agree that the truth value of the statement is uncertain. Deryck Chan (talk) 10:43, 3 December 2019 (UTC)
@Ghouston: I think that indeed it shouldn't. Per what people have said above, the distiction seems to be:
Use sourcing circumstances (P1480) if in principle an exact unqualified value for this could exist (even if these sources don't give it). -- eg there was definitely some date on which the person was born, even if the cited sources can't give it exactly; there was definitely some painter who painted this painting, even if the cited sources can't be sure that it was this person.
Use nature of statement (P5102) if this is not the case, ie if even with correct and perfect knowledge, there would still need to be some clarifying qualification made to the statement.
But then we run into a potential ambiguity: did the source say that something was possibly (Q30230067) correct? Or are we saying that the source was possibly (Q30230067) correct, or alternatively incorrect, in what it said? I don't think we're to the end of this road yet. Jheald (talk) 18:05, 6 December 2019 (UTC)
I do struggle with this. If the source says an item was made by “workshop X” or “workshop Y”, does that mean that the source says it's “possibly” by X? - PKM (talk) 20:16, 10 December 2019 (UTC)
@Richard Arthur Norton (1958- ): - I’ve had good results sending corrections to VIAF using the “send us a comment” link on the bottom of their page. I included the VIAF and wikidata IDs and an external reference for the particular problem, and they fixed it in about a week. - PKM (talk) 20:33, 10 December 2019 (UTC)
Excellent! It looks like they are making good use of our data. Now all we need is for Findagrave to recognize duplicates and merge them. --RAN (talk) 23:45, 10 December 2019 (UTC)
Reinheitsgebot
@Magnus_Manske:Please look at Ixusko Ordeñana (Q64492900) and see how Reinheitsgebot made an error for the death date. I have left the error in. Can someone examine the bot to see why it is making such obvious errors. I have fixed a dozen or more over the past month. In this example it says that it added a death date for "Ralph Leroy Belknap" but when I click on the link it is for "Martin, Leslie, 1908-2000". The death date it inserted was "1960" which does not appear in the entry for "Martin, Leslie, 1908-2000" and Ixusko Ordeñana (Q64492900) is alive. The Wikidata entry is for Ixusko Ordeñana (Q64492900). There may be many more errors that do not trigger the error tracking queries Wikidata:Database reports/items with P569 greater than P570 or Wikidata:Database reports/unmarked supercentenarians. Also please let me know of any other error tracking queries you can think of, that we should be making for humans. Thanks! --RAN (talk) 15:55, 6 December 2019 (UTC)
I fixed the dozen or so errors, but someone still needs to see if these errors are systemic and not caught by one of the error red flag searches. --RAN (talk) 23:47, 10 December 2019 (UTC)
Delete all un-referenced statements ?
Just wondering what people think. While ultimately all statements should be referenced and most statements can actually be referenced fairly easily, many statements aren't even though we imported them from a generally reliable resource and some checking don't tell us these are implausible.
The question is now what to do with these statements. Most data users can easily filter such statements and Wikidata users can complete them as they come across them.
Seems like a daunting task that will destroy wikidata. Sure, it can be rebuilt, but the current project would be devastated. There is way more unreferenced data than referenced data. Maybe instead of working backwards, it would be better to look forwards by making future edits contain a reference before it can be posted? Quakewoody (talk)
Some statements defines the item. (P31 for example) Their nature is axiomatic. Delete those and the whole item has to be deleted. And all the items relying on them has to be deleted. According to Gödel's incompleteness theorems (Q200787) Wikidata can never be fully referenced.
So, no! Random deletion only because of lack of reference is a bad idea. So also always demanding it in the future. But we could be more open to removal and rephrasing of badly referenced statements. I for example often remove statements sourced by Wikipedia, where Wikipedia obviously is wrong. 62 etc (talk) 14:10, 7 December 2019 (UTC)
Sensitized I now see external ID statements from mix'n'match without ref everywhere in medbiochem. Please fix Mix'n'Match ASAP! --SCIdude (talk) 14:34, 7 December 2019 (UTC)
If so, Sapphorain should be blocked. The solution to no reference is to find a reference. We don't incrementally build wikidata by stupid pointy behaviour. --Tagishsimon (talk) 14:38, 7 December 2019 (UTC)
+1. I think at this stage we need User:Sapphorain to confirm that he understands the will of the community on this, and will abide by it; otherwise a community topic-ban would be in order on all statement removals by this user. @Sapphorain: Do you understand, and will you comply? No more statement removals for being unreferenced. Jheald (talk) 19:13, 7 December 2019 (UTC)
I agree that some statements don’t necessarily need to be sourced in order to stay in Wikidata. If a statement is that the university of Zurich has approximately 25’000 students, it can stay without any date or reference, as it is plausible and the approximate figure is easily verifiable. But if the statement is that the university of Zurich has exactly 25732 students, with no date and no reference, then it is a useless and absurd statement that either must be sourced or be deleted: otherwise, what shall we you do if some contributor enters a second unsourced and undated value of say 25998, and yet another one an unsourced and undated value of 26002?! Let them stay on the ground that « a solution to no reference is to find a reference »? What if there does not exist a reference? Let the data stay forever? This is stupid. I will continue reverting new statements without any reference: either the contributor knows a reference, so why did he or she not provide it? Or he or she has no reference, and the assertion is dubious, and must be deleted. Sapphorain (talk) 20:30, 8 December 2019 (UTC)
In the future of WD most statements (at least statements other than simple relations P31/P279 and meta-statements) should have a reference (other than P143-pseudosource), there is really no other way. But this can be achieved only by requiring references for new statements and adding references to existing data (just how it works now in pl.wikipedia). You can't delete most of WD data, usually correct data, only because of the lack of reference. Wostr (talk) 15:17, 7 December 2019 (UTC)
@Wostr: The other day to remedy the P143-pseudosource situation I imported some sources that are required by Zmarli w... series of articles (e.g. Zmarli w grudniu 2019 = Deaths in December 2019) in the Polish language Wikipedia. Deaths dates are rather sensitive data so this practice is doable and advisable IMO. Kpjas (talk) 12:04, 8 December 2019 (UTC)
This issue was discussed in detail in September. I will reiterate that statements should be verifiable even if not currently referenced, and should not be removed merely because they currently lack a reference, unless they are clearly vandalism, bogus, or violations of Wikidata:BLP. Generating references is in most cases too damn tedious: in some cases one might need to create an item for a book, then an item for an edition of the book (none of which is automated), along with page number, etc. just to accurately reference a simple fact like birth date. We need more tools to auto-generate items/references and better bots to mine and parse existing simple data from External identifiers. Easier generation of references will result in more referenced statements. -Animalparty (talk) 18:07, 7 December 2019 (UTC)
Support Yes, please, references should be MUCH easier to add in Wikidata, currently it is painful. Why does the same reference have to be added on every data item, for instance, rather than having it once on the page with a reference to it, the way it works in the regular wikipedias (yes, I know there's a tool to make this easier, but it's still annoying). And I frequently default to just a "reference URL" link when it should maybe be something better structured. ArthurPSmith (talk) 19:14, 9 December 2019 (UTC)
Yes. Common sense should always take precedence over pedantic adherence to technicalities. If an uncontroversial, self-evident/self-referencing statement is added in good faith, let it be. -Animalparty (talk) 18:51, 7 December 2019 (UTC)
And 2 more cents: Wikidata is a volunteer project with little standards, few guidelines, and lots of confusion, omissions, redundancy, and inefficiencies. It's basically the Wild West, anything goes, whatever works. There is no editor-in-chief. The data should thus never, ever be presumed to be authoritative, complete, or accurate (even if referenced!). Any and every good-faith effort to improve the database, even if unreferenced, should be welcomed. Any effort to remove data rather than improve data (which makes the project worse, not better) should be strongly discouraged. -Animalparty (talk) 19:07, 7 December 2019 (UTC)
Some confusion may be due to the proposed guideline Wikidata:Verifiability, which currently contains the statement "The majority of unsourced statements, and statements not supported by the source provided, will be removed from Wikidata." I think this is terrible advice, for reasons already discussed. -Animalparty (talk) 19:32, 7 December 2019 (UTC)
Yes. It looks like nobody has really engaged with that page since January 2015, and not much even then. In its current form it is profoundly misguided, does not represent the practice or norms of the community, and ought to be fixed. Jheald (talk) 20:11, 7 December 2019 (UTC)
Support removing the line ""The majority of unsourced statements, and statements not supported by the source provided, will be removed from Wikidata" immediately.
Additionally, if we're agreed that external identifiers and book/article metadata are self-referencing, then we should stop counting such statements as unreferenced for reporting purposes, especially as the number of scholarly articles increases so rapidly. - PKM (talk) 21:05, 7 December 2019 (UTC)
My philosophy, and I think one that should be widely adopted across all Wikimedia projects, is "be a wrench, not an axe". Improve, don't remove! Be constructive, not destructive! Fix it, don't nix it! Correct it, don't wreck'd it! (I'll stop now...) Sometimes there are cases where removal is better (as in clearly erroneous data, or cases of overly detailed content and excessive external links in Wikipedia articles), but in a project like Wikidata, where issues of aesthetics tone, relevance, and composition are largely moot, a hatchet approach is generally unhelpful. -Animalparty (talk) 23:35, 7 December 2019 (UTC)
Many statements have implicit reference, like an id to another online database. Many P31's are never referenced, and there is no need for it. This request would just delete all work done in the last 8 years of Wikidata-existence. My answer to the question "should we nuke the whole project" is simply F... O... Edoderoo (talk) 23:41, 7 December 2019 (UTC)
I brought up the same question about Nikkimaria (talk • contribs • logs), she deletes information from Wikidata when the information is deleted from English Wikipedia based on voting at English Wikipedia. Religion was removed from English Wikipedia by a vote and she is deleting that information here too. --RAN (talk) 23:20, 7 December 2019 (UTC)
Chiming in with the rest to say that I find the deletion proposal incomprehensible. How to improve referencing, though, is a worthwhile subject of thought. I think Help:Sources is quite well-written. But there is no allusion to references in either of the Tours. And the "How to get started" section of the Introduction contains a link to Help:Items but not Help:Sources.
I myself, having only been working with wikidata for less than a year, find the issue of what and how to reference one of the hardest things to figure out. For instance, for humans (not living ones, that's another can of worms), you may not have a source that explicitly states so-and-so's gender but you are 99.99% sure that his gender was "male" because he was referred to as "Mr.," he attended an all-male school, etc., etc., etc. In that case either no source is an appropriate reference for gender or everything is. This is not one of the cases discussed in Help:Sources/Items not needing sources. Levana Taylor (talk) 04:06, 8 December 2019 (UTC)
Thanks! my general point was that there needs to be more guidance and help for new/casual editors. I'm mulling over concrete ideas for what that might be. Levana Taylor (talk) 15:33, 8 December 2019 (UTC)
If the only way to push people to add references when creating statement is to threat that unreferenced statement will be deleted, then I am in favor of keeping that threat. Statement without references should be the exception not the rules. Snipre (talk) 21:10, 8 December 2019 (UTC)
Snipre Adding references to statements is great, but often easier said than done. It is often prohibitively tedious. Would you want to go through all 10 steps at Help:Sources#Books every time you wanted to reference a statement? -Animalparty (talk) 00:55, 9 December 2019 (UTC)
why all 10 steps? some are not obligatory ;) In some areas it may be quite annoying, e.g. if we have one information per source, in some areas it's easier (hundreds or thousands of values from one source). But I see no other choice that to create new items to reference statements, even if it's annoying. Maybe it's that way, because I came from a project where every new non-evident piece of information has to be referenced (or it's deleted). Not every statement needs a reference in WD, some are self-referencing, but many do need a reference. The only thing that stops me from referencing some new statements I add (that I think should have a reference) is not that creating items for sources is tedious or annoying, but that I don't know how to properly create items for that sources (e.g. legal acts and amendments to them). BTW I think that there was a bot in en.wiki that was used to populate cite templates – there should be something like that in WD (user creates new item with ISBN or DOI and bot imports the data). Wostr (talk) 05:42, 9 December 2019 (UTC)
User interface - For me, if we had a dedicated reference box that allowed you to slide a website into it, that would be great. But after adding data(sometimes needing multiple changes because you only need to add 1234 to certain statements because the abc part of the address is automatically implied), you then need to click this and add that, which pops up a box and then.... think like a new editor - if references are to be required, it shouldn't be a hassle to figure out how to source the one piece of data you are going to add. I mean, I am not a novice editor and I still get tangled up with certain edits. Could you imagine a first time editor being forced to add a reference using the current method of adding references. Quakewoody (talk) 14:47, 9 December 2019 (UTC)
Almost any piece of data about a living person might be controversial; anything that's individually challenged or might be challenged should be supported by a reliable public source or may be subject to removal. In particular properties that are instance of (P31)property likely to be challenged (Q44597997) should be supported by suitable references when applied to living persons. In the case of a dispute, the burden of evidence rests with the editor who adds or restores material.
Genealogy Bank and Ancestry developed an AI that could identify obituaries in the OCR of scanned newspapers and pull the data into a database, we can do the same to reference most data fields from prose text. --RAN (talk) 23:55, 7 December 2019 (UTC)
The quoted part of the policy speaks about "indiviudally challenged" to not allow someone to simply challenge a group of statements without looking at them individually and making individual decisions to challenge them. The RfC did ask whether there should be a bot that deletes unsourced statements and the consensus was against having a bot that automatically deletes such statements. ChristianKl ❪✉❫ 21:21, 8 December 2019 (UTC)
How to?
Thanks for your comments. Maybe I should clarify that many changes Sapphorain does are actually quite helpful: some review of the references we have or don't have is needed. Maybe we should be more explicit about various categories: what do with clearly wrong statements (with or without reference), what to do with unlikely statements, what to do with statements that are plausible but currently unreferenced. Personally, I think all statements can be referenced .. --- Jura13:25, 9 December 2019 (UTC)
I think clearly wrong statements without references should be removed. Clearly wrong statements with supporting references should be deprecated and information about why they are wrong. When it comes to unlikely statements without references I see that it makes sense to remove them.
When it comes to plausible unreferenced statements I wouldn't remove them. I find it important to keep the barrier to entry to edit Wikidata low and there will be plenty of new users who enter claims without references. I proposed in the original RfC that it would be good to have a bot that writes users who add new statements to living people that are privacy sensitive messages urging them to add references. ChristianKl ❪✉❫ 13:52, 9 December 2019 (UTC)
@Jura1: Re: "all statements can be referenced", sorry if I'm sounding like a broken record here, but do you feel that image (P18), image of interior (P5775), etc. require a reference? And, if so, what would be an example of referencing them appropriately? How about
@Jmabel: I'm just saying they can be referenced, but I don't think they have to be referenced nor would it be efficient to spend time to add references to them before many other things. Even for the statements you mention, this can be crucial:
Going through items for paintings, I found that sometimes it would be useful to add a reference that an image on Commons is one of that item. Many potential creator's signature (P7457)-values lead to a similar experience.
❶ Using multiple lexemes (example: Großeltern/Grosseltern (L3146) vs. (L228305)) would seem to require a lot of duplicate work and data to be maintained, as most non-spelling-related (and non-language(-variant)-related) statements tend to be identical across the respective lexemes. However, it does allow for each lexeme to be assigned its own language (variant) (Großeltern/Grosseltern (L3146) is German (Q188), (L228305) is Swiss High German (Q1366643))—which in turn would have to be taken into account in queries.
@BlaueBlüte: I thought the second approach was the preferred one, and I don't follow your statement that "it doesn’t allow the variants to be described as belonging to separate language( variant)s" - it actually requires that, I believe. Your example Urgroßvater/Urgrossvater (L34049) has one version labeled "de" and the other "de-ch". ArthurPSmith (talk) 19:02, 9 December 2019 (UTC)
@ArthurPSmith: What’s labeled de and de-ch, respectively, are the individual-string representations of the multilingual string that is the lemma (essentially the multilingual value in the statement having the property wikibase:lemma), whereas the lexeme itself is marked as being of <http://purl.org/dc/terms/language>German (Q188)—which shows up in the web interface on the line “Language” just underneath the page heading “(L34049) [lemmas]”, where only a single item is allowed as its value (containing German (Q188) in this case). (See also this SPARQL query looking behind the web interface. Also, Wikibase developers, please correct me on my terminology as needed.) So I’m not sure how I would query for ‘all specifically Swiss High German (Q1366643) words’, for example. But it probably wouldn’t be as easy as a clause like WHERE { ?lexeme dct:language wd:Q1366643 }. (Cf. this minimal query that works for the (L228305) case.) And similarly, the individual-string representations of the forms (values for <http://www.w3.org/ns/lemon/ontolex#representation> statements, see query) are marked as de and de-ch, respectively. Again, I’m wondering how one would query, e.g., ‘only Swiss High German (Q1366643) forms’. So, yes, in the case of approach ❷, too, in a sense the variants can be labeled as languages or regional variants, albeit ISO-defined ones rather than Wikidata items, and not in a way that would strike me as particularly helpful for querying.―BlaueBlüte (talk) 01:40, 10 December 2019 (UTC), updated BlaueBlüte (talk) 04:53, 10 December 2019 (UTC)
Thanks for informing me of this discussion, @BlaueBlüte! I don't really have too much to add since I'm not well versed in the technical side of Wikidata lexemes. It does seem confusing and impractical that Großeltern/Grosseltern and Urgroßvater/Urgrossvater have different representations, though. And it's not clear that language codes are sufficient to represent the Serbian case, where there are Latin and Cyrillic versions of each word (e.g., Latin "pȁs" and Cyrillic "пас", meaning "dog"), but there aren't distinguishing language codes. TJones (WMF) (talk) 14:59, 10 December 2019 (UTC)
@TJones (WMF): At least as for the case of Serbian (where, I think, varietal differences are more limited to pure spelling in either of two alphabets, whereas varieties of German involve regional and lexical differences more pronouncedly), RFC 5646 specifies that w:en:ISO 15924 writing-system codes can be used to amend language codes so as to convey which writing system was used for the language in question. See for example вода/voda (L2068), even though those codes (sr-ec and sr-el), per RFC 5646, would mean ‘Ecuadorian Serbian’ and ‘Serbian of the undefined EL geographical region’, and should perhaps be replaced with sr-Cyrl and sr-Latn, if RFC 5646 is indeed supposed to be applicable. Pinging @Infovarius as the main editor of вода/voda (L2068).―BlaueBlüte (talk) 20:33, 10 December 2019 (UTC)
@BlaueBlüte: Ah I see your concern. Maybe the data model here does need some adjusting? However, for these cases of variants isn't it generally true that there is a general "language" that applies ("German" in this case) while the specific lemmas are regional variants (or perhaps "standard" and "variation")? ArthurPSmith (talk) 19:09, 10 December 2019 (UTC)
If we do get into revisiting the model, we might want to think also in terms of historic spellings. E.g. Romanian had a spelling reform that (among other things) changed the use of â and î, but older works would have the old spelling.
What do we currently do about the different systems for writing Chinese? - Jmabel (talk)
2:
How can I filter or look for a certain Amsterdam code (P6434) (10240 for example)? I only found filtering on Q entities, year and labels. I would like to do this so I can have a query which shows the municipalities replacing each other through the time. My query: https://w.wiki/DcH.
Has anyone already made a Phab ticket for this? quickstatements appears to be down with the message "You can't create a new batch, because you are blocked on Wikidata". Did someone block Magnus?!? Jane023 (talk) 19:51, 9 December 2019 (UTC)
We will investigate on this issue as soon as possible. If you're encountering this issue, please let me know, so we can estimate who/how many people are impacted. Thanks for your patience! Lea Lacroix (WMDE) (talk) 08:13, 10 December 2019 (UTC)
@Lea Lacroix (WMDE): I have the same issue as well (actually, I went to this page to ask about it, but it is already here). The problem appears for both my base and my bot account (neither are actually blocked), from different browsers and from different devices (not platform-related issue), with and without VPN (so it is not an IP-issue).--Tohaomg (talk) 10:49, 10 December 2019 (UTC)
Is there any gadget or something that would allow me to add more than one predefined descriptions to an item? Sometimes I correct items that had incorrectly added descriptions based on instance of (P31). There is a good tool to clear all descrptions, but I would like to add correct descriptions, at least for several languages (and doing it manually is quite annoying, especially if I know the proper description, but I don't have specific language defined on my userpage). Wostr (talk) 22:59, 10 December 2019 (UTC)
@Wostr: Are you thinking of something like labelLister (a gadget you can enable in your preferences)? That'll help with adding languages not on your userpage, and the UI is a little faster than adding everything. Vahurzpu (talk) 00:15, 11 December 2019 (UTC)
@Vahurzpu: I have it enabled, but still it is not the way to add descriptions in reasonable time. Usually I delete more than 50 incorrect descriptions using dataDrainer, but I'm able to add only a few descriptions (and I could be adding from several to over 50 descriptions, depending on the corrected item). Correcting an item (changing P31/P279, checking ids, linking to other items) takes a few minutes, but adding even a dozen descriptions in languages I don't have on my userpage would double this time (copy-pasting from other page). Even a tool that would allow to add descriptions to item X from labels of item Y would be great. Wostr (talk) 00:35, 11 December 2019 (UTC)
I have an inelegant solution for that problem, I use LabelLister to export the labels of item Y and then make a Google Sheets template to add those values as descriptions, which I use as input for a QuickStatements batch on item X. I have about 30 of those sheets I keep for adding everything from “British fashion designer” to “artistic theme”. - PKM (talk) 01:24, 11 December 2019 (UTC)
Aucun en particulier Ayack :). Je posais la question car il m'arrive fréquemment de tomber sur des éléments supprimés dont il est apparemment impossible de connaître le sujet sans être admin/leur demander. Nomen ad hoc (talk) 16:47, 11 December 2019 (UTC).
Par exemple, quand je regarde la liste des éléments créés par un utilisateur, si ceux supprimés excèdent une dizaine, je ne vais sans doute pas demander sur l'AN. Nomen ad hoc (talk) 19:04, 11 December 2019 (UTC).
No, there is no other option than asking an admin. It is sort of the purpose of a deletion to make the page content invisible to the general public. —MisterSynergy (talk) 17:01, 11 December 2019 (UTC)
Right. But some do mention the former item title in the admin's deletions comment (and I find it helpful). Is this a fair processing? Nomen ad hoc (talk) 19:07, 11 December 2019 (UTC).
Some background: the deletion form has two inputs; the first one is a drop-down menu which is fed from MediaWiki:Deletereason-dropdown (see source for all contents), of which one of the reasons has to be selected; "other reason" is the default entry. The second input is a free-text input form called "Other/additional reason", and it is pre-filled with the Content was: "SOME LABEL" string by the software. The software concatenates both inputs during a deletion as the seletion summary. As an admin, one can decide to either empty the free-text box, or modify the pre-filled text, or replace it with something completely different. I personally prefer to use the pre-filled text in most cases, but sometimes I need to remove sensitive information or the entire "additional reason" string as well. Other admins do it differently, and there really is no policy what exactly to write in the free-text input field. Mind that there are sometimes also people generally unhappy with providing information in the deletion summaries, as e.g. in phab:T239012. —MisterSynergy (talk) 07:34, 12 December 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Moebeus (talk) 01:26, 19 December 2019 (UTC)
@Pigsonthewing: I usually use <affiliation>=[university] qualified with <position held>=professor emeritus (Q1240569) and a start date if I have it (and sometimes I add <stated in reference as>= "Professor Emerita of Anglo-Saxon History" or whatever in the references). - PKM (talk) 21:00, 10 December 2019 (UTC)
I'm not sure that there's much consistency in the title: some holders may still be employed and receiving a salary, and others may be fully retired. Ghouston (talk) 02:08, 11 December 2019 (UTC)
TLDR version: see here for what I think is the best available source. The long version:
The Indonesian Wikipedia article does not cite any references except the Ensiklopedia Sastra Indonesia Modern (Encyclopedia of Modern Indonesian Literature). That is probably a perfectly good source, it was reviewed in a scholarly article for one thing. However, its contents are not accessible online.
Asmara Hadi (1914-1976) berpendidikan MULO Taman Siswa Bandung, pernah giat di pers dan politik, sempat menjadi anggota Konstituante dan anggota MPRS (1966). Dia terkenal dengan sajak-sajak perjuangan yang penuh keyakinan. Kepenyairannya telah dibahas J. U. Nasution dalam Asmara Hadi Penyair Api Nasionalisme (1965).
(Google translate: Asmara Hadi (1914-1976) educated with MULO Taman Siswa Bandung, who was active in the press and politics, was a member of the Constituent Assembly and a member of the MPRS (1966). He is famous for his poems of struggle full of confidence. His words have been discussed by J. U. Nasution in Asmara Hadi Poet of Nationalism (1965).)
There are plenty of other Indonesian biographical sources in books but only seen in snippet view via Google. Luckily, enough can be seen of Ensiklopedi sastrawan Indonesia, Volume 1, page 124 (Indarti Yuni Asuti; Jakarta: Permata Equator Media, 2008) to give a birth date. "Asmara Hadi A. Riwayat Hidup smara Hadi adalah tokoh pergerakan nasional yang JL A.juga tercatat sebagai tokoh perintis surat kabar di Tanah Air. Ia terlahir dengan nama Abdul Hadi pada tanggal 18 September 1914 di Bengkulu." (Google translate: Asmara Hadi A.'s life history is a national movement that JL A. is also noted as a pioneering figure in the Homeland. He was born under the name of Abdul Hadi on September 18, 1914 in Bengkulu.)
An English-language article in Indonesia Circle no. 51, page 34, likewise, in a brief allusion, gives dates of 1914-76 and calls him a "Sumatran poet from Bengkulu." (male pronouns used for the poet in this article).
All the other books & articles I've seen in snippet view--about 10 of them--state a birth date of 1914 but not a day.
There is an online Ensiklopedia Sastra Indonesia which has a lengthy article about Hadi & gives a birth date of 8 September 1915. No references cited for that. Likewise without references is the death date of 3 September 1976.
However, I think the best online source is this periodical article in Koran Sulindo -- it cites no references but is detailed and apparently informed. Says he was born 8 September 1914, in Lubuk Ngantungan, Bengkulu, and died 3 September 1976 in Jalan Cilantah, Bandung.
There are various other unreferenced online bios which all say 8 september 1914. I think the Koran Sulindo article is the primary reference to go with; the "18" rather than "8" in Ensiklopedi sastrawan Indonesia would be an easy misprint to commit. -- Levana Taylor (talk) 20:22, 13 December 2019 (UTC)
Thanks, fixed now and the typo identified. It was his date of death that was wrong. I find a dozen each week of transposing number typos. --RAN (talk) 23:18, 13 December 2019 (UTC)
Short movie with people being interview but not real cast
Hi,
I have a question. I made a new item for a youtube movie for which several people where interviewed. I tried to list these people as cast member, however then I get a constraint error. The people interviewed are not actors but scientists. How can I state which people are appearing in the movie / documentary without them being actors?
“News” does NOT allow “Publication date”, “Published in”, “author” nor “full work available at”
How is one supposed to code a news report?
Example: Q78512157 documents a front page article in the New York Times. The first thing I'm asked to document is "Instance of" (P31). So I select "news". Then when I supply “publication date" (P577), "Published in” (P1433), “author” (P50), and “full work available at” (P953) each one is flagged, "This statement has some potential issues". When I click on the exclamation point in a circle, I see that it violates a "type constraint", saying that "Entities using the ... property should be instances of one of the following classes (or of one of their subclasses), but U.S. Ramping Up Major Renewal in Nuclear Arms currently isn't:", one of the acceptable classes is "work". I made "news" a subclass of "work", but that didn't fix the problem.
I updated the instance of (P31) value of your item from news (Q38926) to news article (Q5707594) in Special:Diff/1076221973, and this fixes your problem. "I made news a subclass of work" is usually not a good approach, except for the rare cases when there are clear errors in the ontology (we get much more new errors due to such fixing attempts than actual fixes). Btw. your incorrect fixing attempt did not remove the warnings because you made news (Q38926) subclass of work (Q6958747), but work (Q386724) would have been the one you were looking for. Nevertheless, as long as you raise this problem here, there is nothing that you really messed up, so no offense! Such problems can appear quite complicated sometimes. —MisterSynergy (talk) 22:12, 14 December 2019 (UTC)
External identifiers unavailable outside specific countries
Are there any ways i can mark an identifier in order to indicate that the links are unavailable outside of specific countries? This is a problem i often encounter whenever someone makes an identifier about an online film/television streaming service. --Trade (talk) 22:31, 14 December 2019 (UTC)
Hi, I started to categorize these terms in subcategories e.g. category theory terminology (Q77544558). I'm using petscan. When I increase the depth I get unexpected results. Can someone explain why ordinary numbers is in a subcategory to sublevel 5 of this query https://petscan.wmflabs.org/?psid=14146700&al_commands=P31%3AQ77544558?
Maybe I'm not using the tool correctly or I need to exclude common mathematical concepts from the results.
I see now that category theory is actually an area of algebra, among these other areas: Abstract algebra, Category theory, Elementary algebra, K-theory, Commutative algebra, Noncommutative algebra, Order theory, Universal algebra. source
It can also be devided in Abstract algebra, Linear algebra and Elementary algebra. (this is what wikibooks does it seems)
@so9q: Does category theory studies category theory terms ? No. Category theory object is categories. In the same spirit, items about category theory in Wikidata are not about category theory terms. This would be the same as saying « a Wikidata item is about its label » … meaningless. A category theory item on Wikidata is about a category theory notion (category, functor and so on). It’s not about the names « category », « functor », … Terms are for the lexical part of Wikidata. author TomT0m / talk page14:52, 6 December 2019 (UTC)
Interesting. I use P31 as a device to easily e.g. easily be able to round up all lexemes in a query that is related to mathematics or a specific subfield of mathematics.
I welcome anyone to improve all these Q-items that e.g. have no other statements than the general P31 I gave them (and remove the P31 when we have another feasible way of selecting it in a query). Take a look at e.g. https://tools.wmflabs.org/reasonator/?q=Q281126 and tell me how you could easily select this as a law concept in a query (ignoring the P31 law term).
My goal is to be able to make a dictionary of law terms (based on our lexemes) with a single reasonably simple query that include all q-items that satisfy some conditions, e.g. part of law, P31 law term, P279* (this is liable to time out for big taxonomies like mathematics because of the way blazegraph work it seems) or what ever. This is currently impossible because many q-items lack proper P279, studied by, part of or P31 statements.
My second goal is to make a tool that can take a text and describe all e.g. legal terms found it it (similar to https://tools.wmflabs.org/ordia/text-to-lexemes) which could be an invaluable tool for e.g. students of a certain profession.
You're really not getting this, So9q. P31 is not a coathanger for any old crap you care to fill it with in order to make your simple queries work. You have multiple people here telling you you are mistaken in your use of P31. Your response seems to be "someone else can remove the wrong P31 crap if they come up with some other means of keeping my queries simple". The ideal here is to do thing *correctly*, not to do things "easily" because you cannot be bothered with complexity / cannot be bothered to learn how wikidata works. I'm sorry you lack the imagination to see that your simple approach screws wikidata up for other users who naively expect that P31 will be used as intended, not as some sort of bookmark system for your personal queries. --Tagishsimon (talk) 20:29, 6 December 2019 (UTC)
I really dislike the tone of you reply. Could you please calm down before writing replies? I'm still learning the ropes here and if this kind of tone is accepted here I'm not sure I wan't to be part of discussions here or a part of the community at all. I have no problem whatsoever adapting my way of doing things if I get good arguments and a friendly point in the "right" direction. This is not the first time I find a reply of yours @Tagishsimon: to be out of line. I hope others who read this can recognize the toxity of this kind of communication style. Would you be willing to make an apology?
To anyone else reading this I would like to point out that I followed a praxis (way of doing something) above that was established and NOT confronted by anyone (to my knowledge) until I raised the subject here in a way that to me looks amicable and geared towards finding a solution acceptable to everyone. Maybe I'm stepping on a really sore toe here without my knowledge?
To attack me as a newcomer for doing something that Tagishsimon himself in september 2019 (and the others pinged above) has contributed to is just plain ignorant. Don't kill the messenger. I urge anyone reading this to instead focus their energy on fixing the problems. I can roll back my latest edits if anyone wants me to do that or change them from P31 xx term to P361 xx terminology. As I stated above I don't really care how we agree to do this, I just think it should be doable to link the power of lexemes with the diversity of Q-concepts.--So9q (talk) 21:32, 6 December 2019 (UTC)
@Wikisaurus, TomT0m: Is this Q-item better/acceptable (I use part of xx terminology instead of P31 xx term)? I had to create the whole taxonomy of this terminology because it did not exist. I feel like breaking new ground here so I would be very happy if other would chime in and tell me (preferably in a constructive & friendly way) what to change/improve if anything.--So9q (talk) 21:32, 6 December 2019 (UTC)
I changed Q77544558 to be a part of algebra terminology which is a part of mathematical terminology. I will now proceed with changing all P31 Q77544558 to P361 Q77544558 instead. Does this take care of your expressed concerns?--So9q (talk) 09:56, 7 December 2019 (UTC)
@So9q: No, not really. This is a database of concepts and how they relate, not of terminology. The terminology (Q1725664) subtree should be avoided to the greatest degree possible. Recourse to it usually inidicates that you haven't thought enough about what something actually is. Jheald (talk) 19:31, 7 December 2019 (UTC)
Oh, thanks for voicing this. I understand your concern. Is this concern shared by a majority of the community? If yes whey are the current x terms and x terminology tolerated (no deletion requests?)? I see no discussions going on on the talk pages voicing this concern at all either.
Maybe we could see this terminology work as a precursor to a fully described wikidata where all concepts related to psychology have at least 5 statements (not 0-1 as they do now). Then we can remove the terminology statements because we then have other effective ways to select what we want in queries e.g.--So9q (talk) 21:15, 7 December 2019 (UTC)
@So9q:, can you undo your categorization yourself? I do a number of edits on mathematical entries and see that they are still there, although the way to categorize via instance of (P31) and "smth term" is certainly wrong, as you can see from comments above. Wikisaurus (talk) 21:47, 8 December 2019 (UTC)
@wikisaurus: Sure, I was waiting for you first to answer my question above. Is this Q-item better/acceptable (I use part of xx terminology instead of P31 xx term)?. I understand from Jheald that it is not perfect but I hope it is acceptable until we have a higher quality items in general.--So9q (talk) 22:09, 8 December 2019 (UTC)
@So9q:, well, it is the same - a term in a part of terminology, and a concept is not, it is rather a part of science. But blank (Q368951)part of (P361)analytical chemistry (Q2346) would not be a good solution either, because it would pollute part of (P361), so I would really prefer to have a separate property for this. I still see some problems with such a property, like what depth should one chouse (why not absurdly specific blank (Q368951)studied bychemistry of blanks?), but it would be much more neat. What do you think about creating a property proposal? Wikisaurus (talk) 22:17, 8 December 2019 (UTC)
Yes, I like that too. These special concepts as the ones mentioned in the proposal are not well known outside the academic tower.--So9q (talk) 20:32, 9 December 2019 (UTC)
Best practices for a subject with multiple VIAF entries?
Very definitely add all three to the item, so the VIAF people can see that we think some of their entries refer to the same person. Typically I think they actively pick this up and re-examine such cases within a month.
I would not make any of the identifiers "preferred", as this might make it harder for VIAF to spot that we think they have multiple live identifiers for the person. (Unless somebody in contact with VIAF can say this wouldn't make a difference). Jheald (talk) 23:38, 14 December 2019 (UTC)
They have been very good at using our duplicate entries as a basis for merges at the VIAF end. For more complicated VIAF errors you have to add them to Wikidata:WikiProject Authority control/VIAF errors. These are where people have been merged improperly and they have to be teased apart again. Unfortunately some of these errors are caused from our end, when we make a merge error or when we have a typo in the birth or death date. I have corrected over a hundred errors where we imported the "flourished" date as a birth or death date, or where we imported the dates of the dynasty from VIAF or from Wikipedia, for royals, as birth and death dates leading to implausible longevity claims. See Wikidata:Database reports/unmarked supercentenarians. --RAN (talk) 19:04, 15 December 2019 (UTC)
Data portal
Hello,
in the last months I have thinked about how the data of Wikidata could be used from a lot of people and what of the possibilities are easy and where there are problems for users if they are not people with a bigger knowledge in Programming and Query languages. In some Wikipedia language Versions there is a Portalnamespace. The sites of a Portal help to get a overview about the topics. I think portals could help people to understand the structure of the data in Wikidata and it could be places to collect queries to get the data related to the topic. I think this is something what would help people to get more information and to understand what the data is about. Something what is also iimportant is how the data was created. I often dont understand completely how the data was created. For the overview there are the items and for detailed information and for the needed understanding you need before you can use the data there are the articles. So both things are important. What do you think about establishing Portals for the data. Is there somebody who works in a portal and is interested in testing this. I would help as far as I can write Queries and also if I understand the topic in creating a overview. -- Hogü-456 (talk) 21:07, 14 December 2019 (UTC)
I would be weary of starting a new type of page on Wikidata for this as there a good chance that it will be another place that gets little editing and is in the end not helpful. I think it would make more sense to integrate such content into the existing Wikiprojects. ChristianKl ❪✉❫ 21:32, 14 December 2019 (UTC)
Sounds like you are looking for Wikidata:WikiProjects, and some other more general pages in the Wikidata: namespace that provide technical background (Wikidata:Data access for instance). I agree with you that documentation is important, and quite often we do not do this properly. However, WikiProjects are the venue to document the work in a certain area, and to meet other users that are interested in your field of work to collaborate with them. You can describe how data in a certain field is modeled, where you can import from and who does this regularly, which maintenance jobs are available, create (Listeria) reports to monitor data quality, and so on …. It is unfortunately somewhat difficult to find the correct WikiProject sometimes, but I guess that this would not be any different for "Portals" either. There are no formal barriers to pass when you want to start a new WikiProject, but if you do so, do it properly in order to create something useful… Many existing WikiProjects have "participants" lists where you can simply add your name and start/improve documentation. —MisterSynergy (talk) 21:49, 14 December 2019 (UTC)
Wikidata Projects are great, but what we desperately need is something akin to a Project:Manual of Style (Q4994848), a collection of "Best practices" recommendations with the goal of standardizing data management, reducing confusion, and improving the efficiency of queries. It's become clear that almost nobody knows how things should be done, they just do whatever. Confusion is rampant about what to use for the fundamental property instance of (P31) in many fields like books, occupations, and businesses (do we use book (Q571) or written work (Q47461344) or novel (Q8261)? Should we use business (Q4830453) or supermarket (Q180846)? Both? Who knows?!) It seems there are hundreds of secret discussions happening in the bowels of a dozen of subpages, but consensus is sorely lacking. And if there is consensus, it's not widely circulated. Confusion persists. -Animalparty (talk) 22:30, 14 December 2019 (UTC)
I understand that you mean that adding more subpages is not helpful. I think it would help if it is possible to get from a item what is a subclass of another item with a link in a property to the related WikiProject. Then it is more easy to find the right wiki project. If the information needed is not there then it is possible to ask a question at the discussionpage of the WikiProject. Maybe it is also possible to create a property with a link to the description how the data about the topic should be modelled. -- Hogü-456 (talk) 21:03, 15 December 2019 (UTC)
How to handle dead url in reference ?
Hi,
I was looking for versions of Natron (Q20707858). For versions > 2.3.5, most of the time, there are 2 references, nearly the same; the earliest reference leads to a dead url. What should be done? Is there a property to notify about it or should it be just removed? Thank you !
In general I wouldn't advise removing a reference URL just because it no longer works. It could be replaced by an archive.org version if one exists. But in a case like this where there are two references and one of them is nonfunctional, in my opinion it would be fine to just remove the one that doesn't work. ArthurPSmith (talk) 14:42, 12 December 2019 (UTC)
Recoin displays information about the relative completeness of a Wikidata item by comparing its statements with those found on other similar items. Especially useful for editors working in a knowledge area that is new to them.
Jmabel: Its only valid sitelink was deleted for lack of notability, then the IP blanked it. The same would have occured here, despite having different notability criteria. Esteban16 (talk) 21:13, 16 December 2019 (UTC)
Hello. Why is this language named Crimean Turkish on Wikidata?
"Crimean Turkish" is not used at all
And it's widely known that the first name of this language is Crimean Tatar, not Turkish
See these links:
@Devlet Geray: please restore the statement and alt labels you just deleted on Q33357. Wikidata attempts to include all variants. Do not attempt to suppress the ones that don't fit your POV. --- Jura12:20, 10 November 2019 (UTC)
Somebody who experienced in opening tickets should propose to fix the tooltip that's now showing Crimean Turkish. There are around 1500 ethnic Turkish people living in Crimea, and historically Crimea belonged to Turkey/the Ottoman Empire, so it may sound like Crimean Tatar is a kind of Turkish dialect. Even if this term is in use by somebody (in several English- and Turkish-speaking sources), it doesn't stop being simply incorrect. --Wolverène (talk) 11:01, 11 November 2019 (UTC)
It's not "simply incorrect", IANA does list both as valid names for the language but it lists Tatar first and there's no good reason to prefer Crimean Turkish over Tatar, so I agree that a ticket would be good. ChristianKl ❪✉❫ 13:24, 11 November 2019 (UTC)
As usual, requests are made in the issue tracker: http://cldr.unicode.org/index/bug-reports . crh doesn't have a locale, so this would be a change to the English name in the "Locale Display Names" rather than an "autonym" change. Remember to cite sources for your claim and don't consider anything to be obvious or "widely known" (in Western languages, by the way, "Turkish" is about as ambiguous as "Indian", another word which was used across the centuries for widely different things).
I opened a ticket with CLDR almost a year ago to change the default English name of Qırımtatarca to Crimean Tatar (I cleverly go the names backward in the ticket title, but added a comment explaining the right thing). Making comments there can't hurt, and might help. I also have a Phab ticket from a year and a half ago for Mediawiki to override the CLDR name for Qırımtatarca in English. It stalled; I think comments there would help. For Wikidata, I just opened a new ticket. Feel free to comment there, too. TJones (WMF) (talk) 15:57, 10 December 2019 (UTC)
Wikidata is a American database, so it's not required to honor the right here. Data consumers should notice that.--GZWDer (talk) 12:28, 10 December 2019 (UTC)
So we should be cautious while importing EU databases. Others does not have such problem. In Europe various databases (including non-EU based ones) may be protected by database right.--GZWDer (talk) 12:38, 10 December 2019 (UTC)
No, the EU database directive protects EU-based databases only. If we were to ignore the sui generis database rights of EU-based databases, we’d effectively render Wikidata useless for all EU-based users due to legal risks. I can hardly imagine that this is what we want. —MisterSynergy (talk) 12:46, 10 December 2019 (UTC)
I removed the warning as it might encourage people to upload data protected by sui generis rights to Wikidata in it's current wording. ChristianKl ❪✉❫ 07:00, 17 December 2019 (UTC)
Hi, I'm confused about the relative purposes of solved by (P1136) and proved by (P1318) when stating the person who first proved a mathematical theorem. It seems to me like proved by (P1318) would be a better fit whenever the entity is about a theorem, and solved by (P1136) should be reserved for when the entity is about a problem or conjecture -- but I have often seen solved by (P1136) applied to theorems. What would you think is the right way to proceed here? --A3nm (talk) 14:53, 14 December 2019 (UTC)
At English Wikisource, in the past few days, pages moved in our Author: namespace are not automagically being updated in Wikidata. I am only noticing it in the one namespace, though there are probably not enough moves in other namespaces to draw conclusions at this point. Is that a problem to be reported and managed here, or does it get reported elsewhere. What information would be required for further analysis? — billinghurstsDrewth07:01, 15 December 2019 (UTC)
Not updating here when paged are moved or deleted have been reported on svwiki too. Change of namespace then does not matter. 62 etc (talk) 09:02, 15 December 2019 (UTC)
This problem has also been occurring on English Wikipedia and Commons for the past couple days: following page moves or category moves, WD sitelinks fail to update, and thus become connected to redirects (often breaking the data connection). I don't know if it's just a a lag in servers, but previously Wikidata would change nearly instantaneously with page moves. -Animalparty (talk) 00:01, 16 December 2019 (UTC)
Comment agreed that is a related ticket. I recent move updated instantaneously, which would indicate that 1) it is fixed; or 2) it is sporadic. — billinghurstsDrewth05:03, 17 December 2019 (UTC)
phab:T240518 is the main ticket. It was a problem with a job queue that affected many tasks, not only renames of accounts, but also mass messages, wikidata updates etc. It is now fixed. --Kitayama (talk) 05:58, 17 December 2019 (UTC)
Not to derail, but this is very common in the music field as well, and it's getting messy when an artist releases music under 3-4 different names, but they all have to be linked to the same "performer" with wildly varying use of qualifiers, if at all. It's not trivial reflecting what releases where released under which names when querying for example. Maybe we should rethink the practice of not creating items for aliases, at least when they are well established with external identifiers from several sources? Moebeus (talk) 15:41, 15 December 2019 (UTC)
Yes, merge them. Add the pen names to the "also known as" list, as well as make a pseudonym statement. Add the multiple VIAF numbers with the 'named as' qualifier.
As to musicians (it could apply to authors and more) - something I learned from other sites (even though I disagree with it) is that sometimes individuals make music under different 'projects', and we treat the project as a "band" even though they are one person. It is most prevalent in the electronic/EDM type stuff. Where one man releases 7 different albums under 7 different names because each name is a "project" with its own "mood". Quakewoody (talk) 19:46, 15 December 2019 (UTC)
WD is kinda the odd one out here, as Discogs, Rate Your Music, All Music, Musicbrainz and most other music databases operate with separate IDs for separate artistic aliases. The practice of recording under different aliases is as old as time and occurs in all genres, there can be contractual reasons, artistic reasons, etc. I like Musicbrainz approach here - rather than trying to "correct" they try to adhere to "artistic intent" if identifiable. Mainly though, I'd like to see some SPARQL tutorials on how to query releases by different release artists when all linked to the same "performer" QItem. Ideally using Richard D. James as an example of course ;-) Moebeus (talk) 20:48, 15 December 2019 (UTC)
It should be relatively straightforward to identify the name under which the work was published using a object named as (P1932) qualifier on the creator (P170) or author (P50) statement. A query could require this in the usual way, to distinguish works released under different names.
Of course, P1932 is string-valued, so we lose the advantages that item-valued properties impose a limited controlled vocabulary. For that reason a case could be made for creating items for pseudonyms. But myself, I prefer the advantages of the present system, which helps collate together (and therefore make readily discoverable) all works by a particular creator, under whatever name. Jheald (talk) 22:30, 15 December 2019 (UTC)
@Jheald: To note that pseudonyms can be for joint authors, or some pseudonyms are also factory names. So it isn't going to be one-size fits all. I think that items for/as pseudonyms are valid, they are pseudonyms alone, and not the humans, they don't have any elements of humans. The issue then comes about how we list works for and against which? Suggest to check against instance of (P31) -> pseudonym (Q61002) — billinghurstsDrewth00:37, 16 December 2019 (UTC)
Of course, those other sites you mention have the luxury of serving a niche market, dealing only with music. Which is far different than WD which deals with everything and anything. You have to worry less about William Bailey (the song, and the name of Axl Rose) being confused with the many authors, scientists, President, book titles, germs, etc. Quakewoody (talk)
as a matter of fact, for decades, libraries have been creating new authorities for each identity, because there was no system to unify them... but now, with FRBR-FRAD, (requirements for authorities), which is now renamed IFLA-LRM, then unify authorities, and merge the different names under which the references were recorded... as it was said in a recent meeting about authorities (for librarians) : "une personne est une personne" -> a person is 1 person !
by asking to create an item for each ID, you want to perpetuate the old model, which is now in the way of being completely remodeled, even by the most old-fashioned libraries...
as said above, it is (and has been) the constant practice on wikidata to add pseudonyms on the person's item using P742, and to indicate on works items the identity under which it was published through qualifier P1932
the only "pseudonym" items should be for "collective pseudonyms"...
as for VIAF giving multiple IDs for the same person, it should not be an indicator to create multiple items, as the work of merging (they call it "clustering") authorities from many libraries, is still unfinished, and there are also a lot of authorities which only give a name (undifferentiated)[7], without dates or other infos, that, in fact, link works from different authors sharing the same name - those cannot be linked to a wikidata item, which are defined... --Hsarrazin (talk) 14:25, 17 December 2019 (UTC)
I have spoken to ISNI numerous times. They are often more than willing to make corrections, additions, and merges. They just have to be made aware. Quakewoody (talk) 15:02, 17 December 2019 (UTC)
Gus Lawson
Can someone delete the link to the photo at the article in English Wikipedia of w:Gus Lawson. A family member notified me that the image is his cycling partner. I corrected the image link here and at Commons but I am perma-blocked at the English Wikipedia. Normally I just leave the errors at English Wikipedia, and correct them here as best I can, but this one a family member asked me to correct. --RAN (talk) 23:06, 15 December 2019 (UTC)
Thanks! The collage was a place holder until I made a crop, and prevent the incorrect image from coming back. --RAN (talk) 02:00, 16 December 2019 (UTC)
Vote about structure of cities taxonomy following our november discussion
Hi, I suggest we have a vote for how to deal with the current mess. Infovarius recently undid my edits (which removed the constraints) on big city (Q1549591). I therefore propose that we details some proposals here and vote which one to follow. As I see it there are these options voiced in the previous discussion:
keep current mess with implications to data quality (unclear what to do with big cities, constraints or not, etc.) - hard to query in the current mess
change to a flat taxonomy as in Q5 where all cities/towns are P31:settlement and the size grouping is done via sparql based on the population property. Additionally a settlement can have a P31 describing its size as expressed by the local populace/local statistics bureau or historical classification but not any Q-items with constraints because we don't want to enforce a data model on the items.
multiple taxonomies side by side, one for Sweden based on SCB definitions, one for Denmark, etc. This makes quering internationally really really hard.
I'm trying to understand: are you saying that we get rid of designations based on charters? (e.g. in the state of New York distinguishing "city" from, for example, "village" or "unincorporated community", or distinguishing what type of entity is City of London (Q23311)vs.Greater London (Q23306))? - Jmabel (talk) 09:11, 14 December 2019 (UTC)
No, as long as you can provide a reference for the statement I'm fine with calling it whatever you want. The New York City (Q60) have 5 different P31 statements (but unfortunately none of them are referenced, yet). What I'm suggesting is that we get rid of the subclass taxonomy with e.g. city P279:human settlement because it leads to time outs during queries when traversing the whole tree as Andrew exampled in the discussion linked above.--So9q (talk) 17:52, 14 December 2019 (UTC)
Background
We have a lot of human settlements in WD but the way we categorize them leads to pains when querying and to us enforcing a size constraint like e.g. in big city (Q1549591). See the previous discussion.
Proposal
I propose the following change to our taxonomy of human settlements:
none of the P31 on settlements are allowed to have constraints based on size of population and should only describing the settlement (including its size) as expressed by the local populace/local statistics bureau or historical classification (ideally with a reference).
This is very similar to the human (Q5) taxonomy and hopefully leads to better and more consistent queries on settlements.
Voting
Voting started 14/12-19.
Please indicate whether you support, oppose or abstain below. E.g. {{support}}--~~~~
Thanks for taking the time to comment :). I understand you are in favor of the subclass approach, but unfortunately it leads to timeout because of the sheer number of human settlements on the planet and our WDQS limits. My proposal is believed to solve that issue because P31 is more efficient to traverse than a diverse subtree if I understood Andrew correctly in the discussion linked above.--So9q (talk) 17:52, 14 December 2019 (UTC)
Whether or not it takes more effort for Wikibase to transverse, is an implementation detail of how Wikibase decides to cache data. That implementation detail is not fixed and even if the current implementation would need more effort which you haven't demonstrated, that's no reason to simplify our powerful data-model. If there's a need to speedup those queries it would be better to do more caching inside Wikibase for specific queries like this. ChristianKl ❪✉❫ 20:29, 14 December 2019 (UTC)
Oppose You seem to want to denormalise in one area (P31=settlement), and forbid denomalisation in another area (P31=Big City). That's inconsistent and unreasoned. Nor am I convinced that you know enough about wikidata nor SPARQL to be proposing sweeping changes to wikidata. Nor am I convinced that a vote on any proposal is required; if there are genuine problems - and I'm not certain what you think these are, since you have not specified them here - they can probably be sorted out by adding more statements, rather than mandating a prohibition on the addition of statements. It might be that your proposal solves an issue you have; I'm not sure that you've done any analysis of the effects of your proposal on other uses of the items. --Tagishsimon (talk) 18:14, 14 December 2019 (UTC)
Support making arbitrary distinctions by size, things like big city (Q1549591) for populations over 100k, is arbitrary and complicates queries for no benefit. Should we have items for each population size in 10k intervals, or 1k intervals? Why not, it would be the same principle. Ghouston (talk) 23:12, 14 December 2019 (UTC)
The principle I'd apply is that where there's a property, in this case population (P1082), then it's redundant to duplicate this in a class hierarchy, and redundancy in databases is generally bad. Ghouston (talk) 23:45, 14 December 2019 (UTC)
Up to a point, Lord Copper. Wikidata does in fact endorse and recommend redundancy in a number of areas, such as P17 values, and reciprocal values - has part / part of, mother / child. --Tagishsimon (talk) 02:00, 15 December 2019 (UTC)
"Lord Copper"? When you start resorting to name calling, maybe you are on shaky ground? I would also say that inverse statements are an example of bad database design. I think these inverse pairs have been created only because of limitations in the Wikidata software, that it doesn't treat statements symmetrically / reversibly. I.e., X is the mother of Y should automatically imply that Y has mother X. Ghouston (talk) 22:41, 15 December 2019 (UTC)
Strong oppose It's a lot more useful for a Commons infobox to be able to say "hamlet", "village", "town", or "large city" (translated into Gaelic or Welsh or Breton or whatever) to explain what sort of place a category is for, than just "human settlement". As for the point on querying, a query for subclasses of human settlement (Q486972) takes 1.5 seconds (https://w.wiki/DtM). This is so small as to be essentially irrelevant to total query time. So my very strongly held view is for no change. Jheald (talk) 23:25, 14 December 2019 (UTC)
You seem to have misunderstood my proposal. I do not propose to get rid of the categories "hamlet", "village", "town", or "large city" for a settlement, what I propose is that every settlement should have at least P31:human settlement and not be "hidden" in a subclass of human settlement. Also, the query that times out is this:
1) It's not the only way to do that query and 2) having a homogenous P31 value does not necessarily solve the timeout problem - try ?item wdt:P31 Q5. Bottom line is, your proposal builds on your fragile grasp of SPARQL. That's no basis whatsoever for such a proposal. --Tagishsimon (talk) 13:03, 15 December 2019 (UTC)
Thanks for the information about P31 not being a solution, I got this idea from Andrew in the previous discussion. It seems you in multiple chats now have stated a judgement of my SPARQL-skills. I would rather prefer you used you energy constructively and provided an example query (or a way of querying) that does not time out instead, say a query that shows a map of the 15 biggest settlements of every country in the world. That would be useful and demonstrate your skills (=do good). To do this I will remind you that you will have to consider multiple P31:xx as not every large settlement yet have P31 city, large city or whatever you choose in your query e.g. Kyoto (Q34600) (missing city and/or big city), Kiruna (Q83232) (missing any city-like item), etc. I would also welcome a tutorial like video from you like Lucas recently published where you can show how to handle these very large taxonomies without timeouts (if you know how), alternatively a written guide would also be very welcome I'm sure as I'm most probably not the only one here new to SPARQL who experience problems when querying our cities/humans. Maybe we should create a Wikiproject for cities where it is detailed how to create a good city item (with e.g. P31:largest city, P31:port city for port cities, population best practices, maintenance queries, etc.)--So9q (talk) 14:08, 15 December 2019 (UTC)
It's very constructive to point out that people without expertise in SPARQL should make proposal based on arguments about what they believe will optimize SPARQL queries for which they lack expertise. If we would change Wikidata based on what people who know little about SPARQL believe will improve performance we would end up with a mess. ChristianKl ❪✉❫ 15:11, 15 December 2019 (UTC)
@So9q: Your requested query is difficult, for reasons that have essentially nothing to do with the class hierarchy we have for settlements. It would be just as difficult if you wanted to find eg the 15 oldest living people in Wikidata for each country, or the 15 largest 'big cities' (if that class were reliably populated). The class hierarchy is not what makes this query difficult for the sparql engine.
Here's a query which finds the (up to) 15 largest settlements for each country, that have a population of more than 2 million: https://w.wiki/DvL
The key difficulty with the request is that SPARQL has no fast built-in method to rank items in subgroups.
If you asked me just for the largest 1000 settlements in the world, that would be easy. We could ask SPARQL to ORDER the results by population (a process taking time roughly proportional to n log n, where n is the number of items in the solution set at that stage), and then LIMIT the returns to the first 1000 entries.
There is no corresponding method for ranking within subgroups. The query shows the best I know how to do, which is to do an n x n join on the solution set with itself, matching by country, and then to compute a rank by COUNTing how many items in the subgroup have an equal or larger populuation. This is an n x n process, so if you have a solution set of 5000 going in to this stage, that requires an operations over 25 million rows to complete the stage, which is right at the limit of what the SPARQL engine can manage in 60 seconds.
That, and not the class hierarchy structure, is what makes this requested query difficult; and is why I have had to limit your request to settlements over 2 million in population to get it to complete. Hope this helps. Jheald (talk) 16:25, 15 December 2019 (UTC)
Thanks a lot for the (friendly and) very informative reply. I understand now why my proposal does not win any ground and thats ok. :)--So9q (talk) 18:00, 15 December 2019 (UTC)
Oppose, yes it's messy but the reality is messy (like, in France, where it can really messy, officially there is not really cities, only communes - that can have one, or more, or less, "city" - but it doesn't fit well the concept of city, urban unit (Q1479822) probably being the closest fit). For me, quality here cannot be obtain by uniformisation but by conformation to the reality. Cheers, VIGNERON (talk) 17:31, 15 December 2019 (UTC)
Oppose I choose to oppose my own proposal because this brought about good arguments for keeping our structure and data model around cities. Thanks for your participation :)--So9q (talk) 18:00, 15 December 2019 (UTC)
Shame, I still consider distinguishing big city (Q1549591), based only on population size, to be silly, and there's really nothing to stop me creating an item for cities of size 336k (336000-336999) using that one as a precedent. Ghouston (talk) 22:49, 15 December 2019 (UTC)
I'll concede that this item and similar, although arbitrary, needs to exist because there are Wikipedia articles about it, based on some definition from a conference in 1887. However, not every item that exists needs to be used when modelling other items. However, given that everyone else disagrees with me, I'll have to live with this one. Ghouston (talk) 05:27, 16 December 2019 (UTC)
I don't think an item for "cities of size 336k" would be notable. As far as "big city" goes defining it that was is how the term is defined officially in some jurisdictions. ChristianKl ❪✉❫ 07:58, 18 December 2019 (UTC)
Ghouston I would definitely agree that while keeping the current hierarchy of urban-area types is reasonable, I would absolutely support a specific proposal to stop using big city (Q1549591) as a classification - by modern standards, many of these places are not particularly large, and by local definitions, many of them are not cities. It doesn't seem to be a very useful definition for the world at large. If it is a generally accepted classification in some countries, we can use it there, but no reason to use it everywhere.
If we do keep using it in instance of (P31), perhaps we should rename it to a substantially less confusing "settlement with a population over 100,000" or similar? Functional description for a functional classification. But calling Swindon (Q200942) a "big city" just feels deeply awkward. Andrew Gray (talk) 12:49, 18 December 2019 (UTC)
See unpatrolled edits
Hi all! Recently @Sakretsu: created on my request this gadget, User:Sakretsu/unpatrolled edits.js, which shows, when you visit an item, if the last edit/edits has/have not been patrolled. This is very useful in order to avoid the sedimentation of vandalisms, which can otherwise survive unnoticed for years (I've found many cases). I write here to advertise the gadget and to ask if you support, as I do, its addition to the Preferences. Bye, --Epìdosis10:51, 15 December 2019 (UTC)
Comment IIRC patrolers see unpatroled edits in the edit history. If some gadget can detect this for all logged-in users, maybe the ! should be shown for all users with auto-patrol rights. –84.46.52.19708:40, 17 December 2019 (UTC)
Changes needed to the Wikidata:Notability for the lexeme project to be able to succeed
Hi, today I experienced a vigilant administrator who deleted an item I created (without any discussion) to express a concept needed for a lexeme sense.
The editor cited Wikidata:Notability which to my surprise does not contain anything about items needed for the lexicographic project to succeed.
I suggest we include an additional notability criteria:
9. items needed to express a sense of a lexeme e.g. Q76822222 is considered notable as long as at least one lexeme link to them.
It would be helpful to have some more examples. Right now you point at an item you created, in which you aver that an alias for 'have' is 'must'. Must is not an alias of 'have'. It's very much less than clear what your Q76822222 item is and is not. It is no basis for an - is it your third or fourth or fifth? - attempt to legislate in an area in which you conspiculously fail to demonstrate any mastery of whatsoever. --Tagishsimon (talk) 22:53, 16 December 2019 (UTC)
Meanwhile - again demonstrating that you have not understood the territory in which you dabble - why exactly would we need your criteria 1.9, given criteria 3: "It fulfills a structural need"? --Tagishsimon (talk) 22:56, 16 December 2019 (UTC)
As Tagishsimon already said, we do have a structural needs provision. The problem is that your items for "have" are not needed to express concepts. If you take something like Q76656353, it would be superior to use an item like ownership (Q6498684). Because the concept can already be expressed that way, there's no need for Q76656353 and Q76656353 is going to be deleted.
Hi, thanks for the polite answer! I requested Q76656353 for deletion and improved on ownership (Q6498684) instead. The concept that was deleted was not important to me so I'm not going to discuss that particular one, but next time it happens I will cite the structural needs provision. :)--So9q (talk) 09:29, 17 December 2019 (UTC)
Adding "have" as alias isn't an improvement. The name for the concept in the English language is ownership. Subject item is not about a link to an item for the word of the sense but for an item about the underlying concept. ChristianKl ❪✉❫ 07:43, 18 December 2019 (UTC)
Correct way to model controversies and tie them to their subject
Was directed to the Wikipedia page for Donald Trump sexual misconduct allegations (Q27243341) recently and I was curious to see how this is modelled in Wikidata as it's not the kind of thing I've edited before. I have two immediate questions:
I'd make it a subclass of allegation (Q2839009), since it refers to more than a single accusation, hence they form their own subclass. I suppose the same item can represent both the accusations and the controversy, in which case it can also be an instance of controversy (Q1255828). It would be an instance it can all be wrapped up as a single controversy, although I suppose you could also say it's a class of controversies (perhaps some of the accusations may be considered controversies in their own right). Ghouston (talk) 01:45, 19 December 2019 (UTC)
merge error notification
I apologize for not seeing the entire message - I went to merge Q78364215 and Q5558202. I used the regular gadget found my user preferences. And a box popped up that said something like "merge started, now you can focus your browser", and then it went away. But the item never merged. The page never refreshed. In case something got stuck or delayed, I didn't want to close my window before it was done, so I opened a new window. I checked the history, the merge didn't happen. I clicked refresh on the new window, no merge. I clicked refresh on the old window, no merge.
I tried to merge a 2nd time to see if I would be able to read the message and see what happened, but the merge went through as normal.
I've not seen this message before. What did it say? Did I do something wrong? What do I do the next time it appears? Any answers to questions I did not ask think to ask? Quakewoody (talk) 12:18, 23 December 2019 (UTC)
No worries, they did merge. But if you open both items, you see that one is redirected to the other (and the other, obviously is not redirecting ;-). They end on exactly the same item. Like you would expect from a merge. Congratulations with your first succesful merge, thanks for your contributions and concern! Edoderoo (talk) 05:22, 24 December 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Nomen ad hoc (talk) 08:05, 24 December 2019 (UTC)
Difference between Undo and rollback
Hello,
what is the difference between Undo a Edit and Rollback a edit. I dont have Rollback rights but I can Undo a Edit as I have seen here a few minutes ago. I dont understand the difference. Can someone please tell me the difference. -- Hogü-456 (talk) 22:09, 13 December 2019 (UTC)
de:Hilfe:Wiederherstellen is pretty helpful; the only difference is that you get rollback rights at dewiki automatically after some time, which is not the case at Wikidata where you need to apply for it at Wikidata:Requests for permissions/Other rights#Rollbacker. In a nutshell: "undo" requires you to confirm your action, and you can (should) provide an edit summary that explains your action; "rollback" works without confirmation and without an edit summary, which makes it suitable for reverting obvious vandalism only (i.e. no explanation required). "undo" can be done by all users (even IPs), "rollback" requires the rollback right and it is practically only relevant for those who often deal with vandalism. —MisterSynergy (talk) 22:39, 13 December 2019 (UTC)
It is mostly the same idea. With an undo you can add a decent edit summary, with a rollback you can revert multiple edits by one vandal more easily, and anybody including IPs can get similar effects by editing a last known good version and click on "save" after adding their reason in the edit summary. Rollback is implicitly about vandalism, and if that's not the case a rollback can be considered as personal attack. You don't need the rollback right, unless you are very active in recent change patrol. –84.46.52.19707:59, 17 December 2019 (UTC)
As I understand it, "rollback" also automatically marks the reverted edits as "patrolled" (if they are from an IP or new user), which I don't think is true for "Undo". ArthurPSmith (talk) 19:37, 17 December 2019 (UTC)
That could be a side-effect of "rollback" rights combined with "patrol" rights; admittedly I'm curious if any rollbacker without patrol rights exists. –84.46.53.22800:17, 20 December 2019 (UTC)
In my real query, in place of the bind, I have some code which select some wikidata entities. The goal is to get a wikidata description of these entities
In most wikis SL Benfica only have one article both for multi-sports club and for the football team, as we can see if we see the articles from the two items. The articles from the two items are basically the same (direct translations, as we can see in the wiki pt and en, for example, one in one item and one in another). With this fact, since the data in the articles of both items are basically the same it makes no sence to have 2 items, so, in my opinion, all articles should be merged in the item about the multi-sports club (with the exception of the wikis that have 2 articles).
@JozeSlb: the items S.L. Benfica (Q131499) and S.L. Benfica (Q2895813) can't be merged as long as the Dutch Wikipedia has articles for both topics: the "multi-sport club" and the "football club". I don't know whether there is actually a separate football club, or at least some kind of internal division of the club dedicated to football. If so, S.L. Benfica (Q131499) will correspond to an external entity. If not, it's just for the structure of (Dutch) Wikipedia. There are also others, some of which are "clubs" like S.L. Benfica (Q3321835) and some are "teams" like S.L. Benfica (Q765125). Separate items for teams are logical. Ghouston (talk) 04:13, 20 December 2019 (UTC)
Relationship between a scholarly society and a publisher
How to model the relationship between a scholarly society and a publisher that publishes something (e.g. a journal, conference proceedings or a monograph series) on behalf of the society? We have publisher (P123) to map from the published item but nothing established for the "on behalf" part to express the relation to the society, although the society is often (also) indicated as the publisher. This question came up in the context of Coalition letter opposing lower embargoes (Q79108241) and a Twitter conversation around it. --Daniel Mietchen (talk) 03:31, 20 December 2019 (UTC)
Can Wikidata be extended to contain information other than encyclopedic one?
For example information about opening hours of a nearby shop. This information is too specific to be put in encyclopedia, however it is useful for local residents.
I'm fairly certain that we already have a property related to opening hours. You are welcome to nominate it for deletion if you think it doesn't belong here. --Trade (talk) 15:13, 20 December 2019 (UTC)
Can Wikidata be extended to be also a primary source of information? en:Primary source
I would like it to contain also information not published elsewhere. For example information about opening hours of a nearby shop that has no website. Because the shop has now website I cannot cite it. However I know its opening hours because they are written on its door.
You can enter data information here, even if en-wiki does not allow the information to be used on their wiki. The policies of en-wiki are not made on WikiData. Other wiki's might have an easier policy. Edoderoo (talk) 14:33, 20 December 2019 (UTC)
The problem with having items about individual shops like that on Wikidata is that there are many shops and when a user would search for any particular one, they are unlikely to find an item. When they find an item they don't know whether or not the opening hours are up to date and given that a lot of items aren't updated it's likely sooner or later out of date. ChristianKl ❪✉❫ 20:18, 20 December 2019 (UTC)
This feels like it would fit most appropriately on OpenStreetMap, which definitely does want this information, and has a suitable data-model for storing it. It's not something we generally record on Wikidata, even if we do have an item on an individual shop (which we usually don't). Andrew Gray (talk) 19:08, 21 December 2019 (UTC)
PagePile
Is there a way to get all wd-items for categories in one pagepile? With PetScan it ran into troubles, with PAWS the SparQL-query gives result, but it can not add that easily to a PagePile. At first I need the pile-id, but I would be interested how to make it myself. Thanks. Edoderoo (talk) 10:04, 21 December 2019 (UTC)
For me is the unification of similar sounding names suspicious, so what says the community to this? For me, Czech Tat'jana is a name on its own, similar for names in other latin scripts. --Florentyna (talk) 09:51, 17 December 2019 (UTC)
It's wrong and it's frustrating seeing the tedious work establishing transliterations separating the different spellings and script systems suddenly destroyed by a merge from "red editors". This is not the first time though, I think it happens because wikipedia editors want all name articles to be grouped. Which is a legitimate goal, but not necessarily how WD is/should be organized. Moebeus (talk) 10:09, 17 December 2019 (UTC)
Would it be possible to receive an notification whenever another user links to an item created by me? I'll really appreciate that. --Trade (talk) 01:29, 22 December 2019 (UTC)
It would be nice if you disable it for particular items. I have a few which are sporadically popular and will sometimes generate a few hundred notifications in one day. Ghouston (talk) 20:55, 22 December 2019 (UTC)
Actually, I think I'll just turn the feature off, since the vast majority of the notifications I get seem to be for the same boring items. Ghouston (talk) 21:02, 22 December 2019 (UTC)
I misunderstood, then. When I went to the page, it looked like what needed to be done was to fill in the images of the paintings. - Jmabel (talk) 16:41, 21 December 2019 (UTC)
@Animalparty: this week is the first time I heard about this model (P2634). In the context of a portrait with a single person on it, that person, the sitter, is the main subject (P921).
Would it be possible to exclude cases when the date of the painting does not match the years of life of the perspective sitter? I removed two cases by hand, but they are probably going to return with the next bot update.--Ymblanter (talk) 21:01, 21 December 2019 (UTC)
You could try fiddle around a bit with the SPARQL query? Probably easier to just create the items for the missing people in the paintings. Multichill (talk) 19:33, 23 December 2019 (UTC)
It seems like O2 (Q1759255) is falsely conflating the brand O2 and the company Telefónica Europe (at least going by the linked English Wikipedia page which is specifically for the brand). Would it be appropriate to split this item into two?
This company/brand seems particularly confusing to me with multiple subsidiaries all operating under the same brand name:
I'm not sure, but it sounds like Telefónica Europe no longer exists; I'm not sure a separate item for it would be very useful. If there is something to be said about it I guess - how it relates to the existing companies, the history, etc. then maybe it should be created. Otherwise this item probably should be just limited to the brand O2. ArthurPSmith (talk) 18:22, 19 December 2019 (UTC)
The item is for the company formerly called O2 plc. The English article was recently moved to "O2 (brand)", but is still about the company; other language articles are more clearly about the company, as are most statements in Q1759255 and all links to it. Peter James (talk) 11:44, 21 December 2019 (UTC)
The English Wikipedia article seems to be a bit of a mess, but my understanding is that the brand O2 is owned by a company Telefónica - a subsidiary of which (Telefónica Europe) happened to run all European branches and no longer does thanks to a restructure in 2014. This item seems to be conflating multiple concepts to me (the brand O2 which is used by multiple subsidiaries; the company O2 plc which was purchased by Telefónica; the company Telefónica Europe which is a subsidiary of Telefónica), but I'm not convinced I understand the situation enough to properly fix it. --SilentSpike (talk) 16:40, 21 December 2019 (UTC)
O2 plc was renamed to Telefonica Europe plc, and is now Telefonica O2 Holdings Ltd. The O2 brand (or at least its UK trademark) was transferred from O2 Holdings (a subsidiary of Telefonica Europe) to O2 Worldwide (a subsidiary of Telefónica, but not of Telefonica Europe) in 2015.(https://trademarks.ipo.gov.uk/ipo-tmcase/page/History/1/UK00002233188) So it is currently:
Do you think WD needs to track such company shuffles and cadrilles as were so eloquently described by @Peter James:? Corporate officers do such adjustments all the time for a variety of reasons, and I don't think these hold much public interest. So I think we should track the major holdings/conglomerates and their stable/significant/distinct subsidiaries (eg Space X is very distinct from The Boring Company), but don't need to care about the minutiae of their evolutions --Vladimir Alexiev (talk) 12:54, 23 December 2019 (UTC)
@Vladimir Alexiev: I'm not so concerned about capturing that detail, but since items and Wikipedia pages seem to exist for multiple regional subsidiaries it would be good to correctly capture their relation to the O2 brand. My particular interest is in having a stable item for the O2 brand itself which can be used by external projects - in this case, Name Suggestion Index (Q62108705). I'm not sure that Q1759255 (the item that has been used until now) fits that criteria since it's for Telefónica Europe specifically. --SilentSpike (talk) 16:39, 23 December 2019 (UTC)
Possibly not, it was just easier to find the information (that the former O2 plc was renamed and became a subsidiary, and that an O2 trademark was still under that subsidiary but isn't now). Peter James (talk) 12:39, 24 December 2019 (UTC)
If you never saw the original, and got your data from the archive, isn't it intellectually dishonest to claim the original you never saw as your reference? - Jmabel (talk) 17:26, 22 December 2019 (UTC)
So, similarly, if (for example) you were to see Richard Hofstadter (Q982698) quote James Madison (Q11813), it would be OK to cite the latter as the source? It's a while since I was in academia, but from what I remember of citation standards (and which I apply on WMF projects), it is not appropriate to cite a work you have not actually seen, except with a qualification of whose citation/quote of that source you are following. Are we simply trusting that archives are so reliable that we can consider ourselves to have effectively consulted the underlying source? Or what? - Jmabel (talk) 02:46, 23 December 2019 (UTC)
For exactly the same reason as, in my example, you would trust Hofstadter. But it is generally considered, in academia a least, as intellectually dishonest to attribute your information directly to a source you have not personally seen. In the above, you would definitely cite "…Madison as quoted in Hofstadter…", not simply "…Madison…". The difference is whether your actual source is Hofstadter or Madison. Similarly, it seems to me there is a difference between having something first-hand only from the archive, and that is your actual source, vs. having it from the original source, and quoting the archive only as a convenience link. While we generally trust the archive to be accurate, we don't necessarily trust it to be so perfect and complete as to be interchangeable with the original. For the most obvious example, the original may have illustrations that are missing from the archive. - Jmabel (talk) 16:20, 23 December 2019 (UTC)
Could you reword that last? I'm not sure I can make sense of it.
I believe that typically, the reason we end up with an archive url is that originally someone cited something from a particular (non-archive) page as it stood on a particular date, and the archive url is intended to back that up because (or in case) that information is no longer available at the original source. - Jmabel (talk) 22:04, 23 December 2019 (UTC)
I believe, and have said above, that by traditional academic standards, reference URL (P854) for the archive URL would be the most accurate. If we make a decision for purposes of convenience to deviate from that, I think the decision should be explicit. I could imagine a policy of using the archive URL with archive URL (P1065) and the dead URL with reference URL (P854)with some way to mark the fact that the page at the dead URL was never actually examined.
By the way, just today an issue came up on Commons that a particular archive is adding some snippets of their own that were not in the original of the pages they archived. - Jmabel (talk) 00:56, 25 December 2019 (UTC)
A quantity-based overview is available at wikiscan, which has tons of other numbers as well. Yet, item creation is to a far extent automated, thus the numbers do not have much meaning. —MisterSynergy (talk) 11:39, 24 December 2019 (UTC)
Tools like Quickstatement allows to use your regular account to do automated edits, without a bot flag. It’s a (very) common practice, I suspect far more used than regular bots, so hiding the bots is far not enough if you mean to exclude automated creations. author TomT0m / talk page12:27, 24 December 2019 (UTC)
At the moment the Greek media legend here does not match the Greek description on commons, the Greek legend here claims to be English, and the also existing English legend does not exist on commons. In other words, FUBAR, there must be only one description per language, and I'm not supposed to see Greek, because my ru-0 Cyril skills are not good enough to decode it, and my browser anyway does not indicate that I can sometimes decode Cyril. –84.46.52.8419:27, 24 December 2019 (UTC)
It looks like just a side-effect of the way it's implemented. It deletes the entire statement, including qualifiers, then adds a new one. Ghouston (talk) 00:00, 25 December 2019 (UTC)
Typically that would make sense, a deleted legend == description for, say, a non-free photo of a living person in Athens 2019 won't fit for an older free photo of the same living person in Thessaloniki 2011. –84.46.52.8402:57, 25 December 2019 (UTC)
These are just files that have been renamed, so it's the same image. I suppose it's possible that somebody used the wrong image in the wrong place because it has the wrong name, but sometimes it will be fine. Ghouston (talk) 03:28, 25 December 2019 (UTC)
collapsing long list of values
Is there any tool or gadget to collapse the long statements for big items? Scrolling up/down with mouse wheel is really inconvenient. – Kwj2772 (talk) 01:08, 25 December 2019 (UTC)
There are still several sites where categories would have to be merged. Q5611149 is "Foods", not "Food", or at least it is on English Wikipedia, which also has Q7134763 as a subcategory, but the distinction between them is unclear so I may propose merging them. Q5645580 contains these, but also includes drink, and food manufacturers and production I would expect Premier Foods (Q7240301), for example, to be in a subcategory of Q5645580, but not in any other category mentioned here (actually it is, but the categories don't apply to it; it isn't a pie brand - or used prominently as a brand name on any of its products - and it isn't particularly associated with Cornish cuisine). Peter James (talk) 13:36, 25 December 2019 (UTC)
OK, the amount of bad categories is 5 now (further growth possible). The comprehensive description of both Category:Cooking (Q6010682) and Category:Cuisine (Q9703849) is (in 17'000'000'000 dialects) "Wikimedia category". Many of them are shared between wikipedia and wiktionary, Category:Food products (Q7134763) is linked to only 2 wiktionaries, and Category:Cuisine (Q9703849) does NOT include any wiktionary. The simple problem on wiktionary is what lemmas have to be categorized where.
What is the scope ?
Does the category include only "hard" food (bread, bananas), or even drinks (beer, water) ?
Is only consumable material (food + drinks) included, or are even tools like spoon or colander, or even industrial manufacturing, ... ?
English Wiktionary is using "cooking" for both cooking and cuisine - definition 3.2 of cooking is synonymous with cuisine, but cooking typically means specific processes. So sashimi (Q190715) is part of Japanese cuisine, therefore in Q9703849, and is also a food product, and belongs in Q7134763, but not in Q6010682, whereas baking (Q720398) belongs in Q6010682, but not Q9703849 or Q7134763, and bread (Q7802) belongs in both Q6010682 and Q7134763 (or Q5611149), but not Q9703849. Q5645580 is a parent category of all of these categories and probably more not mentioned here. Peter James (talk) 16:18, 25 December 2019 (UTC)
Map with items
Not sure where (it might be a tweet by Wikipedia) I found a map with Wikidata items on it. I think the map was about Items near you that miss images. That is also how I found out that a railway station was at the wrong position on the map. Now I can't find the map anymore. I found Wikidata:Tools/Visualize data but the map I'm looking for isn't on the list. Any ideas? --D-Kuru (talk) 18:43, 25 December 2019 (UTC)
The first one seems to be for Wikipedia only. The second one is pretty close, but the map didn't require to enter the coordinates and radius and had coloured circles (I think red for no image, green for existing image, and blue for something else). the third one doesn't work for me so far --D-Kuru (talk) 20:26, 25 December 2019 (UTC)
In other words, my question here: Is the use of different from (P1889) reserved for use on items that look similar in English only or is the use of the property extended to any two items that look similar in any language? --Meno25 (talk) 09:48, 18 December 2019 (UTC)
I can't think of any reason to fail to add that property if there is confusion in any language that is at all widely used (which Arabic certainly is). Not sure we should do this if the confusion existed only in, say, Aranese, or Yiddish (and I say that as someone with some vocabulary in both of those). - Jmabel (talk) 16:13, 18 December 2019 (UTC)
@Infovarius: Thank you for the explanation. I understand that the two items are about different concepts (I am a native Arabic speaker by the way.) but the problem is that both the two items appear on the projectmerge list as candidates for merging. In such cases, I usually add different from (P1889) to them to remove them from the list. (I prefer to use different from (P1889) over the whitelist as usage of the property makes it easier to track such cases using different database queries.) So, in light of the comment by Jmabel, I can see that there is no harm in readding the property to the two items. I am going to wait for any final comment by you, and if there is not, I am going to readd the property. Thank you and keep up the good work, Infovarius. --Meno25 (talk) 17:39, 18 December 2019 (UTC)
Hi again, @Infovarius:. I don't think that this is necessary unless there is an exact match between site links such as the case described above. Best wishes to you. --Meno25 (talk) 11:06, 26 December 2019 (UTC)
I'd like to propose a new "componentSteps" property containing the steps of a certain exercise, dance move, or similar. For example, the componentSteps for the grapevine dance move would be "1. Side step, 2. Step behind the support foot, 3. Side step, 4. Step across support foot". Unfortunately, it seems like this would require translatable properties in Wikidata, which it doesn't seem to allow yet.
Is there a way to add a property like this to Wikidata? Are there plans for handling translatable properties like this in Wikidata?
@Mirek2: Could you explain what you mean by a "translatable property"? If you just mean being able to have a different name in different languages, all properties have that.
The "SPARQL (new)" links on property documentation boxes do not really work as they output only a small subset of actual violations. It relies on the "query constraint violations with WDQS" function that has not been completed yet. You can go with this simple query instead. —MisterSynergy (talk) 18:40, 25 December 2019 (UTC)
Hey I'm new to this stuff
And i have a question
How can i make page for popular person on Wikipedia ? – The preceding unsigned comment was added byBlitzawy (talk • contribs).
And I think it may be worth noting - at this stage of your editing experience, the person you are going to create an item/article for is not as popular as you think, otherwise they would already have an item/article. Quakewoody (talk) 22:29, 27 December 2019 (UTC)
The other Alexander Taylor died in 1867 [12]. I suppose we should create a new item for him if one doesn't already exist, since he has a memorial fountain and it would help avoid confusion. Ghouston (talk) 00:01, 29 December 2019 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Ghouston (talk) 06:56, 2 January 2020 (UTC)
Link UCUM unit symbols
Hi, I think it would be useful to link each item related with a unit of measurement with the unit symbols of the UCUM specification, which defines a set of codes to unambiguously represent all unit symbols currently used internationally in science, engineering, and business. It's focused on machine-to-machine communication, so it's a good candidate to be linked within wikidata :-)
Whether SI prefixes can be used with this unit or not (e.g. "km" means 1000 meters, but "kmin" cannot be used to indicate 1000 minutes)
Thefore, if each wikidata item about a unit of measurement is linked with the UCUM code, it would be possible to automatically extract/check from the UCUM specification some information like the definition factor for unit conversions and so on. Or detect missing items in Wikidata.
I was thinking that it would be enough to indicate the case-sensitive code in each Wikidata item. My doubt is the best way to do it properly. Currently it already exists the property unit symbol (P5061), so I tried in a couple of items (metre and cubic yard) to use this property to indicate the code, adding as a reference the UCUM item (with the stated in (P248) property). However, the unit symbol (P5061) property just accepts a monolingual text, and therefore a language code must be indicated. I saw that ISO 639 already dedicates a few special codes like "und" or "zxx", but these language codes are not accepted by Wikidata for monolingual text (which is understandable). So I just used "English", which is a bit problematic since in both cases it is already defined for English (with the same value in the first example, both as "m", and with a different value in the second, with "yd³"), and more over it does not convey that this is a special code.
Of course it would be possible to propose a new Wikidata property specifically for this ("UCUM code" or something like this), but any other idea that can just use existing properties? Thanks in advance for all your feeback! —surueña06:34, 21 December 2019 (UTC)
This query does the work. wdt: only considers best rank claims; thus, if there are any number of P31 statements with preferred rank, it does not see all normal rank P31 statements of that item. With p:P31/ps:P31 instead of wdt:P31 you get all results regardless of their ranks. —MisterSynergy (talk) 22:43, 22 December 2019 (UTC)
Thanks, did not notice that one statement had a different rank. And it's even the one which needs to be removed because of being redundant which is marked as preferred. Ahoerstemeier (talk) 23:00, 22 December 2019 (UTC)
Note there are parts of WD with deep concept trees where such removal would be vandalism (chemistry, medbio, molbio). One statement is always the tree top (or one of a few main nodes that are often searched), and the second statement is the most specific, as usual. Without this the query server would time out because of endless tree walking. --SCIdude (talk) 15:16, 28 December 2019 (UTC)
Should the OÜ entry be merged into the general PLC entry?
Osaühing is an Estonian kind of a private limited company. The Osaühing article seems to describe only Estonian PLC and not PLC in general, and the same time I could not find a separate article about PLC in general in Estonian Wikipedia. The following Q16917171 entry has some translations for exactly this kind of PLC, and does not contain any interwiki links. I'm not sure if it should be merged into the general Q18624259 entry.
There seem to be cases when state-specific kinds of PLC have their own Wikidata entry. The Q15829892 describes only GmbH, and refers to Wikipedia articles in many languages as well. But neither this entry nor the general PLC entry refer to the Deutsch Wikipedia: instead, there are several articles about GmbH in several German-speaking countries! Though, GmbH has historical significance for the whole phenomenon of PLC, while Osaühing is not. Bodqhrohro (talk) 23:57, 27 December 2019 (UTC)
Entered languages is not equal to translations of article?
Hello, I have a minor question that at first I wanted to ask on the chat but that doesn't connect! So, I hope it's OK to ask it here.
At CSV Apeldoorn (Q2140396) there is the language box that folds out when "All entered languages" is clicked. At any other item I've looked at, all the languages that item is translated into on WP are listed here. However in this instance, Hebrew is listed among the languages while there is no corresponding Hebrew language article of this club. The label is "No label defined", but there is a description though that Google Translate tells me means "A football team from the Netherlands". Why is Hebrew among the languages here? --Dutchy45 (talk) 13:19, 28 December 2019 (UTC)
On Wikidata there are basically four area's that are connected to languages:
Label, every item has one label, that is usually the article name (without the brackets)
Description, every item has one short description that explains what the items actually is
Aliases, alternative labels, helpful when searching for wikidata items
sitelinks, a link to an article on a language wikipedia, or to wikinews, wikibooks, etc
I'm only blocked on nl-wiki. Just two more days to sit on my hands. There is more then enough that can be done, no worries, my time is filled. Edoderoo (talk) 15:59, 28 December 2019 (UTC)
Wikidata dump files
Hi everyone! I have some questions about the Wikidata dumps. I'm importing the JSON dumps into a database, but the dumps do not have redirect information. Because of this, some relationships point to non-existent entities, which is causing data consistency issues. I found Phabricator tasks mentioning that the JSON dumps don't include redirect info, but there is not any recent activity on them. I'm ignoring the missing entities for now, but I need to find a longer-term solution that takes redirects into account.
To work around the issue, I tried parsing the owl:sameAs predicates out of the RDF files and creating an entity for each subject. That created some duplicate entities, which leads me to believe that some of the subject entities in the list of redirects exist in the JSON dump and some do not. This makes sense if the dump files are not snapshots in time, or if they are but the JSON and RDF dumps are run at different times.
To try to understand this better, can someone answer the following for me:
Within a single dump file (JSON and/or RDF), is every entity that appears in a relationship guaranteed to be defined in the file? In other words, does the dump file represent a self-consistent snapshot of the database at a moment in time?
If the above is true, then is it also true across the JSON and RDF files in a given dump directory?
My understanding is that a redirect is created when duplicate entities are discovered in the database, at which point the duplicate entities are redirected to one canonical entity. This is done so that links that point to the duplicate entities will automatically resolve to the canonical entity without having to update the links (e.g. in relationships). Is this understanding correct?
If the above is true, is it also true that, once an entity is changed to a redirect entity, it will never be changed back?
Should I be using the RDF dump instead of the JSON dump? I'd prefer JSON because of the availability of standard parsing tools, and because the one-entity-per-line format makes parsing much easier.
If there is a more appropriate place for me to ask these questions, or if you need more information from me, please let me know. Thanks in advance for the help! --Mventimi (talk) 19:03, 28 December 2019 (UTC)
There's a bot that resolves statement links to redirects to the target item, so maybe you'll only find them for items that have been redirected recently. An option would be to query the live database for the redirect target. You can't say for sure that "once an entity is changed to a redirect entity, it will never be changed back", since redirects can be reverted if somebody thinks they were a mistake. Ghouston (talk) 23:18, 28 December 2019 (UTC)
Do you know how often that bot runs? I'm finding a good amount of redirects still present in statement links. I've downloaded the 12/9 and 12/23 dumps, so I'll go through them both to quantify "a good amount" and to determine whether the bot cleaned up the redirects between the 12/9 and 12/23 dumps.
What you said about reversions makes sense. I was hoping that I could come up with a hard rule to resolve the case where an entity shows up as both a defined item in the JSON dump and a redirect in the RDF dump. However, reversions mean that I can't really know which of those take precedence.
Do you have an idea about the dump consistency questions? Looking through Phabricator, I found some statements in T144103 that imply that a single dump file is consistent, but that different files are not. However, T128876 implies that even a single file is not consistent. --Mventimi (talk) 00:40, 29 December 2019 (UTC)
The redirects are fixed by KrBot. I merged one item at 2019-12-19T10:17:49, and the bot fixed up a reference at 2019-12-22T00:27:44, less than 3 days later. I don't know anything about the dumps, sorry. Ghouston (talk) 01:28, 29 December 2019 (UTC)
Biblical categories changeup at Commons
Commons has recently renamed all categories pertaining to persons named in the Bible. The category moves should be reflected in the interwiki links, however all of the links from Commons category (P373) will need to be updated. Example: Adam (Q70899) has Commons category (P373) linked to the old Category:Adam, but the interwiki now points to the new commons:Category:Adam (Biblical figure).
Note: many of these categories still bear discussion labels, so it's possible the category renames will be challenged and the whole process begin again, but the linked discussion seems to have ended in October, and the moves have just been undertaken this week. --EncycloPetey (talk) 21:45, 28 December 2019 (UTC)
Wikidata is increasingly popular and consequently it becomes increasingly problematic to contribute to Wikidata. What I do at the moment is work on science and scientists about/from Africa. It is a subject that has a huge gap compared to general science. It is not possible to do maintenance for imho no good reasons. Quality is bandied as a reason to deny the use of SourceMD while at the same time a known quality deficient, bias generating project is accepted.
My request is to accept the use of SourceMD and thereby enable the work on science relevant to Africa. Be clear, more lives are saved by improved neo-natal and maternal care than cancer care. We do not find that in the numbers of Wikidata and it is an indicator of the bias we propagate. Thanks, GerardM (talk) 08:40, 26 December 2019 (UTC)
WD:SourceMD is blocked, but the "Too many requests" notification is due to bot traffic to the WD:Author Disambiguator, as discussed here. I agree that both issues make it hard to contribute in the areas (supposed to be) served by these tools, and we do not currently have good solutions to address them. --Daniel Mietchen (talk) 07:53, 30 December 2019 (UTC)
I should probably do some consulting on Wikitech/WMF cloud services about this, but if anybody here knows more about things like (A) blocking web crawlers that ignore META (noindex, nofollow) tags, (B) getting more resources for tools at wmflabs (currently the author disambig service seems to be limited to 4 PHP fastcgi processes), or (C) speeding up certain SPARQL queries - please post a note on my talk page or suggest something on github (the link Daniel Mietchen posted will get you there) - thanks! ArthurPSmith (talk) 20:32, 30 December 2019 (UTC)
The revised version of the very powerful PetScan offers to edit or create items based on Wikipedia categories, search, SPARQL, links with filters by label or properties
Other Noteworthy Stuff
New tool: Dwynwen, adapted from Crotos, provides users with a search interface designed specifically for artworks and other visual digital content from Commons, using Wikidata.
"History" and "watch/unwatch" tabs are auto-hiding shortly (in ~0.5sec) after item loading for me. It is very inconvenient to live without them. I know at least one more user with the same problem. Can anybody help us? --Infovarius (talk) 21:39, 24 December 2019 (UTC)
Even turning off all gadgets and cleaning all my common.js doesn't help - these two tabs simply disappears. What's happening? Does anybody experiences this too? Browsers Opera and Firefox. --Infovarius (talk) 13:14, 25 December 2019 (UTC)
The problem is in Monobook skin probably - in Vector tabs are all present (though sometimes floating somewhere in wrong places). --Infovarius (talk) 23:28, 30 December 2019 (UTC)
Property part of (P361) is used to link an instance of an individual object to an instance of a group that contains the first object. It is not used to link to a class. The inverse of the part of (P361) property is has part(s) (P527). So, for example, extending on musical group (Q215380) immediately above:
Sports teams are something similar, and for some reason have their own subproperty member of sports team (P54). There are also some quite large musical groups like symphony orchestras. So should Help:Modelling/General be changed in this instance? Ghouston (talk) 23:37, 14 December 2019 (UTC)
I share this frustration, I wish we could finally have a "band members" property. "member" to me implies a membership, and naturally goes with member count and members have occupation. It's not a natural fit for the looser structure of bands. And "part of" is way, way to broad. I've said it before, but Project Music has a lot to learn from the sports guys, they know what they want and then they get it done (member of sports team must have been one of the first properties ever created?). Moebeus (talk) 00:55, 15 December 2019 (UTC)
I don't have any problem with member of (P463) being used as a generic membership property. The kind of thing that it's a member of should already be apparent from the subject item, so it doesn't seem useful to me to have a different property for each case. member of sports team (P54) was indeed a very early property, and pre-dated the generic item. Ghouston (talk) 08:04, 15 December 2019 (UTC)
I would like to see the inverse property - member of band, and band member. On the band page you say someone is a member, and on the person page you say he is a member of. Quakewoody (talk)
I guess that's why people prefer using "part of" and "has part". Ideally, inverse statements shouldn't be created, because they are redundant. But it's a limitation of the Wikidata software that it doesn't treat statements symmetrically. Ghouston (talk) 22:35, 15 December 2019 (UTC)
There are no plans as far as I know to change how the software works. So the question is whether the inverses are indispensable. Are they used on templates in Wikipedia, with no way to rewrite such templates? If the inverses are needed, then an inverse membership property could be created so that "part of" doesn't have to be used. Ghouston (talk) 06:20, 16 December 2019 (UTC)
E.g., at es:The_Beatles there's a template that lists the 4 main members. If you delete the members in the template and preview the result, it will display the 8 members from Wikidata instead. I notice that the 4 main members have been given preferred rank in Wikidata, which I'm not sure is for any good reason supported by Help:Ranking, and in any case is ignored by this particular template. Ghouston (talk) 06:29, 16 December 2019 (UTC)
Propose either a generic "member" property, as an inverse to member of (P463), or alternatively, a new band member / member of band pair. I think the generic property would be sufficient, since the band context will be obvious, and part of (P361) / has part(s) (P527) are already doing the job despite being generic. If such property is approved:
1/1 would support. For what it's worth, I think it's preferable to have a "member" property on the band, rather than a "member of" on the musician. After all, a band is nothing more than the sum of it's members, while Ringo Starr simply rulez; with, or without, the Beatles ;-) Moebeus (talk) 22:37, 22 December 2019 (UTC)
But what is the notable entry - the collective musicians, or the individual musician. For example, I never would have created Q66424356 if he wasn't a member of Q458561. While the Coral Reefer band wouldn't meet notability if it wasn't for Jimmy Buffett. Which is why I think the inverse would also be necessary. Quakewoody (talk) 01:15, 23 December 2019 (UTC)
I don't think it really matters. If you only need one item to hold all the information, then just create one item. On proposing an inverse "member" property, I'd suggest that this would only be auxiliary to member of (P463), and only used where it's likely to be needed by templates. It would be silly to enter all the members of National Academy of Sciences (Q270794) onto that item. The member property would have a constraint that the inverse "member of" exists, but the reverse would not apply. Ghouston (talk) 08:58, 26 December 2019 (UTC)
And I understand everything (or at least most of) what everyone is saying. But I can't help but see a situation where 'Band X' has 4 members (which is typical of the modern rock band). We create the band, and create the members, we link the members of the band to the band but we don't link the band to the members. The members could then be deleted because nothing points to them. And then, once all the members are deleted, nothing is pointing to the band. Which is why I think member of band and band member would be needed. Create the two way relationship. Quakewoody (talk) 14:12, 26 December 2019 (UTC)
I still don't see why "member of band" / "band member" would be better than a generic "member of" / "member" (which may also be useful in other situations.) Ghouston (talk) 00:12, 27 December 2019 (UTC)
I suppose with "member of band" you could consider making the inverses mandatory in both directions. However, I'm not sure that that would be a good idea. There may be some very large and long-lived bands, like symphony orchestras, that have had numerous notable members and you neither want to list them all on the item nor include them in infoboxes. Ghouston (talk) 00:17, 27 December 2019 (UTC)
I haven't really considered the practicality of an alternative solution, which is to use make a Lua module that can retrieve the members when processing a page for a band/other organization. I think the Commons Infobox does this kind of thing all the time. Ghouston (talk) 22:33, 27 December 2019 (UTC)
I believe this feature requires making additional queries to Wikidata, beyond what's available to a Wikipedia page from its linked Wikidata item by default (because the API is not symmetrical on the left and right side of statements). That's why a Lua module (or Javascript in the user interface) is needed. Ghouston (talk) 04:48, 28 December 2019 (UTC)
Yeah, it uses the a module for Lua in the Wikibase client. Unfortunately, this only allows querying from items from the left side of statements, it doesn't work if the items you want appear on the right. That can only be done as far as I know using the query service, which isn't used from Lua in Wikipedias. I think it would come under "automatic list generation" at phab:T67626. Ghouston (talk) 06:03, 31 December 2019 (UTC)
The notability question seems artificial, to me. The relevant line in Wikidata:Notability that would permit band members to be included is "It fulfills a structural need, for example: it is needed to make statements made in other items more useful." If it was accepted that listing the members of a band was a "structural need", then it doesn't seem to matter exactly how it's expressed in statements. I'm not sure that there is such a structural need: some bands are very large and some members may stay for very short periods. Ghouston (talk) 04:31, 31 December 2019 (UTC)
If you have an item A that's notable and an item B that wouldn't be notable without the structural provision then having a statement on A that references B would infer notability to B. An statement on B however doesn't create notability for B. It's not making statements on items other then B more valuable. That said band membership should usually be able to be demonstrated through sources and thus there should be material for the notability of band members. ChristianKl ❪✉❫ 12:52, 13 January 2020 (UTC)
It would be the same "structural need" that would say we need to add items for all of the residents of a particular city, since the city is notable for Wikidata, and we should be able to construct a list of residents since it's information about the city. However, we collect information only about notable residents of the city, and perhaps only about notable members of a band. Ghouston (talk) 22:38, 31 December 2019 (UTC)
I disagree with that last. It's interesting to know that two notable bands had (for example) the same drummer, or possibly even to be able to trace that one orchestra has functioned as something of a "farm team" for another (I'm less certain that we care about non-notable players in something as large as an orchestra, though we certainly do have a very low threshold for analogous athletes to be considered notable). It's not particular interesting to know that there are many people who have lived in both New York City and Los Angeles. - Jmabel (talk) 23:56, 1 January 2020 (UTC)
Support Yes, please, references should be MUCH easier to add in Wikidata, currently it is painful. Why does the same reference have to be added on every data item, for instance, rather than having it once on the page with a reference to it, the way it works in the regular wikipedias (yes, I know there's a tool to make this easier, but it's still annoying). And I frequently default to just a "reference URL" link when it should maybe be something better structured. ArthurPSmith (talk) 19:14, 9 December 2019 (UTC)
Done, but such requests are handled on Wikidata:Deletion requests. --Esteban16 (talk) 21:12, 16 December 2019 (UTC)
CLDR's and ICU's Mark Davis et al. definitely cook water like everybody else for their coffee or tea. –84.46.52.197 07:44, 17 December 2019 (UTC)
Comment It seems that the lookahead search is really slow to update. I moved s:Author:John Cleland to s:Author:John Cleland (1709-1789) and even after a couple of minutes the new target is not found in the lookahead. It does show when I have an exact title for it, but until I have a full title ... nada. — billinghurst sDrewth 06:44, 16 December 2019 (UTC)
Oppose It seems like the proposal doesn't understand what subclasses are. Items that are p31city (Q515) are already currently human settlement (Q486972). All queries you could write afterwards are already supported. ChristianKl ❪✉❫ 10:12, 14 December 2019 (UTC)
Strong oppose It's a lot more useful for a Commons infobox to be able to say "hamlet", "village", "town", or "large city" (translated into Gaelic or Welsh or Breton or whatever) to explain what sort of place a category is for, than just "human settlement". As for the point on querying, a query for subclasses of human settlement (Q486972) takes 1.5 seconds (https://w.wiki/DtM). This is so small as to be essentially irrelevant to total query time. So my very strongly held view is for no change. Jheald (talk) 23:25, 14 December 2019 (UTC)
Strong support —Eihel (talk) 00:24, 18 December 2019 (UTC)
–84.46.53.228 00:17, 20 December 2019 (UTC)
⟩ instance of (P31) 
Notified participants of WikiProject Companies
.jpg&diff=prev&oldid=379354756){kind=link}
is used to link an instance of an individual object to an instance of a group that contains the first object. It is not used to link to a class. The inverse of the part of (P361) property is has part(s) (P527)