Wikidata:Project chat/Archive/2020/10

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

Advice on instance of (P31) and subclass of (P279)

Hi. Cdo256 has been tidying up on instance of (P31) and subclass of (P279) but has inadvertently been creating constraint violations. I have started a discussion at User talk:Cdo256 but some of the points raised are outside of my experience. I don't have a solid grasp on when instance of (P31) and subclass of (P279) are the correct option, so it may be the user that is correct and it is the property constraint (P2302) that is set wrong. Can someone please offer further advice to the user? Thanks. From Hill To Shore (talk) 09:40, 26 September 2020 (UTC)

My feeling is that some of the constraints like the value-type constraint (Q21510865) on architectural style (P149), brand (P1716), represents (P1268), and occupation (P106) should be changed to allow both subclasses and instances rather than just instances. Since Xinstance of (P31)profession (Q28640) implies that X is an indivisible profession with no subclasses however many examples exist where instance of appears to be used the wrong way, for example architect (Q42973) is treated both as an instance and a class (for example urban architect (Q2860261)subclass of (P279)architect (Q42973), making architect a class), which as Wikidata:WikiProject_Ontology/Problems states, this is an anti-pattern. In the case of occupation (P106), I think that jobs should prefer to be subclasses rather than instances since a job could be subclassed right down to a specific opening/employee at a specific company. This I think also holds for other properties things like color, shape and texture where it's difficult (or impossible) to define what a specific indivisible shape and color a real world object has. - cdo256 10:41, 26 September 2020 (UTC)

@cdo256: architect (Q42973) is an instance of occupation (Q12737077) but a class of particular jobs. Because architect (Q42973) is a class (by nature of it being an instance of the metaclass occupation (Q12737077)), it's OK to have subclasses, such as urban architect (Q2860261)subclass of (P279)architect (Q42973). On Wikipedia, people are not instances of architect (Q42973) because "occupation of" is the proper way to say that a particular person is an architect, but it is fine to say Chief Architect of Moscow (Q84308954)instance of (P31)architect (Q42973) [1]. — The Erinaceous One 🦔 04:09, 28 September 2020 (UTC)

@The-erinaceous-one: Isn't the correct relationship Chief Architect of Moscow (Q84308954)subclass of (P279)architect (Q42973) rather than Chief Architect of Moscow (Q84308954)instance of (P31)architect (Q42973)? --Oravrattas (talk) 08:10, 28 September 2020 (UTC)

@Oravrattas: I'm not convinced either way. Can you explain your reasoning? If we view Chief Architect of Moscow (Q84308954) as the class of people who hold that position, than "subclass of" makes sense, but if we instead view it as a singular instance of an architect position, then it should instead be "instance of." The description of Q12737077 is worth noting: occupation (Q12737077): label applied to a person based on an activity they participate in (based on Google translate, this matches the description in most languages), so instances of Q12737077 should be labels. Another possibility is that Chief Architect of Moscow (Q84308954) is neither a subclass nor instance of architect, but is just an instance of position (Q4164871) with a qualifier to connect it to city architect (Q53236130).

(Note that if we choose to use subclass of (P279), then there should be no instances of architect (Q42973)---or any other occupation, for that matter.) — The Erinaceous One 🦔 10:53, 28 September 2020 (UTC)

@The-erinaceous-one: I see this as being analogous to positions like Mayor of Moscow (Q1837906), and it's fairly well established how we model those. --Oravrattas (talk) 12:04, 28 September 2020 (UTC)

If that's the way it is already modeled then it sounds fine to me. — The Erinaceous One 🦔 08:56, 1 October 2020 (UTC)

@Oravrattas, The-erinaceous-one: if that's the case then the advice on WikiData:WikiProject_Ontology/Problems needs to be completely changed and Help:Basic_Membership_Properties needs to be amended.  – The preceding unsigned comment was added by Cdo256 (talk • contribs).

User:MPF capitalizing bird names

User:MPF has been changing the English labels of numerous bird species to use Title Case instead of lowercase. According to Help:Label#Capitalization, things which aren't proper nouns should have lowercase labels; this includes birds. MPF contends that many ornithology databases use Title Case, and has changed the labels of numerous bird species as such.

I originally encountered this on Blue Jay (Q199758), where myself and others have restored the lowercase label multiple times after MPF capitalized it. I started a talk page discussion there a while back, but it stagnated so I'm bringing this topic here. I'd like to establish community consensus on how birds should be labeled so we can have some stability here. Thanks, IagoQnsi (talk) 17:48, 29 September 2020 (UTC)

Continuing the quote from where you left off, "In the rare case that something intentionally breaks capitalization rules, the capitalization on Wikidata should reflect this, and not try and correct it". And if you scroll down to the Vernacular names and look at the references, you'll see they are capitalised by the primary references (IOC et al.). The overwhelming standard among bird authorities is to capitalise bird names (and ditto, most other animals, plants, etc.). Consider them as proper nouns of particular species. BTW, I am far from the only person doing so. - MPF (talk) 18:00, 29 September 2020 (UTC)

@IagoQnsi, MPF: please don't focus only on birds. If other groups (animals, plants) in Wikidata are written in lowercase (because of not being proper nouns), then birds also must be written in lowercase.--Estopedist1 (talk) 05:42, 30 September 2020 (UTC)

I wasn't; as mentioned "and ditto, most other animals, plants, etc." - capitalisation is the general rule; they are considered proper nouns. It is with good reason too; most rats are brown, so a brown rat can be any of numerous species, but a Brown Rat is specifically Rattus norvegicus. Similarly, various moth larvae eat various forms of wool-based fabrics, but Tapestry Moth is specifically Trichophaga tapetzella, Common Clothes Moth is Tineola bisselliella, and so on. Capitalisation of specific English names has a long history; see e.g. Miller's 1768 Gardener's Dictionary; what I find strange is this unreasoned hostility to it being expressed by some quarters on some wiki projects. - MPF (talk) 10:04, 30 September 2020 (UTC)

@IagoQnsi, MPF: well, in my mother language (Estonian language), vernacular names are definitely not proper nouns, but I am not English expert. Maybe some explanation can be found here: en:Proper_noun_and_common_noun#cite_note-24--Estopedist1 (talk) 06:15, 1 October 2020 (UTC)

Should these be merged or one be the building and the other be the defunct organization that used to own the building?

Should these be merged? We have University Medical Center of Princeton at Plainsboro (Q7894826) and we have Penn Medicine Princeton Medical Center (Q89034988). University Medical Center of Princeton at Plainsboro (Q7894826) became Penn Medicine Princeton Medical Center (Q89034988) in the same building when a larger organization bought out the hospital. Do we need a separate entry for the defunct organization? Is a hospital an organization or a building? --RAN (talk) 23:51, 1 October 2020 (UTC)

My opinion is, separate items. Not only do you need to distinguish between organization and building (an event happened in the building, a policy was adopted by the organization) but it is often necessary to specify which exact organization -- pre- or post-acquisition -- did some activity. There are plenty of properties like followed by (P156) and merged into (P7888) to link them. — Levana Taylor (talk) 01:43, 2 October 2020 (UTC)

I changed the descriptions to distinguish them, and the instances_of. --RAN (talk) 03:07, 2 October 2020 (UTC)

• If the operator changes, just add a new statement. --- Jura 04:53, 2 October 2020 (UTC)

Problem with Wikidata:Project_chat/Archive/2020/08

Somehow this edit makes that this section isn't available on the next revision of the archive .. I don't quite know how. Possibly some odd formatting in the post that is being archived. Any idea what causes it?

I left a note for operator of the bot that happened to archive it (@Revibot:, but I don't think it's problem of the bot as such. --- Jura 05:17, 2 October 2020 (UTC)

There was a {{P}} with one of the closing brackets missing.[2] When a section with JSON code was archived, the brackets at the end closed the template. Peter James (talk) 14:48, 2 October 2020 (UTC)

Problem with Q1716419

It is attached to https://commons.wikimedia.org/wiki/Category:Lathe_operators -- but it explicitly says that it's "Different from: lathe operators". But I can't figure out how to detach this item from that Commons category. Can someone please do that? Thanks. -- Wesha (talk) 07:46, 2 October 2020 (UTC)

• Somehow the infobox on c:Category:Lathe_operators seems to come from turner (Q1716419), not lathe operator (Q28933489). Maybe purging will eventually fix it. --- Jura 07:55, 2 October 2020 (UTC)

• If a Commons category is moved to a different item, it needs a null edit and then a purge to fix the Infobox. Who knows why. Ghouston (talk) 09:19, 2 October 2020 (UTC)
  • Oddly, even in preview it's not working .. --- Jura 09:27, 2 October 2020 (UTC)
    • It now shows the infobox for Q28933489. Peter James (talk) 14:53, 2 October 2020 (UTC)

"years old"

Could anyone elucidate why there is an item years old (Q24564698) and why one should use it over year (Q577)? It seems to me that "years old" is not a unit but a strange concotation of property and unit. It also causes quite a few constraint violations, such as a few hundred on age estimated by a dating method (P7584) alone.

I can see how it might make sense to people in the context of countries, i. e. Portugal (Q45)age of majority (P2997)18 years old. It's most often used in qualifiers, such Spiros Kyprianou (Q552389)candidacy in election (P3602)1978 Cypriot presidential election (Q4502497)age of subject at event (P3629)46 years old, where I would argue the phrasing as read already feels odd.

And while we're on the subject, is there a reason for the margin of error in years old (Q24564698)conversion to standard unit (P2442)1±0.1 years? Is that why years old (Q24564698)instance of (P31)unit without standard conversion to SI (Q21684377)? --Matthias Winkelmann (talk) 17:16, 29 September 2020 (UTC)

The phrase "x is 3 years old" can be rephrased as "age of x = 3 years". The unit is year, the quantity is age. Therefore, Q24564698 should be eliminated (or merged into the appropriate year item). Toni 001 (talk) 09:40, 30 September 2020 (UTC)

I agree with Toni 001, Q24564698 can be merged into Q577. Ainali (talk) 21:02, 2 October 2020 (UTC)

The phrase years old (Q24564698) have different meaning of use than year (Q577). 'Years old' is more specially used for showing age of a human or organism. This entity is used in Infobox person template in Catalan Wikipedia for showing the age of a person. ie it will show '5 years old' right now. If these two items are merged the template will show just '5 year'. In different language the meaning of these two are different and uses different terms for both. In Malayalam years old and year have different meaning.-❙❚❚❙❙ GnOeee ❚❙❚❙❙ ✉ 04:14, 3 October 2020 (UTC)

That sounds like a presentation issue that could and should be solved by the lua module importing the value. It still just stores a number of time units, which in this case is the same. Ainali (talk) 12:46, 3 October 2020 (UTC)

• Some age-recording systems such as in South Korea don't use the actual birth date, although I'm not sure if it's relevant here. See [3]. Ghouston (talk) 22:50, 3 October 2020 (UTC)

Problem with QuickStatements V1

I always used QuickStatements V1 to create items in the past. When I use it today, it does not response properly, so I have to switch and learn how to use QuickStatements V2.

However, I did not know QuickStatements V1 is actually malfunctioning. Although it does not response properly, it still keeps creating blank items in the background, and I realize that after about half an hour. I have tried my best to reuse some of the blank items, but there are still around 400 blank items unused (between Q99899213 and Q99900750). Fell free to reuse these blank items. --Myerbee (talk) 16:26, 3 October 2020 (UTC) (Updated: All blank items have been used or deleted. --Myerbee (talk) 02:58, 4 October 2020 (UTC))

Goodreads subscriber counts?

I noticed this edit on Euripides (Q48305).

Is the data that is being added suitable for Wikidata, or is it site advertising? Is it relevant for the person whose data item is carrying this information? Nothing in the property proposal or description indicates that these kinds of data would be imported.

So, is the number of subscribers to a millenia-long-dead author on a website relevant data for that author's data item?

And is this simply data, or is it advertising for the author's page on the website? --EncycloPetey (talk) 02:10, 30 September 2020 (UTC)

Wikidata has no standards. Only data matters. If it's data, it's fair game. Resistance is futile. Trivia is the only truth. -Animalparty (talk) 02:56, 30 September 2020 (UTC)

@EncycloPetey: I was importing data for all goodreads authors. There is value in being able to tie authors items here to that database as it has further metadata. Pulling in subscriber counts in addition to everything else was easy so I did it. We already often do the same thing for e.g. youtube and twitter accounts. I cannot imagine how you would think this is advertising. I have no connection to goodreads and am planning to do the same thing again. BrokenSegue (talk) 03:09, 30 September 2020 (UTC)

I share your doubts. I am not sure about the relevancy of these data, mainly for old authors from a non-english area with actually tiny numbers (example), also in a case of popular items too frequent updates may cover vandalism easily. What are the plans for the frequency of updates? --Jklamo (talk) 12:16, 30 September 2020 (UTC)

@Jklamo: How will this cover for vandalism? The edits are done on a bot account. I was thinking annual updates but really had no concrete plans. I don't see why I would import this data for new authors but not all authors given it's literally easier to do the former. Maybe you all misunderstand what "following" someone on goodreads means. It just means that person likes their works. The fact someone is dead or ancient is irrelevant. BrokenSegue (talk) 12:46, 30 September 2020 (UTC)

@BrokenSegue: Frequent update edits may "hide" vandalism edits, as people see only the last edit in their watchlist. But annual updates are OK from this point of view.

I understand the following concept, but still not sure about the relevancy of a small number of followers for old authors from a non-English area.--Jklamo (talk) 13:20, 4 October 2020 (UTC)

From Wikidata how do I view what is linked to Commons by "depicts"?

From Wikidata how do I view what is linked to Commons by "depicts"? Can I click on the equivalent of what_links_here and see what what is all the way over in Commons? I had a few Wikidata entries deleted that were sparsely filled in, but were created because the entry depicted a person or thing at Commons. I want to know if someone deletes here are they aware of what is linked at Commons. --RAN (talk) 01:42, 4 October 2020 (UTC)

Wikidata item use in structured data at Commons (SDC) does not show up in any Special:WhatLinksHere or Special:EntityUsage pages, neither here nor at Commons. Since July this year there is the Wikimedia Commons Query Service (WCQS) which has this data available. However, most admins probably do not look there for item usage. Simply ask for undeletion if something went missing. —MisterSynergy (talk) 07:44, 4 October 2020 (UTC)

Q68258272

Can someone familiar with recurring_sporting_events/sports _season look at 1931 National Air Races (Q68258272) that I created. It is further broken down into the two events that make up the races in each year. I want to make sure it is modeled properly before I do the other years. --RAN (talk) 13:59, 4 October 2020 (UTC)

Three different image-related text tasks in structured data in Commons

I was looking at an image that had a visible caption in it on Commons and wanted to add that text as structured data. So I tried adding a media legend (P2096) statement, but got a warning that the property should only be used as a qualifier. Then I tried adding depicts: text (Q234460) instead with a media legend qualifier containing the text in the image, but got a warning that P2096 isn't included in the allowed qualifiers constraint (Q21510851) of depicts (P180) even though media legend says that the "subject item of this property" is "depiction".

But thinking about it, for most images on Commons the caption would have been cropped out and would be purely metadata, so in those instances attaching it through depicts:text would not seem to be appropriate.

Then a third case I'm considering is text that's in an image but is not a caption, rather is part of the image. For images of certain objects there's inscription (P1684), for text inscribed on an object being depicted, but I'm not seeing a property that would correspond to text within the image itself, or text on a road sign or printed label or something like that.

So how would I go about doing each of these three things? Or should I just not do one or more of them? Thanks, --Struthious Bandersnatch (talk) 11:47, 4 October 2020 (UTC)

• Constraints for properties at Wikidata mainly reflect how they should be used on Wikidata itself. This may be useful for Commons, but in some cases it may not. Eventually, divergent requirements should be modeled with different property constraints, see Wikidata:Property_proposal/property_constraint_for_Commons.

How existing Wikidata properties should be used would generally be discussed on Commons, e.g. at c:Commons talk:Structured data/Modeling and subpages.

On a side note, there is also identified in image by (P7380) for some aspects of text in images. --- Jura 11:56, 4 October 2020 (UTC)

Thanks Jura! I've copied my question into the Commons Modeling discussion. Hopefully someone can point me in the right direction there, or we can start making any decisions that need to be made. --Struthious Bandersnatch (talk) 23:25, 4 October 2020 (UTC)

Model number property?

Is there any property for the model number of devices?

I could not find one.  – The preceding unsigned comment was added by 84.147.38.26 (talk • contribs).

• There is a related proposal at Wikidata:Property proposal/part number. --- Jura 09:55, 5 October 2020 (UTC)

Wikidata weekly summary #436

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

• Discussions
  • Open request for adminship: Fuzheado

• Events
  • Upcoming: Next Linked Data for Libraries LD4 Wikidata Affinity Group call: Professor Toby Hudson on Entity Explosion, a browser extension which allows you to get information about entities on the web page you’re visiting, 06 October. Agenda
  • Upcoming: Wikidata & Wikibase office hour, October 27th at 17:00 CEST (15:00 GMT)
  • Upcoming: plenty of events for the Wikidata birthday, check the calendar and feel free to organise one!

• Press, articles, blog posts, videos
  • What does/doesn’t follow Benford’s law: An analysis of hundreds of sets of numbers from Wikidata.
  • Video: How to create and edit items on Wikidata (in Spanish) - YouTube
  • Video: Introduction to Wikidata and wikiprojects of interest to librarians (in Spanish) - YouTube

• Tool of the week
  • IllWill.js is a userscript that searches Wikidata for foreign-language sitelinks, to populate Template:ill and replace plain red links on other Wikimedia projects.

• Other Noteworthy Stuff
  • omeka-s-wikidata is a new module for the open source GLAM collection management software Omeka, that allows cataloguers to use Wikidata as a vocabulary for persons, places, subjects, etc.

• Did you know?
  • Newest properties:
    • General datatypes: TDD number, twin town sign, character designer, Deutsche Bahn station code
    • External identifiers: Museen Dresden article ID, SSYK 2012 The Swedish Standard Classification of Occupations, LoC HABS/HAER/HALS place ID, Symptom Ontology ID, photoLondon ID, Syro-Malabar Catholic Church ID, Native Plants Database ID, Lower Sorbian place name ID, Les roses cultivées à l'Haÿ en 1902 ID, Maniadb album ID, USA Water Polo Hall of Fame ID, Présence Compositrices work ID, Regional Council of Tuscany ID, Indonesian prison database ID, Encyclopaedia Beliana ID, Twitter topics ID, Naver movie ID, Bollywood Hungama movie ID, BoardGameGeek family ID, London Stock Exchange company ID, Région Île-de-France ID, Slovník českých nakladatelství 1848-1949 person ID
  • New property proposals to review:
    • General datatypes: Infobox based on, CompositionToday composers showcase ID, parallel street, World Factbook country code, port calls, described at Wikipedia page, erected by, PIM award ID, Stadtwiki Dresden article
    • External identifiers: Quebec Dams Directory ID, Doktori.hu ID, CompositionToday composer ID, Slovník českých nakladatelství 1848-1949 publishing house ID, ACER code, Dictionnaire biographique du fouriérisme ID, NSW State Archives and Records Authority ID, SwissLipids identifier, Scottish Brick History Brick & Tileworks ID, Minecraft UUID, Minecraft username, Science Museum Group ID, Kicker.de player ID (actual scheme), RNACentral ID, NPS people ID, Directory of Belgian Photographers-ID, Dresden Street ID, Spotify playlist ID
  • Query examples:
    • Birthplace of people named Michael or any other variant of the name (Source)
    • Map of outdoor warning sirens, color-coded by manufacturer. A layer selector lets you toggle individual manufacturers on and off. Clicking a siren shows its make and model if available. - OSM/Wikidata query
    • Soviet cartoons in descending order of popularity in the world (Source)
    • List of ISO countries showing each country's synonyms, labels, etc. (Source)
    • Norwegian politicians who have been members of at least 3 parties (Source)
    • Streets in Leiden named after people/things (Source)
    • Maps of arts and humanities research grants funded by Arts and Humanities Research Council in UK between 2013-2019. Sorted by year, Sorted by budget (Source)
  • Newest database reports:
    • EntitySchema directory - feedback welcome on the report talk page.

• Development
  • First pieces of the Query Builder are coming together and we are working towards making it generate the first dumb query
  • Talking to people about their thoughts on checking Wikidata's data against other databases. If you're interested reach out to Mohammed.
  • Got some first internal feedback for the REST API specification draft. We'll publish it for feedback in the next days.
  • Fixing an issue with certain language codes no longer being available (phab:T264294)
  • Working on making sure blocked bots don't waste Item IDs when trying to create new Items (phab:T232620)

You can see all open tickets related to Wikidata here. If you want to help, you can also have a look at the tasks needing a volunteer.

• Monthly Tasks
  • Add labels, in your own language(s), for the new properties listed above.
  • Comment on property proposals: all open proposals
  • Contribute to a Showcase item.
  • Help translate or proofread the interface and documentation pages, in your own language!
  • Help merge identical items across Wikimedia projects.
  • Help write the next summary!

Read the full report · Unsubscribe · Mohammed Sadat (WMDE) 16:05, 5 October 2020 (UTC)

Merge request

Is there someone that can kindly merge Category:Zecchino d'Oro singers (Q65626334) and Category:Zecchino d'Oro singers (Q9201295)? Thanks in advance!!! --2001:B07:6442:8903:212F:CE8C:ABB7:9A15 16:44, 5 October 2020 (UTC)

 Done--Ymblanter (talk) 20:22, 5 October 2020 (UTC)

follows/followed by (P155/156) as qualifiers only

Have a look here for some suggestions for the migration of these properties from main statements to qualifiers.

On 11 July 2019 @Matěj Suchánek: added a property scope constraint to followed by (P156) (copied over shortly after to follows (P155)) that establishes that these should only be used as main statements or as qualifiers. This, however, appeared to contradict discussion on the latter property's talk page that suggested that it should only be used as a qualifier, and so a few days back I removed the "as main statement" qualifier on that constraint. At present, however, these removals have been questioned and reverted. What follows is an elaborated version of an argument I made in favor of the mandatory requirement of use as a qualifier four years ago:

• Consider the date January 1, 2016 (Q22266535). We can say that it "follows" December 31, 2015 (Q22266533) and is "followed by" January 2, 2016 (Q22266537), but this is only with respect to being a calendar day of a given year (Q47150325). As a January 1 (Q2150), it follows January 1, 2015 (Q18705603) and will be followed by January 1, 2017 (Q28094918). (As a Friday (Q130), it follows December 25, 2015 (Q22266524) and is followed by January 8, 2016 (Q22266544), and as a first Friday in January (Q51155371), it would follow January 2, 2015 (Q18705600) and be followed by January 6, 2017 (Q28135175).) See how this date follows and is followed by so many others with respect to so many different criteria?
• Or consider the number 25 (Q79986). It is natural, deficient, square, Friedman, odd, centered octagonal, centered octahedral, and centered square, and for some of these values of instance of (P31) different values of follows (P155)/followed by (P156) qualifiers are present, the relocation to statements of which would de-emphasize the other sequences of which the number is a part.
• Or consider the race 1994 San Marino Grand Prix (Q171552). Depending on whom you ask, some might say it "follows" 1993 San Marino Grand Prix (Q516041) and is "followed by" 1995 San Marino Grand Prix (Q385403) (by virtue of it being a San Marino Grand Prix (Q9122)), but others might say it "follows" 1994 Pacific Grand Prix (Q424174) and is "followed by" 1994 Monaco Grand Prix (Q419410) (by virtue of it being part of the 1994 Formula One World Championship (Q69713)).
• Or consider the episode Gethsemane (Q5554456). To say it "follows" Demons (Q5256287) might seem unambiguous, but the "followed by" counterpart could be either Redux (Q50279722) (since the episode is part of the series The X-Files (Q2744)) or "no value" (since it is the final episode of the series The X-Files, season 4 (Q3468921)), and to use one (or even both!) as main statements would introduce ambiguity.
• Or consider the album The Pearl (Q3522197). With respect to this being part of Brian Eno's albums in chronological order (Q98208209), it "follows" Music for Films Volume 2 (Q6941775) and is "followed by" Thursday Afternoon (Q7799409), but with respect to this being part of a similar chronologically ordered list of Harold Budd's albums, it "follows" "Abandoned Cities" and is "followed by" Lovely Thunder (Q6692291) (and the current P155/P156 values on that album's item do not reflect this difference in ordering particularly well).

To detach the ordering of any of these items from the sequences to which they are with respect (by not having P155 or P156 as qualifiers to some other clarifying property) seems rather disingenuous. Numerous other sequences (of recurring events, of multiple series of texts, of story arcs in TV shows, or of really anything else) may be adjusted in exactly this same fashion. A rather similar argument can be made with respect to the use of similar sequence properties such as replaces (P1365)/replaced by (P1366) as qualifiers only (and in fact has likely been made before).

I thus ask what circumstances from a data modeling standpoint (we all know that badly modeled data on Wikidata hurts its users) prevent the adoption of a constraint on these properties requiring them to be used as qualifiers on some other property. @Infovarius, Jon Harald Søby, Jklamo, Billinghurst: as active users from the linked talk page discussion, @Hjart: who has taken issue with my line of reasoning recently, and @Multichill: if you want to pummel den pummel again. Mahir256 (talk) 05:06, 4 October 2020 (UTC)

From my PoV if it follows et al, then logically it has to be as a qualifier as there is always some criteria for it to be following. What is it doing otherwise? If someone writes spy novels and film scripts and children's book, then something could have multiple follows, so unless you are specifying the item it is ambiguous. Sure sometimes it is obvious what it follows, however, that doesn't mean that it still shouldn't be as a qualifier.  — billinghurst sDrewth 07:13, 4 October 2020 (UTC)

There are currently roughly 1M uses as main statements and ~550k uses as qualifiers for both properties. In other words: direct use as main values is pretty wide-spread, in fact more than use as qualifiers. Before restricting the property to qualifier-only use, I would like to see a migration plan. In many situations it is not that obvious where the main statements should be moved to. It does not help to consider two times 1M statements as constraint violations when nobody cares about them anyways. —MisterSynergy (talk) 07:59, 4 October 2020 (UTC)

See Property_talk:P156#Constraint . -- MovieFex (talk) 08:34, 4 October 2020 (UTC)

@Billinghurst, MisterSynergy, MovieFex: Here are some suggestions for the migration of such statements. I am open to modifications, though probably not outright shelvings, of the proposals therein. (@Jura1, Bouzinac, Ghouston, Jane023, Jklamo: as others who have opined in this thread who may be interested.) Mahir256 (talk) 21:38, 4 October 2020 (UTC)

As a general note, I will leave you to fix up the problematic uses, I was more stating that I saw them as qualifiers for how we direct people to use them and the constraints that we consider appropriate.  — billinghurst sDrewth 00:35, 5 October 2020 (UTC)

As I wrote in Special:Diff/1281493739, my intention was just to track invalid use of these properties in references (often confused with qualifiers). If there is a consensus for qualifiers-only, no problem with it, the constraint can be updated. But as MisterSynergy says, there is still usage of both. --Matěj Suchánek (talk) 08:43, 4 October 2020 (UTC)

• There are cases where this works well as main property, but frequently it's preferable to use this as qualifier only. In the way Wikidata works, I don't think we are able to sort between the two efficiently without having properties for each usecase type of use. --- Jura 09:20, 4 October 2020 (UTC)

Was also puzzled when modelling for instance that item Fallon (Q98506059) : depending on USA/UK PoV, it is followed either by Banon (Q98505191) or Jara (Q98506539)... Bouzinac ^{💬●✒️●💛} 09:28, 4 October 2020 (UTC)

In some cases it may not be clear which statement you'd attach the qualifer to. E.g., on Never Gonna Give You Up (Q57) the followed by (P156) is apparently the next single performed by Rick Astley. To generate that as a list, you don't need a P156, just select the items using instance of (P31) and performer (P175) and order by publication date (P577), but it couldn't be used for an infobox, like on en:Never Gonna Give You Up. Presumably you'd delete the P156, on the grounds that it's meaningless without context and there's no place for it as a qualifier, and since the infobox isn't using Wikidata anyway. Ghouston (talk) 10:32, 4 October 2020 (UTC)

I use follows/followed by for paintings that considered to be related but only specifically separate in time (still don't know how to relate similar paintings not considered to be separate in time). So if the time sequence is A,B in which B is either copied from A or inspired by A, or strongly similar to A without being a strong copy of any aspect, then I use these properties. If however, B is documented as a copy of A, then B is based on A and is a derivative work of A (note the difference). If B is documented as inspired by A (because it could be inspired by some other "C" that is inspired by A or came before A), then I do not add B as a derivative work of A, but I say B is inspired by A and A is followed by B. In any case I agree that these should not only be qualifiers, but admit that I often ignore constraints because I can't even. Jane023 (talk) 10:50, 4 October 2020 (UTC)

We are using P155/156 as main statement at Wikidata:WikiProject Companies/Properties for mergers, demergers and spin-offs, as there is no special property (except merged into (P7888)). --Jklamo (talk) 13:12, 4 October 2020 (UTC)

@Jklamo: Isn't it a qualifier to instance of (P31)? Yes, some of these cases where it has been considered "obvious" it is that they are meant to be qualifiers of the base property. FWIW I find companies, especially book publishers problematic from 18th -> 19th -> 20th -> 21st they end up being a bit of a gludge in the allocation.  — billinghurst sDrewth 00:23, 5 October 2020 (UTC)

Qualifiers to instance of (P31) are not generally a good idea as far as I'm aware. The case of organizations seems a very clear use for these properties as main statements. I don't understand this push to prevent such obvious use. ArthurPSmith (talk) 15:01, 5 October 2020 (UTC)

@Jklamo: In the case of individual organizations, there is certainly a good argument for having properties for organizational spin-offs (we already have has spin-off (P2512)) just as there are for mergers; the relationships between such legal predecessors/successors deserves to be made more unambiguous in those cases. @ArthurPSmith: A specific organization by itself is not inherently part of a specific sequence for which "follows"/"followed by" is needed, but forms part of it with relation to some property/facet/quality/(whatever you want to call it) of the item, and I don't understand what's wrong with making "[the] series of which the subject is a part" (i.e. this relationship between these two) more explicit. Have a look at the migration suggestions linked in my reply to MisterSynergy/MovieFex's comments above. Mahir256 (talk) 15:23, 5 October 2020 (UTC)

@Mahir256: You know I think I was confused - we generally use replaces (P1365) and replaced by (P1366) for organizations (or merged into (P7888) where applicable), not P155/P156. Though I think the semantics is quite similar. But see for example the sequence of departments of agriculture in Australia - Department of Agriculture, Fisheries and Forestry (Q17000710) replaced by Department of Agriculture (Q21533128), replaced by Department of Agriculture and Water Resources (Q4294668) replaced by Department of Agriculture (Q65044771), replaced by Department of Agriculture, Water and the Environment (Q85756401). Those seem like they definitely belong as main statements, not qualifiers. ArthurPSmith (talk) 15:38, 5 October 2020 (UTC)

@ArthurPSmith: Couldn't these qualify instance of (P31) ministry of agriculture (Q1364302) (qualified with of (P642)/applies to jurisdiction (P1001) Australia (Q408))? I'm pushing less for every P1365/P1366 use to be moved to qualifiers (hence why my suggestions on Property talk:P156 do not mention those properties), but still contend that many situations in which they are used can be clarified with a move to qualifiers. Mahir256 (talk) 15:50, 5 October 2020 (UTC)

I generally support the idea that where possible succession information should be moved to qualifiers, rather than having them as main statements, but I also agree with ArthurPSmith that instance of (P31) is almost always going to be a bad place for these. If nothing but P31 seems reasonable, then that's probably a sign that we're missing a more useful property. For things like government departments/ministers, using P31 (and often multiple P31 statements) to essentially denote the make-up of the 'portfolio' is already at its limits in many cases, and has caused some rather contorted histories where key elements have been shunted around different departments. Having a distinct property for these would make it easier to be more explicit what actually happened when a "Department for Culture, Art, and Sport" was replaced by a "Department for Culture and Heritage", and a "Department for Sport and Leisure", for example. --Oravrattas (talk) 16:23, 5 October 2020 (UTC)

@ArthurPSmith, Mahir256: replaces (P1365)+replaced by (P1366) combinations don't always work for territories, especially for items that are about Crimea, Palestine, etc. Because they can be replaced each other by several billion times, and wasting memories of Wikidata servers to save all values of both may cause fatal errors when visiting those items. --Liuxinyu970226 (talk) 23:17, 5 October 2020 (UTC)

Spontaneous idea: has anyone already considered doing this the other way, i.e. P155/P156 only as main statements, and an optional qualifier to indicate which series this main value refers to? This would avoid the difficulty that we would need to find some property where P155/P156 fit as qualifiers. I also think that the P155/P156 claims would be much more visible and accessible that way. —MisterSynergy (talk) 09:46, 6 October 2020 (UTC)

Recently I had the same idea but forgot to share... --Matěj Suchánek (talk) 09:50, 6 October 2020 (UTC)

• Interesting idea. So instead of <series>qualified by<next> we would have <next>qualified by<series>. --- Jura 09:53, 6 October 2020 (UTC)

That could become pretty unreadable if there are multiple statements that each have their own "followed by". position held (P39) more often takes replaced by (P1366) as a qualifier than followed by (P156), but nevertheless consider Q9576#P39 as an example of just how many sequencing qualifiers an item might take. Jheald (talk) 10:15, 6 October 2020 (UTC)

I think P39 should NEVER use P155/156. --- Jura 10:22, 6 October 2020 (UTC)

Here's a query for items with the largest number of genuine followed by (P156) qualifiers: https://w.wiki/fMt

I am not sure what I think about its use to qualify properties has part(s) (P527) and has part(s) of the class (P2670), but it's evidently something we need to take into consideration. Jheald (talk) 10:36, 6 October 2020 (UTC)

Thanks for that query; here is a modified version that takes into consideration the main statement property where the P155/P156 qualifiers are sitting. I would say they are either select special cases, or poorly modeled anyways. —MisterSynergy (talk) 11:15, 6 October 2020 (UTC)

Some of them maybe could be better modelled. But the uses with part of (P361) look appropriate, eg for letters which occur in different versions of the alphabet (with different conventional orderings). It also highlights that the 'followed by' may be naturally associated with other qualifiers, eg series ordinal (P1545), that may make sense to keep together with each other. (Most commonly co-occurring qualifiers: https://w.wiki/fQ3 ) Jheald (talk) 14:48, 6 October 2020 (UTC)

This seems like it could get awkward if someone has held exactly the same position multiple times — e.g. it's not uncommon for someone to be Prime Minister of a country, lose the next election, and then win the one after that again. So they would replace / be replaced by someone else in the same office multiple times (sometimes even succeeding and being succeeded by the same person in turn multiple times). We would thus need to further qualify each of those replaces/replaced statements somehow to know which is being referred to each time, and I'm not sure how we would distinguish those other than by copying the dates across, which gets unwieldy and repetitive quite quickly. (Norodom Sihanouk is an especially extreme case here.) --Oravrattas (talk) 14:04, 6 October 2020 (UTC)

@Jheald, Oravrattas, MisterSynergy: While this seems to pertain to P1365/P1366 main statements (the movement of which as I noted to Arthur above I am pushing less for compared to P155/P156, which titles this talk page section and my suggestions on the talk page of which have not attracted much attention as yet), I also find this rather unworkable for two reasons: 1) it decouples a sequence (noted by a main statement with a given property) to which an item belongs from the items in that sequence (qualified with that same property or something similar), making it prone to desynchronization (avoiding which is the whole point of my proposals in the first place), and 2) it appears to suggest that an item inherently has some position in an ordering (indicated by P155/P156 being more prominent than the sequence property qualifying it) that end users (no matter how much you inform them about any qualifiers that might be present) may well mis(s/interpret), whereas confining the scope of P155/P156 to a particular property will generally avoid this problem. Mahir256 (talk) 14:45, 6 October 2020 (UTC)

Periodicals' start and end

Some works published over a period time use start time (P580) for the publication date of its first part and end time (P582) for its last part. Obviously, one could also use additional inception date for when the work was first conceived.

Periodicals currently use either start time (P580)+end time (P582) or inception (P571)+dissolved, abolished or demolished date (P576) for the date of publication of the first and the last issue. #Venue item properties suggests the later. To harmonize them in one way or the other, I'd use just the first group also for periodicals. --- Jura 09:29, 5 October 2020 (UTC)

I would just focus on why there are still peoples that use P580/582 as main statements, what are they reflecting? --Liuxinyu970226 (talk) 13:46, 6 October 2020 (UTC)

Backs of postcards

See Commons:Category:Backs of postcards. It would be quite usefull to have data-item for structured data. We have Q192425 and Q68345931 but not the combination wich have functional properties, such as handwritten text and poststamps.Smiley.toerist (talk) 12:52, 1 October 2020 (UTC)

PS:I dont seem to able to use Q68345931 as a property of Q192425. P31 can only have one value. Smiley.toerist (talk) 13:06, 1 October 2020 (UTC)

• There is image of backside (P7417) for use directly on Commons. --- Jura 04:54, 2 October 2020 (UTC)

I tried to apply this property to File:AK - Fröhliche Weihnachten - 1914 A (cropped).jpg, I could not add a qualification to the postcard. Another problem is that there date-item: 'posted on'. This is not the same as a publication date.Smiley.toerist (talk) 19:24, 2 October 2020 (UTC)

• You need to ask on Commons to have the datatype of P7417 made available. Currently, I think they started only with string-, date-, and item-datatype properties. --- Jura 06:25, 3 October 2020 (UTC)

I created a new Wikidata-item: Q100140585 Smiley.toerist (talk) 16:21, 6 October 2020 (UTC)

Adding values for another language in constraint clarification (P6607) as a qualifier

Does anyone know how to add more values for constraint clarification (P6607) as a qualifier in another language? For example, in typically sells (P7163), there is one constraint clarification (P6607) value in English as property generally should have just 1 or a few values. But if you change your interface language to, say French, and open an item with "Suggestions" in typically sells (P7163) (like flea market (Q385870)), you would retrieve the French value of it Cette propriété contient généralement une seule valeur if you click on the flag. But if you open the page for typically sells (P7163) you would only read the English value for constraint clarification (P6607) and I think I also can't find how to add another language. RXerself (talk) 09:53, 7 October 2020 (UTC)

@RXerself: The message you see on flea market (Q385870) is an interface message of the WikibaseQualityConstraints extension, not the constraint clarification (P6607) of the constraint. (Even in English, you’ll see This property is generally expected to contain only a single value., whereas the constraint clarification (P6607) is property generally should have just 1 or a few values.) Translations of the message (wbqc-violation-message-single-value) are maintained on TranslateWiki.net. --Lucas Werkmeister (WMDE) (talk) 10:25, 7 October 2020 (UTC)

@Lucas Werkmeister (WMDE): Thank you! The message group also turns out to be the one I was looking for a month ago. RXerself (talk) 11:17, 7 October 2020 (UTC)

Sindhi Language Improve to Survive

i am a sindhi and want to submit any project about our sindhi language.  – The preceding unsigned comment was added by 42.201.234.34 (talk • contribs).

Welcome on Wikidata! It would be very helpful if you add Sindhi language labels and descriptions to items. (guide) --Pyfisch (talk) 13:28, 7 October 2020 (UTC)

Call for feedback about Wikimedia Foundation Bylaws changes and Board candidate rubric

Hello. Apologies if you are not reading this message in your native language. Please help translate to your language.

Today the Wikimedia Foundation Board of Trustees starts two calls for feedback. One is about changes to the Bylaws mainly to increase the Board size from 10 to 16 members. The other one is about a trustee candidate rubric to introduce new, more effective ways to evaluate new Board candidates. The Board welcomes your comments through 26 October. For more details, check the full announcement.

Thank you! Qgil-WMF (talk) 17:09, 7 October 2020 (UTC)

Modelizing Scrabble (Q170436)

How to state one can play Scrabble (Q170436) in French (Q150), Arabic (Q13955), German (Q188), English (Q1860), etc + numbers of letters ? More info on Scrabble languages can be found https://en.wikipedia.org/wiki/Scrabble_letter_distributions here Bouzinac ^{💬●✒️●💛} 19:11, 7 October 2020 (UTC)

Brackets in labels of scientific articles

Hi, I noticed that these 2 [4] [5] have superfluous brackets and full stop in the labels and titles. It should be fixed IMO and the bot owners should be contacted to clean the data better before import. Any takers?--So9q (talk) 07:32, 1 October 2020 (UTC)

• The brackets have a meaning, see User:Research Bot/issues#non-English article.

Wikidata:Property_proposal/translated_title proposes a solution. --- Jura 07:35, 1 October 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ From which, "To do... remove brackets from English label". They should indeed be removed. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:59, 3 October 2020 (UTC)

• It's a step after the info was added in a structured way.

Ideally we would need a property to add the title as well. Somehow I doubt pubmed does literal translations. I wonder led people to conclude that. Sounds somewhat offensive. --- Jura 11:14, 8 October 2020 (UTC)

Update database report

Wikidata:Database reports/Constraint violations/P4633 is clearly not up-to-date. Section "violations of Q7777570 and Q204854" should list a lot of stuff, for example Iron Man 3 (Q209538), but it doesn't. How to force the update of report? --Kanzat (talk) 13:01, 7 October 2020 (UTC)

subject type constraint (Q21503250) does only work for as main value (Q54828448).[6] But I have made a query for number of "violations of Q7777570 and Q204854" if you are interested. https://w.wiki/fmi -Premeditated (talk) 12:16, 8 October 2020 (UTC)

Link to list of showcase items in Main page

The main page of Wikidata has a link to Douglas Adams (Q42) as a showcase item. Is there a list of showcase items for every item type? (or at least for the most popular ones) If not, need to create it. I.e. showcase movie, politician, football player, book etc.. Once it's ready - we can mention it in the Main page, it would be very helpful, because Wikidata is still too complex for most users. --Kanzat (talk) 13:40, 7 October 2020 (UTC)

First, there are no showcases promoted since 2018. Second, we need a clear rule of how to promote (or demote) a showcase.--GZWDer (talk) 14:56, 7 October 2020 (UTC)

There's model item (P5869) which has a lower standard than showcase item but may give you more examples to work with.

The following query uses these:

• Properties: model item (P5869)   

  SELECT ?type ?typeLabel ?item ?itemLabel
  WHERE 
  {
    ?type wdt:P5869 ?item.
    SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
  }
  

Try it!

- cdo256 16:32, 7 October 2020 (UTC)

• Given the dynamic nature of Wikidata, what may be showcase today wont be a month later.

A regularly updated, random selection might work better, e.g. for films: Wikidata:WikiProject Movies/reports/random/film/samples. --- Jura 11:23, 8 October 2020 (UTC)

Stone lanterns

Would someone have a look at stone lantern (Q24577890)? Doesn't look well modeled, especially has part(s) (P527). - Jmabel (talk) 01:21, 8 October 2020 (UTC)

I added made from material (P186) stone (Q22731), but I'm not sure about the other part building stone (Q11250576). Ghouston (talk) 01:50, 8 October 2020 (UTC)

I think there's some confusion between a stone as a natural object and stone as a material. The two commons categories are c:Category:Stone (material) and c:Category:Stones, and there are separate Japanese Wikipedia articles. Ghouston (talk) 21:14, 8 October 2020 (UTC)

Russian: Merge Q427948 and Q4456859?

Is the Russian article about the same subject as the others? The English label is the same ("scope statement"). Pyfisch (talk) 11:28, 8 October 2020 (UTC)

No, I've updated labels and added few statements for clarification. --Lockal (talk) 15:27, 8 October 2020 (UTC)

Items for Sockpuppet Categories?

@Magog the Ogre: you are creating pages for sockpuppet account categories[7] such as Category:Sockpuppets of Masato Koizumi (Q99292357) or Category:Wikipedia sockpuppets of Noalla (Q67824369). I read WD:N, but did not find that it excludes "adminstrative, project internal categories" or something similar. Are these categories useful to the project, should they be kept?
By the way Anjooran (talk • contribs • logs) already found a use for Category:Suspected Wikipedia sockpuppets of Sajith Thomas (Q99658738) and turned the item into a monument for the puppeteer.
--Pyfisch (talk) 22:04, 8 October 2020 (UTC)

Wikidata:Notability/Exclusion criteria is important here —MisterSynergy (talk) 22:20, 8 October 2020 (UTC)

Thanks. These pages are easy to miss if you are focused only on the rules text on the main page. --Pyfisch (talk) 22:35, 8 October 2020 (UTC)

This is new to me. When the issue came up before they'd decided it was OK. Magog the Ogre (talk) 20:13, 9 October 2020 (UTC)

Animal Glue: Merge Q98479449, Q1552509 and Q21782673?

It would seem all three items are about "Animal Glue" and I have no idea how to merge them. Could someone take a look or point me to the right tutorial? Thanks! --Agnat (talk) 11:53, 9 October 2020 (UTC)

WD:Merge. However, these three items shouldn't be merged, only two of them. Wostr (talk) 13:14, 9 October 2020 (UTC)

One abuse filter

Discussion started but all thank that I want to edit user page (but I only want change blocking mechanism) and archived without result or refuse. 217.117.125.72 14:30, 9 October 2020 (UTC)

• I think there is still an open question for you and it was explained why not. --- Jura 14:44, 9 October 2020 (UTC)
  • Which question is still open? 217.117.125.72 15:56, 9 October 2020 (UTC)
    • My question isn't answered. ChristianKl ❪✉❫ 16:28, 9 October 2020 (UTC)

Link issue with Q12629755

For some reason the links to this item don't want to show up on the English article, and the Piedmontese article links to a different English title, can somebody figure out what is going on? --Glennznl (talk) 14:37, 9 October 2020 (UTC)

• There is an old style interwiki at https://pms.wikipedia.org/w/index.php?title=Lenga_oriya_desiya&action=edit --- Jura 14:42, 9 October 2020 (UTC)

@Jura1: Thanks, I fixed that now, but the English page still does not want to show the links. I tried deleting and re-inserting the link two times now to no avail. --Glennznl (talk) 14:54, 9 October 2020 (UTC)

I have purged the en.wiki page, which seems to have cleared the issue there. --Malcolmxl5 (talk) 18:49, 9 October 2020 (UTC)

What language code is to be used for scientific binomials? Or is there an alternate proprerty to replace "named after"?

As is, named after requires the language of the name in question to be entered. This is problematic whenever that name isn't a natural language, as is the case for scientific nomenclature. Organisms and genera get frequently named after localities, people, sometimes also mythological/fictional figures. Entering a "named after" datum now requires incorrect labelling, since eponymic scientific nomina are not Latin or Greek, but also not the language of the name used for the eponym. Some eponyms are not any language, such as "Etia" (derived from the initials of Ethelwynn Trewavas plus suffix -ia), or "Wahydra" (derived as an anagram of Hayward).

-- Pitke (talk) 05:02, 9 October 2020 (UTC)

• @Pitke: named after (P138) itself doesn't require the language to be specified. One of its qualifier may, if you deem it necessary.
  There some discussion about applicable language code at Property_talk:P225#IETF_tag_for_values_and_its_parts (re: taxon name (P225)). Maybe a new (IETF) subtag for some language would be useful. --- Jura 11:32, 9 October 2020 (UTC)
• mul (or mul-lat) would be the language code when the name comes from multiple languages. ChristianKl ❪✉❫ 13:48, 9 October 2020 (UTC)

Thank you both! --Pitke (talk) 17:50, 9 October 2020 (UTC)

@ChristianKl: Turns out "mul" cannot be used as a language code for this. See statement level "named after" at Q742818. Attempt to add qualifier "applies to name of item". A language code is forced, and "mul" does not exist as an option. --Pitke (talk) 17:57, 9 October 2020 (UTC)

• I fixed the qualifier value. Maybe "la-x-taxon" (until there is "la-taxon") should be used. --- Jura 06:26, 10 October 2020 (UTC)

named after (P138) qualified with applies to part (P518)=epithet (Q207869) is another possible solution. BTW: Infact Phasianus impejanus (Q100250410) was named after her. --Succu (talk) 18:32, 9 October 2020 (UTC)

Exact values, ±0

Wikidata allows storing upper and lower bounds for numbers: A number like 1.23 on its own does not indicate how "well" the value is actually known. 1.23±0.02 indicates that the actual value is believed to be between 1.21 and 1.25. Now comes the tricky part: When entering 251 in Wikidata, it is interpreted as a decimal, without any assumption about is exactness, not as a precise integer as one might assume. To make the value precise, one enters 251±0, indicting that the lower and upper bound coincide, thereby giving the value no room for variation.

There are various places where one knows a value exactly, typically if one defines a quantity to have a certain value. Examples include the modern speed of light expressed in SI units and certain unit conversions (1 km = 1000±0 m, 1 yard = 3±0 feet).

Historically, Wikidata displayed numbers differently and to work around it, people added ±0 to hide a decimal point, not because the number was known exactly. Now that Wikidata's display of numbers has been fixed, those ±0+workarounds are being removed.

I've noticed over the past year that ±0 have been removed even in cases where they were justified. I hope that this little summary will help to alleviate the myth that "±0 is bad" and that it now serves a legitimate purpose.

Best wishes. Toni 001 (talk) 07:58, 9 October 2020 (UTC)

In spite what you say, I am wondering whether integer constraint (Q52848401) and no-bounds constraint (Q51723761) aren't being misinterpreted/misused. --Matěj Suchánek (talk) 10:34, 9 October 2020 (UTC)

Why do you think they are being misinterpreted? I worked with them quite a lot and I think I even proposed them here, so I am genuinely interested in your opinion. —MisterSynergy (talk) 10:57, 9 October 2020 (UTC)

My point was: When entering 251 in Wikidata, it is interpreted as a decimal – doesn't this beat the purpose of Help:Property constraints portal/Integer? (I've read Help:Property constraints portal/No bounds which explains the purpose, so no problem with it.) --Matěj Suchánek (talk) 11:19, 9 October 2020 (UTC)

I did not fully understand this part of the original comment, thus a clarification: for any quantity data type property, you can input float numbers (i.e. quantities with decimal places). This does not always make sense, so for some properties decimal places should usually not occur. The "integer constraint" warns if a quantity data type property claim has decimal places, but shouldn’t. —MisterSynergy (talk) 11:34, 9 October 2020 (UTC)

With decimal I was referring to the semantic web representation of the values. Look for instance at the Turtle (.ttl) representation of an item (copy the "concept URI" of an item, paste it into the address bar and add ".ttl") and locate the value corresponding to, say, the population. This aspect is relevant for external tools which don't look at the user interface, but only at the machine readable data. Toni 001 (talk) 09:22, 10 October 2020 (UTC)

Some remarks:

• Per mw:Wikibase/DataModel#Quantities and Help:Data type#Quantity, the exact meaning of bounds is undefined. We should thus be very careful when using them.
• In the beginning, "±0" had always been added to quantity data type claims by default. This had been changed so that they are only being added when the user explicitly wants them, and we removed the unintentional bounds from most places where we thought they do not belong to.
• Furthermore, bounds should be used as in the source where the value was imported from. We should not calculate any bounds by ourselves.

—MisterSynergy (talk) 11:02, 9 October 2020 (UTC)

Academia.edu profile URL (P5715)

Hi! Is anyone adding data to P5715? Is there a guideline about this type of data, regarding how and who is linked in wikidata? Is a profile at academia.edu reason enough to add it to wikidata? Mysteriumen (talk) 16:39, 8 October 2020 (UTC)

I guess, if it's a self-edited site, it can be added for existing items but doesn't suggest notability itself, much like an account on Facebook, Twitter or LinkedIn. Ghouston (talk) 21:04, 8 October 2020 (UTC)

I does seem like a social media, doesn’t it. What do you mean by self-edited site? If it is based on user contribution? Is there a policy/guideline on wikidata for this? Mysteriumen (talk) 21:49, 8 October 2020 (UTC)

A related question is of course if Academia.edu publication ID (P7896) should be populated with articles, and according to a certain criteria? Some are preprint articles scheduled for publication Mysteriumen (talk) 21:49, 8 October 2020 (UTC)

It's generally helpful to add most available identifier to existing items. This includes both Academia.edu publication ID (P7896)/Academia.edu profile URL (P5715) --- Jura 13:57, 9 October 2020 (UTC)

Jura1 Thanks Mysteriumen (talk) 20:55, 10 October 2020 (UTC)

Why does LOC Classification have a distinct values constraint?

I must be misunderstanding what Library of Congress Classification (P1149) is used for -- is it not supposed to be applied to works, to indicate what classification the LoC cataloguing-in-publication data states? Because for that, a distinct-values constraint makes no sense. Even if it's to be applied to items that could be topics, there might not be distinct values because of lack of one-to-one correspondence between Wikidata and LoC classification schemes. — Levana Taylor (talk) 22:15, 9 October 2020 (UTC)

• I think you are looking for Library of Congress Classification (works and editions) (P8360) --- Jura 06:03, 10 October 2020 (UTC)

Oh, OK. Now it makes sense, sort of. — Levana Taylor (talk) 06:54, 10 October 2020 (UTC)

That mistake's been committed some 250-300 times; I just fixed about 40 of them (other than my own) — Levana Taylor (talk) 09:10, 10 October 2020 (UTC)

Library of Congress Classification (works and editions) (P8360) is recent. Initially, it was a bit of mix-up. --- Jura 16:10, 10 October 2020 (UTC)

How to cleanly split an item into separate items?

So there's a big problem with a lot of Wikidata items concerning anime and manga, in that they often share the same item. For example, there's a manga named "Horimiya" (Hori-san to Miyamura-kun (Q3786774) which has a 4 episode short anime adaptation which released from 2012-2018. It also has an entirely separate upcoming anime adaptation coming next year, but that's not relevant for now.

Essentially, I'd like to take the "MyAnimeList anime ID", "Anime News Network anime ID", and "AnimeClick anime ID" out from this item and create a separate item for the 2012-2018 anime. Is there a good tool for that or should I just create a new item with those attributes and then delete them off the item for the manga? There are plenty of other examples of this being a problem for anime/manga/light novels, and sometimes even video games. For example, Baccano! (Q484579) represents both the Baccano! anime _and_ light novel, when they should almost certainly be separate items.

Just curious if someone has a better idea of how to go about remedying situations like these. Thanks! Nicereddy (talk) 05:32, 10 October 2020 (UTC)

Another good example of this problem is Nichijou (Q483120), which represents the Nichijou manga, anime, and video game 🙃 Nicereddy (talk) 05:36, 10 October 2020 (UTC)

• For items with Q5, there is Help:Conflation of two persons that may help you sort it out. --- Jura 06:00, 10 October 2020 (UTC)

It's unfortunately a quite common problem with animes/mangas/light novels/etc, as pretty much all Wikipedia language versions only have long articles covering all related works (except movies for some reason). And multiple infoboxes that apparently in some cases are expected to pull infos from Wikidata., And some WP projects seem to want to keep IDs for various works on the item connected to the article, so their articles can use the IDs stored in Wikidata. On another project, someone else said to change such items into media franchise items (instance of (P31)media franchise (Q196600) and list the other works with has part(s) (P527), but I'm not convinced that actually works with infoboxes in many cases. So in many cases, I've just created a separate item for anime series and copied relevant IDs and statements there, but also left the IDs intact on the original item. --Kam Solusar (talk) 18:53, 10 October 2020 (UTC)

Byzantine

What is the difference between Q2993777 and Q12544? Of course, almost bij definition these are history articles. In some languages both are used. Example: en:History of the Byzantine Empire and en:Byzantine Empire (Quality article) with the same subject.Smiley.toerist (talk) 13:24, 10 October 2020 (UTC)

• Doe this help? Byzantine Empire (Q12544) history of topic (P2184) history of the Byzantine Empire (Q2993777) Visite fortuitement prolongée (talk) 14:35, 10 October 2020 (UTC)
• The items themselves are not articles. One item is about the country and the other item is about the history of the country.

It's not up to Wikidata to decide about whether it makes sense that en:History of the Byzantine Empire and en:Byzantine Empire are separate articles. That's a decision that's up to EnWiki and it's up to us to provide items that make the distinctions that the Wikis make. ChristianKl ❪✉❫ 16:35, 10 October 2020 (UTC)

There is no country of citizenship (P27) of Austria-Hungary (Q28513)

Tobias1984 Mfchris84 Katjos phaebz Jean-Frédéric PeterTheOne Gittenburg M2k~dewiki emu Haansn08 Maincomb

Notified participants of WikiProject Austria @Peyerk: As of this moment, Wikidata has 12731 items with a country of citizenship (P27) statement of Austria-Hungary (Q28513). There’s only a slight problem: There has never been an Austrian-Hungarian citizenship. (In fact, citizens from the other half of the Austrian-Hungarian Empire were considered aliens.) How should we deal with this problem? Can we set a constraint? Should we delete the faulty data with a bot job? --Emu (talk) 20:55, 3 October 2020 (UTC)

Tobias1984 Vojtěch Dostál YjM Jklamo Walter Klosse Sintakso Matěj Suchánek JAn Dudík Skim Frettie Jura1913 Mormegil Jedudedek marv1N Sapfan Daniel Baránek Draceane Michal Josef Špaček (WMCZ) The photonaut Hartasek Zelenymuzik Gumruch Shadster Dænča M.Rejha Janek Jan Kameníček Eva Vele Linda.jansova Lukša Packa Fukejs Hugo Xmorave2 J.Broukal Lenkakrizova Steam Flow Pavel Bednařík Sanqui

Notified participants of WikiProject Czech Republic

Powerek38 Wostr Wiklol Witia Holek Pit rock Yarl Kpjas 99kerob Matlin masti Upior_polnocy Darellur Bogic

Notified participants of WikiProject Poland --Emu (talk) 21:05, 3 October 2020

@Emu: of course "citizenship" in our current understandig is quite inappropriate in a lot of historic cases, where country of citizenship (P27) is already set by linking to the historic country. i think there is a slight common opinion, that this statement should describe a "relationship" of an human to a specific (historic) country. the german alias "Land der 'Bürger'rechte" tries to describe this. whether there weren't our current rights of citizenship, different citizen rights already exists at least for a small number of people. so i would say, country of citizenship (P27)Q28513{{{3}}} is ok. --Mfchris84 (talk) 21:57, 3 October 2020 (UTC)

I’m well aware that there has been a lot of discussion on P27 in the last years. And yes, there are some major problems before the creation of the modern Nation state. But Austria-Hungary is a clear-cut case as the modern notion of citizenship existed in 1867 (and was arguably one of the reasons for the messy creation of said state). Bürgerrechte doesn’t really help here as it was linked to a municipality. We might of course define P27 in another way, but we never tried to resolve that mess ... --Emu (talk) 22:38, 3 October 2020 (UTC)

Constraints on country of citizenship (P27) also permit using a dependent territory (Q161243) as a value; it was added during a previous discussion at Wikidata:Project_chat/Archive/2018/09#Puerto_Rican_nationality. Ghouston (talk) 22:41, 3 October 2020 (UTC)

Thank you for pointing that out! However, I’m not sure that this is a precedent as there was and is (and arguably has been) such a thing as Puerto Rican citizenship. There has never been such a thing as Austrian-Hungarian citizenship. (On a sidenote: Joseph Roth (Q78509) would have probably preferred to have this P27 but alas he doesn’t have on Wikidata …) --Emu (talk) 23:11, 3 October 2020 (UTC)

What I mean is that the citizenship could potentially be switched to one of the constituent countries of Austria-Hungary (Q28513). Ghouston (talk) 23:27, 3 October 2020 (UTC)

Maybe it helps, but there is a (possibly inaccurate) list of "Austrian" crown lands at lists/crown lands.--- Jura 09:25, 4 October 2020 (UTC)

I am not sure if it is appropriate to label these data as "faulty" and delete them without substitution. Also, I am not sure if it is appropriate to use country of citizenship (P27) Austria (Q40) for prewar citizenship, as it is more likely related to Cisleithania (Q533534) (and home law (Q680977) of each municipality) than Austria (Q40). --Jklamo (talk) 13:44, 4 October 2020 (UTC)

You are right, country of citizenship (P27)Austria (Q40) is inappropriate for pre-war citizens. But there’s at least a constraint giving a warning about that – that’s what is missing for country of citizenship (P27)Austria-Hungary (Q28513) (as there doesn’t seem to be a way to forbid a specific value via normal constraints). We do not have a Property for home law (Q680977) – place of origin (Switzerland) (P1321) could handle it but is defined as Switzerland-specific for some reason. --Emu (talk) 15:34, 4 October 2020 (UTC)

Definitely Austria-Hungary (Q28513) has ("more or less...") two national citizenship (Q42138) (so, country of citizenship (P27) of Austria-Hungary (Q28513) is not quite correct). There is this problem with vaguely defined "State" Cisleithania (Q533534) and its citizenship (on the other hand I believe we are not forced to go to the level of crown land of Austria (Q681026), because there was defined some kind of united citizenship in Allgemeines bürgerliches Gesetzbuch (Q698262)). It seems to me, that we can claim, that residents of Austria-Hungary has citizenship Cisleithania (Q533534) or Kingdom of Hungary (Q171150) (? Transleithania (Q1879239)). --marv1N (talk) 16:04, 4 October 2020 (UTC)

Cisleithania never managed to have a modern citizenship law, but it did have a unitary citizenship. Hungary had a modern citizenship law (and a unitary citizenship in principle and discrimination against non-Magyars in practice). Residency won’t suffice though as the citizenship depended mostly on ius sanguinis. In the majority of cases residence can act as a heuristic though – country of citizenship (P27) is genereally heavily unsourced and mostly relies on guesswork. A (very) long-term goal should be to make clear where it’s guesswork (and on what grounds) with based on heuristic (P887) and where we do know it. --Emu (talk) 19:09, 4 October 2020 (UTC)

You wrote it precisely: It is slightly better to claim, that anybody with citizenship within Austria-Hungary has either asutrian or hungarian citizenship (than "austrian-hungarian citizenship"). For austrian part after 1867 we can use Cisleithania (Cisleithania (Q533534) - this sounds to me a little strange, but I hope correct), not sure what to use for hungarian part. And - as always - there would be lack of sources in lot of cases. --marv1N (talk) 07:25, 6 October 2020 (UTC)

To sum up: I feel that there is a weak consensus that (1.) citizenship should be correct in a legal sense if possible but (2.) incorrect but somewhat useful statements have some merit as well. As a result, I added a property constraint (P2302)none-of constraint (Q52558054) with the constraint status (P2316)suggestion constraint (Q62026391) and the the clarification “this state had separate citizenships for Cisleithania (Q533534) and the Kingdom of Hungary (Q25395037)”. That way, it is made clear that Austria-Hungary did not have a unified citizenship but not as strict as with mandatory constraint (Q21502408) --Emu (talk) 19:33, 11 October 2020 (UTC)

Is it possible to change the message "The value for country of citizenship should not be Austria-Hungary" which says neither why it should not be nor what is the preferred attitude? It also sounds that it is wrong to include such information, the result of which can be that people who are not sure about the citizenship as exactly as required will not include the information at all. Sources do not always give such information. The message should explicitely say that it is preferred to state Cisleithania or the Kingdom of Hungary for country of citizenship if known, otherwise it is OK to state at least Austria-Hungary. --Jan Kameníček (talk) 19:49, 11 October 2020 (UTC)

Almost to 100,000,000

Tracked in Phabricator
Task T44362

Entity Q99000000 was created two days ago. At the current rate, we'll hit Q100000000 somewhere around October 6. (Or sooner than that if any importation bots start up such as were running last Spring.) But if anyone has any code or schemas that assume that Q numbers are smaller than that, that they will always fit in 9 characters, now would be a good time to fix that! —Scs (talk) 21:41, 11 September 2020 (UTC)

Looks like it's about an hour away. —Scs (talk) 03:48, 6 October 2020 (UTC)

We missed Q100000000 and the first item ID over 100 million is Franklin School (Q100000001) and was created by 99of9. — The Erinaceous One 🦔 05:08, 6 October 2020 (UTC)

How do numbers get missed? (Q99999999 was also missed) Were they pre-protected? I'll try to fill out Franklin School (Q100000001) as well as I can. --99of9 (talk) 05:21, 6 October 2020 (UTC)

@99of9: There's some discussion of skipped items in a Phabricator task (linked at the top of this talk page section) and its subtasks. Mahir256 (talk) 05:26, 6 October 2020 (UTC)

Also recently discussed in this previous project chat thread (which incidentally mentions a different Phabricator task). —Scs (talk) 14:31, 6 October 2020 (UTC)

I sent an email to Franklin Early Childhood School to let them know of their involvement in this milestone. — The Erinaceous One 🦔 09:01, 7 October 2020 (UTC)

• seems we have an aerial view of the school: File:Harrison Franklin aerial.jpg (white buildings slight to right of the image center). It could be while was under construction (image is from November 2009, inception statement has February 2019); or it's mis-identified. --- Jura 09:25, 7 October 2020 (UTC)

I tried to edit the news section what is then showed at the main page. I wanted to add the following text:

2020-10-06 Q100000001 is created. After Q100000000 was missed, this is the first Item in Wikidata consisting out of nine numbers.

Do you think that this text is good or how would you write that and can someone please add it at the page when it is good. I dont know how I can put the oldest news to the noninclude tag. This dont worked and so I dont saved the page.--Hogü-456 (talk) 19:54, 6 October 2020 (UTC)

• @Hogü-456: I've done it (and in native English). - Jmabel (talk) 21:11, 6 October 2020 (UTC)
• Or, more precisely, I've edited at Wikidata:News. Apparently there is something an admin has to do to make my edit go "live". - Jmabel (talk) 21:15, 6 October 2020 (UTC)

At Talk:Q100000001, I added a few links to navigate around a few related items, some created these days. --- Jura 17:01, 11 October 2020 (UTC)

Merge request

Is there someone that can kindly merge Bruno of Augsburg (Q75241652) and Bruno of Augsburg (Q460802)? (same person: same parents, same name Bruno, same office bishop of Augsburg). Thanks in advance!!! - Moldur (talk) 19:18, 11 October 2020 (UTC)

• Help:Merge explains how. If you add said to be the same as (P460), that actually prevents them from being merged: the assumption would probably be that not all references consider them to be the same person. --- Jura 19:39, 11 October 2020 (UTC)

Thank you! I will consider that. - Moldur (talk) 20:13, 11 October 2020 (UTC)

Somevalue or novalue in external-id

I found OpenStreetMap relation ID (P402) set to somevalue in Athens (Q1524) and to novalue in Mombasa (Q225641). Does it make any sense? I think using these two special values in external identifiers should be prohibited. Paweł Ziemian (talk) 19:23, 11 October 2020 (UTC)

• Maybe it depends on the identifier. For more static ones, "novalue" means that someone checked the list and didn't find it or it's no longer available. --- Jura 19:35, 11 October 2020 (UTC)

• I think novalue can be useful in specific contexts - "this entity should be in the external database but isn't". If for example some external database says it lists "every mountain in Sweden", but we have added all the entries and then found out it misses out five of them, then setting SwedishMountainsDB:novalue on those items means that we don't keep trying to find them in the database, and that we can feed back "you should add these items, they're known to be missing". But I wouldn't use it in this way on something more general like OpenStreetMap relation ID (P402). Andrew Gray (talk) 20:57, 11 October 2020 (UTC)

Advice on importing information on the devil recorded in historical Scottish witchcraft records.

Hi all, am working with Data Science students at the University of Edinburgh this semester and am again looking at a 6-7 week project to import some of the rich historical data in the Survey of Scottish Witchcraft database into Wikidata. As we have done in creating this Map of Scottish Accused Witches website.
As we have added temporal and geographical data previously, it would be interesting to depict the mentions of the devil as recorded in the historical documents in witchcraft investigations, and visualise that over time and location. The appearance of the devil is variously recorded in meeting notes and the types of demonic pact also. we have previously added items of data on the accused witches, the trials, and persons involved with the trials (judges, prosecutors, sheriffs, lairds, witnesses etc.) so trying to work out to model the information.
e.g. Property = Demonic_Type (refers to the type or motif of demonic pact that was described in the witchcraft documents)
Example values:

• Anti-baptism: renunciation of Christian baptism, apostasy
• Body and soul: giving oneself over to the Devil, body and soul
• Bond/Band: an agreement with the Devil
• Devil's Mark: mark received from the Devil as a sign of pact (often described as not sensible to feeling)
• Head and foot: touching of the head and foot with opposite hands – all between was given to the Devil
• Kisses Devil's bottom: worship of the Devil by inversion/perversion of Christian symbolism
• New name: new name given to a witch by the Devil indicating a rejection of Christian baptism – a re-naming by the Devil
• Paction: general, non-specific pact made with the Devil
• Possession: the accused witch claimed to be possessed by the Devil
• Servant: indicates that the accused had agreed to be the Devil’s servant
• Sex: indicated that the accused had sexual relations with the Devil
• Tacit pact: the accused used power of the pact but did not describe any specific features
• Want nothing: the accused confessed that the Devil promised to provide everything for them and that they ‘should never want’

Not sure about putting forward a property proposal for 'demonic pact' but maybe significant event (P793) Significant event > pact with the devil (Q1506690) added to the accused witch's item with the values above as a qualifier? Any help gratefully appreciated. Stinglehammer (talk) 20:04, 7 October 2020 (UTC)

Devil's appearance in Scottish witchcraft records

Additionally, we also have records on: Property: Devil_Type – This field refers to the type of non-natural being that was mentioned or described in the documents. Example values:

• Animal Devil: the Devil appeared in animal form
• Baby: the Devil appeared in the form of a baby
• Child Devil: the Devil appeared in the form of a child
• Fairy: non-natural being appeared in the form of a fairy, gender not specified
• Female: the Devil appeared in the form of a female
• Female Fairy: non-natural being appeared in the form of a female fairy
• Ghost: non-natural being appeared in the form of a ghost or dead person
• Inanimate Object Devil: the Devil appeared in the form of an inanimate object
• Insect Devil: the Devil appeared in the form of an insect
• Male: the Devil appeared in the form of a man
• Male Fairy: non-natural being appeared in the form of a man
• Other Demon: non-natural being appeared in the form of another non-specified demon
• Spirit: non-natural being appeared in the form of a non-specified spirit
• Unspecified Devil: non-natural being appeared in the form of a non-specified Devil

Looking at potentially adding appears in the form of (P4675) and adding these as values but need to think how to model that they were recorded as meeting with the devil and have appears in the form of (P4675) as a qualifier? And potentially add participant in (P1344) > witches meeting too. Working things out but let me know if you have any thoughts. Stinglehammer (talk) 20:13, 7 October 2020 (UTC)

• What items have such a devil type? Is that on a per-trial basis? When it comes to modeling content like this it's helpful to have examples of what you want to model. ChristianKl ❪✉❫ 14:34, 9 October 2020 (UTC)

It would be on items for accused witches e.g. Jonet Braidheid (Q43390291) and pertaining to the investigation into witchcraft for that accused witch. Discussing whether it should be considered as:

1. charge (P1595) of pact with the devil (Q1506690) and then qualified as to the type of demonic pact or
2. a qualifier of significant event (P793) witchcraft confession (Q66055048) or even
3. a qualifier to participant in (P1344) Witches' Sabbath (Q831942).

whichever way it is the property for adding a 'type of demonic pact' is perplexing me and 'type of devil' likewise. participant (P710) perhaps or appears in the form of (P4675) or has contributing factor (P1479) or even possessed by spirit (P4292)? Stinglehammer (talk) 17:30, 9 October 2020 (UTC)

• • In general "adding a 'type of demonic pact'" is not a good way to think about it. Basically it's a legal case where a person was accused of making a certain agreement. Then we can ask whether we have enough existing properties for this kind of legal case or whether we need additional properties but additional properties likely wouldn't be as narraw as saying anything about demonic pacts or devils. ChristianKl ❪✉❫ 18:53, 9 October 2020 (UTC)

Yeah, point taken. I wouldn't want to put forward something like that. Looking at existing properties and trying to discuss with a professor of history what might work best. charge (P1595) of pact with the devil (Q1506690) and then facet of (P1269) qualifying as to the type of pact? Or does facet of (P1269)facet of work the other way round? Stinglehammer (talk) 13:39, 10 October 2020 (UTC)

including (P1012) comes to my mind as a very general qualifier, e.g.

convicted of	pact with the devil edit including Anti-baptism 0 references add reference
		add value

.

I'm not sure about using charge (P1595) on persons (like at Jonet Braidheid); according to the property proposal it was meant as a property for trials, not persons. The according constraint was altered (https://www.wikidata.org/w/index.php?title=Property:P1595&diff=1188218134&oldid=1183865188), but I see no discussion that shows approval for this change. Probably this should be reverted but then I have no experience in modelling trials and trial-related statements so I would leave this to others. I reverted the change as it left the statement malformed and the Wikidata usage instructions state explicity that it should be used on the item for the trial. - Valentina.Anitnelav (talk) 15:44, 12 October 2020 (UTC)

I realized that there already exist items for each trial (e.g. trial of Jonet Braidheid (Q43394201)): this information should be modelled by putting the charge (P1595) statement on the trial item: trial of Jonet Braidheid (Q43394201)charge (P1595)pact with the devil (Q1506690)including (P1012)Anti-baptism - Valentina.Anitnelav (talk) 17:17, 12 October 2020 (UTC)

Wikidata weekly summary #437

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

• Discussions
  • Open request for adminship: MsynABot
  • Closed request for adminship: Fuzheado (successful)

• Events
  • Upcoming: live SPARQL queries on Twitch and in French by Vigneron, October 13 at 18:00 CEST
  • Upcoming: Wikidata birthday organizers call, October 14 at 12:00 GMT
  • Upcoming: Online Wikidata editathon in Swedish #34, October 18
  • Search Platform Office Hours happens first Wednesday of each month. Come with questions related to Wikimedia search, Wikidata Query Service, Wikimedia Commons Query Service, etc.

• Press, articles, blog posts, videos
  • Wikidata reaches Q100000000, by Lydia Pintscher and Denny Vrandečić
  • A new extension can add support for linking to local media files in Statements (and not just files on Wikimedia Commons) to a Wikibase instance (Demo video, developed by Professional.Wiki)
  • LD4 Wikidata Affinity Group presentation: How Entity Explosion fits into the Wikidata ecosystem, by Toby Hudson
  • Video: Wikidata and COVID-19: Creating a collaborative knowledge graph from CORD-19, by Csisc

• Tool of the week
  • OD2WD automatically converts CSV files from Open Data portals into QuickStatements for Wikidata republishing. Demo video

• Other Noteworthy Stuff
  • noratelimit for bots to be removed on October 20th
  • The new Wikibase/Consultants and Support Providers' directory on Meta lists people/companies that provide professional help with Wikibase -- be it technical support (installation and setup), custom development, or data modeling services.
  • The 2020 Coolest Tool Award is still accepting nominations until October 14th. Please use this form to recommend tools. You can nominate as many tools as you want by filling out the form multiple times.

• Did you know?
  • Newest properties:
    • General datatypes: number of negative votes, number of support votes, x-offset, y-offset, social media followers, fandom, has census, parallel street, entry in abbreviations table
    • External identifiers: Dictionary of Occupational Titles Code (fourth edition, revised), FVLB work ID, Passion Patrimoine ID, Biographical dictionary of Fourierism ID, CompositionToday composer ID, ITHL author ID, Quebec Dams Directory ID, SwissLipids identifier, ACER code, Science Museum Group ID, Slovník českých nakladatelství 1848-1949 publishing house ID, Directory of Belgian Photographers ID, RNACentral ID, Turkey's Culture Portal ID, NSW State Archives and Records Authority ID, Scottish Brick History Brick & Tileworks ID
  • New property proposals to review:
    • General datatypes: ship's captain, planetary coordinates, Date of signature, Unicode name
    • External identifiers: CPV Supplementary, PeoplePill person ID, O*NET OnLine code, Emojipedia ID, GADM ID, CTAN package ID, MinDat taxon ID, AniList anime ID, AniList manga and light novel ID, Normattiva ID, ILO Thesaurus ID
  • Query examples:
    • Map of trees whose species is categorized as threatened on the IUCN Red List - OSM/Wikidata query
    • Women actress less than 70 years old who are descendants of the first king of Navarre, ordered by number of sitelinks
    • Dead Social Democratic Party members who have been ministers in 1992 or later (Source)
    • Artworks with an image in Wikidata by year of creation (Source)
    • Map of mountains in Switzerland (Source)
    • Map of markets in the city of Surabaya (Source)
    • National Anthems by tonality (Source)

• Development
  • The RFC to semi-protect all property pages has been implemented, all properties now require the “autoconfirmed” right to edit (T254280)
  • Working on enabling JSON dumps for Lexemes (phab:T220883)
  • Finishing preventing blocked users from making us skip QIDs (phab:T232620)
  • Continuing work on the Wikibase Manifest extension that allows toolbuilders to easily access configurations of a specific Wikibase instance so it is easier for them to make their tools work with not just Wikidata but also other Wikibase instances
  • Started analyzing what we consider a mismatch for our work around checking Wikidata's data against other databases
  • Continued work on the Query Builder to get it to create the first very simple query. It can do that now but not yet visualize the result.

You can see all open tickets related to Wikidata here. If you want to help, you can also have a look at the tasks needing a volunteer.

• Monthly Tasks
  • Add labels, in your own language(s), for the new properties listed above.
  • Comment on property proposals: all open proposals
  • Contribute to a Showcase item.
  • Help translate or proofread the interface and documentation pages, in your own language!
  • Help merge identical items across Wikimedia projects.
  • Help write the next summary!

Read the full report · Unsubscribe · Mohammed Sadat (WMDE) 15:27, 12 October 2020 (UTC)

Remove noratelimit for bots

Hello all,

As you may know, over the past months we’ve been struggling with more and more bots editing Wikidata at a very high rate, causing infrastructure issues having an impact on the Query Service that couldn’t keep up with the changes and on tools such as Pywikibot.

Over the years, we tried different things (like Add Wikidata query service lag to Wikidata maxlag, increase maxlag or factor, limit the edit rate for all accounts). Wikidata admins also approached individual bot owners to ask them to comply with the bot policy’s limit, sometimes without success.

Recently, we discussed removing the noratelimit feature for the bots group. This would have as an effect to limit the edits to 90 edits/min for most of the bots ; the few bots that need an unlimited rate to function (for example MassMessage) can be added to the existing accountcreator group (with the possibility to rename this group).

Many thanks to bot owners who gave input in the comments of the ticket and helped us frame this solution. If you want to continue the discussion, please have a look at these comments, so we can build from them and avoid restarting the discussion from scratch.

We hope that this solution will allow a fair access to mass editing Wikidata, while preserving the existing infrastructure and avoiding hitting too hard on the Query Service. In the meantime we are working together with the Search Team at the WMF and others on improvements to the Query Service scalability and alternatives to it so some load can be redirected to other systems, as well as general infrastructure improvements.

Unless there’s a strong opposition from the community to this change, we will implement it on October 20th.

If you have any questions or need more information, feel free to add a comment below or in the ticket. Thanks, Lea Lacroix (WMDE) (talk) 11:44, 8 October 2020 (UTC)

Thanks Lea, it is nice to see that this issue is being addressed. Vojtěch Dostál (talk) 13:33, 8 October 2020 (UTC)

• I think I would prefer this not to be bundled with the account creator right but into a new right, so that it's more obvious to people that the right has something to do with the ability of a bot to go over the edit limit. ChristianKl ❪✉❫ 14:00, 8 October 2020 (UTC)
  • See c:Commons:Village_pump/Proposals/Archive/2018/01#Proposal_to_rename_account_creator_group_to_batch_uploaders.--GZWDer (talk) 16:55, 8 October 2020 (UTC)
• Currently we still have such issues: 1. some user groups (e.g. sysops and global rollbackers) have noratelimit and 2. QuickStatements does not use a rate limit at all.--GZWDer (talk) 17:01, 8 October 2020 (UTC)
  • @GZWDer: I believe QuickStatements, since it acts on behalf of a user, is limited by the rate limit on that user account. Of course if 100 users are using it at once that means it could operate at 100 x the individual user limit. Though I assume Magnus has some limit on total simultaneous batches (I think it was 16 but that may have changed?) ArthurPSmith (talk) 19:59, 8 October 2020 (UTC)
    • When 100 users are using QuickStatements it is one hundred users using the service.. IMHO they have precedence over unrestricted bots uploading data that is hardly any value and or not maintainable and or not scalable. We discussed earlier restriction on what datasets may be uploaded, I was even challenged to make a proposal and I did. Thanks, GerardM (talk) 05:08, 9 October 2020 (UTC)
      • Whether or not data is adding value or maintainable doesn't depend on whether it's uploaded through a bot account or QuickStatements. For bots at least there's a community decision that the bot edits add enough value to merit an approval. ChristianKl ❪✉❫ 16:55, 9 October 2020 (UTC)
    • Yes it is limited by rate limit, but it also means it will results in issues such as phab:T264450. In addition, users may sometimes not edit at sll if QuickStatements batches are running. QS should use a throttle that is strictly lower than 90/min.--GZWDer (talk) 20:04, 8 October 2020 (UTC)
  • Maybe edits should be throttled by the number of cleanup edits they require .. so if a bot required more than 1000 cleanup edits, it should be throttled to 10% of max. --- Jura 13:54, 9 October 2020 (UTC)
    • This is not a well-defined figure.--GZWDer (talk) 17:04, 9 October 2020 (UTC)
  • BTW, it works in a similar way for the description suggestion tool. There it's 3 reverts. --- Jura 06:06, 10 October 2020 (UTC)
• Are all edits equally hard on the server backend? would it be possible to consider metering at a level finer than "edit count"? 90 edits/min is probably not an issue for me though. BrokenSegue (talk) 20:15, 9 October 2020 (UTC)

@Lea For a long time regular bots (and it' owners) are second class citizens in the Wikidata universe. What kind of hardware measeures have been done to solve the problem? I'm only aware of restrictions implemented by software. --Succu (talk) 20:39, 9 October 2020 (UTC)

@BrokenSegue: edit count is the usual metric we use. Do you have other suggestions?

@Succu: a few years ago, Wikidata was moved to its own server. The Search Platform team at WMF is currently looking for solutions to improve the Query Service, implement a new updater, etc. But the broad issue (we have a huge database, very frequently edited and queried, with a need for immediately up-to-date results) is extremely complex and cannot be solved by throwing more hardware. For more information, see the updates by Guillaume Lederrey on the Wikidata mailing-list. Lea Lacroix (WMDE) (talk) 11:03, 12 October 2020 (UTC)

@Lea Lacroix (WMDE): I mean I don't know what's feasible but rate limiting weighted by edit size would maybe make sense. Or perhaps edits to some kinds of data are more taxing than others e.g. references v. names v. qualifiers. BrokenSegue (talk) 02:06, 13 October 2020 (UTC)

• On the good side, this is good kick for tool authors to optimize their edit count. Namely, QuickStatements can easily spam one editor's watchlist by editing a single property of a few elements (adding statement -> adding qualifier -> adding qualifier -> adding qualifier -> adding reference, adding statement -> adding qualifier -> adding qualifier -> adding qualifier -> adding reference, etc). --Lockal (talk) 11:22, 12 October 2020 (UTC)

  But bear in mind that tool authors are limited by what the API and the API wrappers they're using offer. The last time I used pywikibot (which seems to be by far the most common), I had to hack together my own function for creating edits in order to do something as obvious as including a reference when adding a statement (instead of adding the reference separately)... but most people are not going to go to the trouble of doing that when the provided functions (which create two edits) work fine. - Nikki (talk) 12:01, 13 October 2020 (UTC)

This should be fine for me, I might need to modify my bot scripts a bit but not a huge deal. I agree with Christian that the rate limit opt-out role should be a distinct role specifically meant for special-case bots. Out of curiosity, was any notification sent on the mailing lists for this change? I don't see one. Nicereddy (talk) 01:58, 13 October 2020 (UTC)

Oh, one thing. @Lea Lacroix (WMDE): just to clarify, this'll only impact edits, right? API calls (e.g. request info about claims or labels) are uneffected? Nicereddy (talk) 02:02, 13 October 2020 (UTC)

The read-only API calls should not be impacted.

The change was mentioned in the Weekly Summary this week, but thanks for the reminder, I will also send a dedicated email. Lea Lacroix (WMDE) (talk) 10:13, 13 October 2020 (UTC)

Item for theoretical concept

Help, please--what is the "instance of" when creating a new item for a theoretical concept? For instance, "split intransitivity" as the author of this article uses it to refer to a particular concept that is part of the theoretical model of language that he's working with. — Levana Taylor (talk) 15:17, 12 October 2020 (UTC)

You could use concept (Q151885) or one of its subclasses [8]. — The Erinaceous One 🦔 21:32, 12 October 2020 (UTC)

Thanks; I've created concept in linguistics (Q100320664) on the analogy of concept in physics (Q33104303) etc. — Levana Taylor (talk) 03:37, 13 October 2020 (UTC)

Feedback needed - REST API draft

Hi everyone,

We regularly get feedback that Wikidata (and Wikibase in general) needs an API that’s easier to understand and use in order to get more people to build tools that use data from Wikidata and other Wikibase instances. The existing action API has several problems. The biggest ones are that it’s not a widely used standard, not versioned and that it is not very well suited for Wikibase’ structured data (as opposed to MediaWiki’s usual wikitext). Over the last weeks we’ve looked at ways to improve the situation and have come up with a draft for a REST API for Wikibase. We’d love to have your feedback on it.

What we want to achieve with it

• Provide a more industry-standard and versioned way to access and manipulate data in Wikibase. This will make it easier for programmers to get started building tools with and around Wikidata and other Wikibase instances.
• Provide an API that is more tailored to the Wikibase data model to, for example, make it possible to get exactly the part of an Entity you need instead of the whole entity.
• Solve a number of issues in the current API that are easier to solve with REST.

A few things to keep in mind

• This is only touching the Wikibase-specific API modules, not any of the others that MediaWiki provides.
• We’ve started with the specification around Items and Properties. Once we are sure the direction is good, we will look at other parts of the data model and content like Lexemes and MediaInfo.
• For existing users of the action API: Nothing changes for you for now. If the feedback is positive, and we go ahead, it’ll take us some time to actually implement the proposed changes. It’d be very important for us to hear your feedback now to ensure that the new API meets your needs in the future.

If you are building tools around and on top of Wikidata/Wikibase, please have a look at the draft and give us your feedback. You can find all the information at Wikidata:REST API. Please leave your feedback on the talk page there, so we have it all in one place.

Thank you!

-Mohammed Sadat (WMDE) (talk) 10:16, 13 October 2020 (UTC)

Wikidata property to identify topics (Q52063969)

Would be correct to say that this item was only intended for online news websites? --Trade (talk) 11:53, 13 October 2020 (UTC)

female form of label (P2521)

Shouldn't that property be deleted and get moved to Lexemes ? Bouzinac ^{💬●✒️●💛} 12:20, 13 October 2020 (UTC)

See Wikidata:Requests for deletions/Archive/2019/Properties/1#female form of label (P2521). --Matěj Suchánek (talk) 12:33, 13 October 2020 (UTC)

Acting officeholders

When the former P794 ("as") was being deleted in early 2018, there was a discussion how to migrate all its existing uses, and one of the main cases was for "acting" or "interim" officeholders. It was agreed to migrate them to subject has role (P2868): acting (Q4676846) (though noted at the time that that's not really a role, and a separate item like 'acting officeholder' might be better).

At that time, nature of statement (P5102) did not yet exist, but as that was explicitly created to allow us to distinguish between things like de jure (Q132555) and de facto (Q712144), it has also come to be used for acting (Q4676846) as well.

My suspicion is that if nature of statement (P5102) had existed at the time of the P794 migration, it would have been the better option, and it would make more sense to (re-)migrate those the subject has role (P2868): acting (Q4676846) to that, but it's certainly possible that I'm misunderstanding what exactly the distinction between these is, and that it might be better to migrate the P5102 qualifiers instead. Either way, it would be good to standardise on one or the other.

(@Swpb, ArthurPSmith, Deryck_Chan, Liuxinyu970226: explicitly pinging people from the 2018 discussion, though of course all views welcome.) --Oravrattas (talk) 06:49, 10 October 2020 (UTC)

• Interesting question. I just noticed that one also uses object has role (P3831)
  Maybe the answer varies depending on the value used:
  Is it acting (Q4676846) or an item relevant to the position, e.g. acting prime minister (Q4676862)? For the later, Wikidata item of this property (P1629) seems a better fit.
  I wonder if nature of statement (P5102) is clearly distinct from sourcing circumstances (P1480). --- Jura 07:34, 10 October 2020 (UTC)

I do not understand why this is an issue.. You have office holders, they replace each other. The office holder is the one with a start date and no end date. It is an inference. I notice quite often statements for office holders, I remove them for being incorrect. The notion of temporary is problematic; either someone is trusted with the office or he is not. Thanks, GerardM (talk) 12:43, 10 October 2020 (UTC)

I'm not sure why the terms "permanent" and "temporary" are used for office holders at all: even if humans had unlimited life spans, it's unlikely somebody could stay in a position forever. Permanent may just mean that no end date has been specified, but many political positions do set an end date. Ghouston (talk) 22:13, 10 October 2020 (UTC)

This seems to be going off at quite an odd tangent. I don't quite understand how "permanent" arose here: this isn't about anything like that at all. This is about the concept of how we record that Alice Bloggs was the Acting Minister of Finance of Queensland in May 1998. We could, of course, explicitly create a parallel "Acting (whatever)" item for every position item, but generally that's not the path that has been taken. The usual approach is for the the position held (P39) item on the person to be for the main position, but with a qualifier of acting (Q4676846). The question here is what property should be used for that qualifier. subject has role (P2868) was chosen when this was last discussed, but that point nature of statement (P5102) did not yet exist. As both are now in use, we should standardise on one, document that, and migrate the others. P5102 seems better than P2868 to me, but I'd like some input from others on this. --Oravrattas (talk) 06:10, 11 October 2020 (UTC)

Maybe nature of statement (P5102) is marginally better, since the role is already Minister of Finance of Queensland, and "acting" is only a role modifier. Or perhaps that's inappropriate conflation of position held (P39) and subject has role (P2868). Maybe there are templates set up to use P2868 and it would be preferable not to change it when it doesn't matter much one way or the other. Ghouston (talk) 06:59, 11 October 2020 (UTC)

This isn't a case where everyone is doing X, and there's a suggestion to change to Y instead. Both X and Y are currently actively being used to say the same thing, in a way that's unhelpful for anyone wanting to query the data. The question is which we try to standardise on. --Oravrattas (talk) 08:44, 11 October 2020 (UTC)

object has role (P3831) that you mentioned would be used on the other side, like the pair on Canadian ambassador to the United States (Q32208) and Kirsten Hillman (Q88720948). It would be easy to get them the "wrong" way round. nature of statement (P5102) has the advantage that it's' a single property. Ghouston (talk) 08:18, 11 October 2020 (UTC)

"as" had that advantage too ;) --- Jura 08:21, 11 October 2020 (UTC)

I agree that that nature of statement (P5102):acting (Q4676846) is better than subject has role (P2868):acting (Q4676846) and object has role (P3831):acting (Q4676846). If there are no infobox uses of these qualifier-target combos I can try another mass-migration. Regarding sourcing circumstances (P1480) "Acting" definitely goes with nature of statement (P5102) but not sourcing circumstances (P1480). @Oravrattas, Jura1, GerardM, Ghouston: Thanks for including me in this discussion. Deryck Chan (talk) 20:33, 11 October 2020 (UTC)

@Deryck Chan: Thanks. I'm also happy to run the migration myself: that's much easier these days thanks to wikibase-cli's move-qualifier. --Oravrattas (talk) 05:54, 12 October 2020 (UTC)

@Oravrattas: You're welcome to use my Jupyter+pywikibot program for this task: Wikipage; github --Deryck Chan (talk) 14:55, 13 October 2020 (UTC)

Since nobody can stop me going off on another odd tangent, I've noticed that Winston Peters (Q1396178) has been acting prime minister of New Zealand on occasion, in place of Jacinda Ardern (Q3606816), although I haven't tracked down the exact dates. Does this mean that New Zealand had two prime ministers during these periods, one real and one acting, or does it mean that Ardern has had multiple terms of being PM, separated by intervals where Peters was PM? I also wonder if acting PMs affect the series ordinal (P1545) and replaces (P1365) qualifiers on the post. Ghouston (talk) 23:19, 11 October 2020 (UTC)

Perhaps it's not such an odd tangent, since the answer may shed some light on the nature of the process. I think the most plausible interpretation is that Ardern has remained the "real" PM, with a single term in office, the 40th PM of NZ, while the position held by Peters is only "acting PM". While I don't think it would be necessary to create a new item for this "acting PM" position, it implies that the subject has role (P2868) qualifier has been used the wrong way around, since it's not a modifier of the person, but of the job. Ghouston (talk) 23:28, 11 October 2020 (UTC)

Or it may be better to say that it's neither the person nor the job which modified, but the way in which it has been filled, on a acting basis. When a job is filled that way, we may want to treat it differently, e.g., when creating lists. Hence, a qualifier of the statement itself, nature of statement (P5102), seems best. Ghouston (talk) 00:06, 12 October 2020 (UTC)

@Ghouston: Not an odd tangent, but to get a proper answer I believe you need a Kiwi constitutional law scholar, not Wikidatans ;) In the absence of expert advice, I think both methods you described above are reasonable. Deryck Chan (talk) 14:55, 13 October 2020 (UTC)

The government has a list at [9], and acting PMs aren't included. Maybe they would include interim PMs, but I don't think the concept has been used in NZ. I think "acting PM" is basically seen a separate job, given that the PM job is already taken. Ghouston (talk) 21:26, 13 October 2020 (UTC)

When a officeholder is temporarily unavailable, and it is part of the function of another officeholder to stand in, it is different from there being an officeholder that holds the position temporarily in the absence of a duly appointed one. The last is relevant, the first not so much (for us to register). Thanks, GerardM (talk) 05:33, 12 October 2020 (UTC)

If there's a difference, wouldn't there need to be two different qualifiers to represent it? He was the official acting PM, sworn in etc., I think in one case for 6 weeks. Ghouston (talk) 05:38, 12 October 2020 (UTC)

When someone stands in as part of his function, he is sworn in for that function I would say. I only add holders as mentioned in the sources I have... Thanks, GerardM (talk) 10:11, 12 October 2020 (UTC)

"An Acting Prime Minister should be distinguished from a Caretaker Prime Minister, which refers to an Outgoing Prime Minister following an electoral defeat, and who by convention does not implement new policies or an Interim Prime Minister who is appointed to perform a similar role to a Caretaker Prime Minister, but who is typically not a Prime Minister at the time of being appointed", according to en:Acting prime minister

You can get somebody "acting" in a role even for a few days while the regular person is on holiday or travelling. I doubt that we'd want to record all those instances. Maybe it's only the "interim" positions that are of interest? Ghouston (talk) 23:42, 12 October 2020 (UTC)

Search results poor

A search for "North Korea" (without quotes), for example, does not return North Korea (Q423) in the first two pages of results.

The top ten results are:

1. Changbai–Hyesan International Bridge (Q5045259)
2. North Korea Uncovered (Q7055811)
3. North Korea flooding (Q16880525)
4. North Korea crisis (Q38107863)
5. Template:Country data North Korea (Q6649686)
6. North Korea Intellectuals Solidarity (Q3344134)
7. Liberty in North Korea (Q3237819)
8. healthcare in North Korea (Q64061621)
9. flora of North Korea (Q8563897)
10. Module:Location map/data/North Korea (Q29128351)

It's so bad, that I've had to resort to using the search on en.Wikipedia, then following the Wikidata link from there.

Is this a known issue? Can something (such as prioritising exact matches in labels, and then aliases; or the number of inbound links, or sister-links) be done about it?

@Lydia Pintscher (WMDE): ? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:06, 13 October 2020 (UTC)

Or how about just allowing us 'hide' the tens of millions of scholary articles. Most of us likely wouldn't have a need to use after them anyways. --Trade (talk) 12:14, 13 October 2020 (UTC)

I mean just look at this mess. --Trade (talk) 12:20, 13 October 2020 (UTC)

The search function seems to be negatively influenced by the "Property" and "Property talk" choices. When removed, it's the fourth result for me (not the first, though). --Matěj Suchánek (talk) 12:30, 13 October 2020 (UTC)

When I remove only the "property talk" choice, it's also the first result for me. Which also happens to be my current standard namespace selection - though I'm not sure what the default selection is. --Kam Solusar (talk) 13:26, 13 October 2020 (UTC)

Now on my computer, the top ten results are:

1. North Korea (Q423)
2. North Korea national football team (Q14132)
3. Pyongyang time (Q20757515)
4. North Korea (Q80218225)
5. North Korean calendar (Q3353159)
6. Korean People’s Army (Q240670)
7. UTC+08:30 (Q7023)
8. North Korea women's national football team (Q839723)
9. flag of North Korea (Q103064)
10. North Korean won (Q106720)

--Liuxinyu970226 (talk) 14:01, 13 October 2020 (UTC)

@Andy: Yeah the issue seems to indeed be the namespaces as Matěj suggested. ElasticSearch can't deal well with having to rank Wikibase Entities (Items, Properties) against pages that are wikitext. This is because wikitext pages don't have statement counts for example so it's hard for it to determine what should be higher etc. The default (when searching via the search box at the top right and then pressing enter to get to Special:Search) is not to have the talk namespace added for all I can tell so the default should be better than what you are seeing.

@Trade: They are already deranked :( We could consider dropping them even more. What you can also do is use a trick like this to exclude them: https://www.wikidata.org/w/index.php?search=%22weather+forecasting%22++-haswbstatement%3AP31%3DQ13442814&title=Special:Search&profile=advanced&fulltext=1&advancedSearch-current=%7B%7D&ns0=1

--Lydia Pintscher (WMDE) (talk) 16:00, 13 October 2020 (UTC)

Instances of scholarly article (Q13442814), academic journal article (Q18918145) and disambiguation pages are, but there are others that could be; the "top ten results" above includes two internal items (a template and a module). Peter James (talk) 22:23, 13 October 2020 (UTC)

The top result has several aliases such as "China North Korea border crossing #2094" from merged items, but some of the items that were merged were not at the same location. It's also unclear where the names are from - in the CSV version of the source, four entries have the same name ("302" - the number of a road or railway line?) and same coordinates, but there are no other names for the China-North Korea crossings. There is a column with "border_crossings_phv.2155" and similar, but the last four digits are all in the range 2155-2214, not 2094, which is a crossing between Spain and France. Peter James (talk) 17:23, 13 October 2020 (UTC) Most of the merged "crossings" didn't exist, the source had inaccurate borders. I removed the aliases. Without the multiple "North Korea" names, it will probably be lower in search results, if it's there at all (the only remaining mentions of the country there are in the labels of other items in statements). Peter James (talk) 18:26, 13 October 2020 (UTC)

Two people mixed in Tony Phillips Q21461790

This item seems to be about the UK painter Tony Phillips, born 1952. However, several of the identifiers are to a different Tony Phillips (New Zealander Anthony Alan Phillips) and a book he wrote (Always ready : Christchurch Fire Brigade : 1860-2010 /). What is the best way to disentangle these two? --MerielGJones (talk) 15:10, 13 October 2020 (UTC)

The wrong statements can be deprecated with reason for deprecated rank (P2241) applies to other person (Q35773207), and optionally create an item for the other person if one doesn't already exist. Ghouston (talk) 21:05, 13 October 2020 (UTC)

I think in this case the wrong VIAF id has been added, and bots have copied various related identifiers. Ghouston (talk) 21:10, 13 October 2020 (UTC)

@MerielGJones: Create a new Item and use User:Melderick/moveClaims2.js to quickly move claims to the new item. --Haansn08 (talk) 21:18, 13 October 2020 (UTC)

OK, thank you both for your suggestions @Haansn08: and @Ghouston:. I will make an attempt to disentangle the people.--MerielGJones (talk) 21:51, 13 October 2020 (UTC)

The "requirement" of references on individual awards

Many items are created because someone by that name received a particular award. The objective is that we have a full list of awardees at Wikidata, the source is a black or red link in a Wikipedia. When there is no article for that award, there is a website for that award where this person is mentioned.

What I find is that a newly instatement for "requirment of a reference" is abused by people to delete these items. They are not interested in the completeness of lists, they have the Wikipedia attitude towards notability and for them no reference is good enough.

I am completely aware that this is an unintended consequence of what is in and of itself not a bad idea. But we have to deal with the consequences and for me a these consequences undo the integrity that is one of the reasons for notability of an item. We cannot change people and their attitude but we can remove this requirement. Thanks, GerardM (talk) 06:06, 8 October 2020 (UTC)

• I don't see why it should be hard to find a reference if the person is in fact awarded a certain award. Without references establishing data quality is hard. ChristianKl ❪✉❫ 12:11, 8 October 2020 (UTC)
  • It is hard because that is not what I do. What I do is include the data of lists from a Wikipedia, a website of an award to Wikidata. There are references, they are the website for the award, the article for an award on a Wikipedia. When an award DOES include sources like the AAS does for its fellows, it is easy to include that data. There is a point because it serves at the same time as a source to write an article. The argument is not about it being difficult to find sources, it is about the unfortunate tendency of some to delete. Thanks, GerardM (talk) 19:31, 8 October 2020 (UTC)
    • If the reference that you actually use is a particular Wikipedia article you could use that Wikipedia article as reference. Wikimedia references aren't ideal but they are better then no references. ChristianKl ❪✉❫ 17:50, 9 October 2020 (UTC)
      • There are whole categories of personal information that do not carry that same requirement and are wrong on a massive scale. Particular nationality. It is assumed that when you teach in a US university, you are USAmerican. This when it is stated that they are born anywhere else in the world.

When you then consider what Wiki work is, a requirement that is used as an excuse for deletion, makes a level 0 quality acceptable. When a container has its reference in links to Wikipedia or websites, it should suffice. In addition a lot of the references to a Wikipedia are actually wrong. Thanks, GerardM (talk) 05:36, 10 October 2020 (UTC)

Wrong references are still helpful because they allow a reader to check the reference and come to the conclusion that a statement isn't supported and thus can be removed. ChristianKl ❪✉❫ 15:44, 10 October 2020 (UTC)

You wilfully do not address the issue. When awards are notable, the integrity of its recipients is what we address. It is not about references; having them is better. It is denying the validity of completeness in awards as an argument NOT to delete. When an award recipient is deleted, at the same time you pronounce the award as non notable.

The notion of a wrong reference as acceptable is unconscionable. You know as well as I do that people hardly ever validate references. Thanks, GerardM (talk) 05:00, 14 October 2020 (UTC)

I’m still not sure what the problem seems to be. Is it newly created items where we only have the names of the recipients? --Emu (talk) 09:08, 14 October 2020 (UTC)

The problem is that a person is deemed to be non notable, is deleted but is the recipient of an award. That should establish notability. There is no such thing as "only for new items"; it is seen as non compliance and consequently it is deemed to be irrelevant. Thanks, GerardM (talk) 09:56, 14 October 2020 (UTC)

• Maybe @Trade: wants to explain [10]. --- Jura 06:10, 10 October 2020 (UTC)

Broadening the scope of P2403 (total assets)

I was thinking of broadening the scope of total assets (P2403) to allow it to be used in entries for politicians. This information is often freely available on the web (at least for Brazilian politicians), and could help with transparency efforts. I think a whole new property just for politicians would be a waste, but I'd like to hear what other editors think about this. NMaia (talk) 08:09, 11 October 2020 (UTC)

• I think the "entity" wording in the property is broad enough that it makes sense to also use this for humans. ChristianKl ❪✉❫ 11:59, 11 October 2020 (UTC)
• Would net worth (P2218) be appropriate here? It seems to be intended for people. Andrew Gray (talk) 14:57, 11 October 2020 (UTC)
  • Yeah, I think that would be a sensible approach. Thanks. NMaia (talk) 13:12, 14 October 2020 (UTC)

A process to collate knowledge on a topic with a group of experts

Hi all

I'd like to share with you and ask for your help in something I've been working on for almost a year. I've developed a process to collate knowledge on a topic with a group of experts to help map the topic area on Wikimedia to know what's missing (redlists) and to build a database for the experts to use and contribute to.

I ran the workshop in the before times with experts on sexuality education have and written up the process, the next steps are Mix n' Matching the data into Wikidata and coming up with different ways to display and use the data ways that are useful for Wikimedia and for the experts who contributed the data.

I'd really appreciate any thoughts on uses for the data and of course some help Mix n' Matching

• Wikidata:Switched_On:_Working_with_experts_to_collate_a_worldwide_database_of_sexuality_topics

Thanks very much

--John Cummings (talk) 14:03, 14 October 2020 (UTC)

All the streets in the world are notable?

Hoi, I looked for papers for the concept "swale" only to find that all the streets and houses of Swale, Kent are included in Wikidata. Not a good idea because by inference all the streets and houses in the world become eligible as a result. Who is going to maintain that? What is the objective, why do we even think it is a good idea when it we only do this only for Britain?? Thanks, GerardM (talk) 13:18, 11 October 2020 (UTC)

• Items such as 72, High Street (Q26627960) all seem to be included because they are historically significant listed buildings; it seems that Swale (a district with four towns) happens to have a large number of historic sites, and as a lot of them don't have individual names they are listed under their street number. I can't see any evidence that "all the streets and houses" are included or that anyone has ever suggested doing so. I don't think there's anything to worry about here. Andrew Gray (talk) 13:54, 11 October 2020 (UTC)

• Do we still have all streets in the NL? --- Jura 14:03, 11 October 2020 (UTC)
  • I do not know nor do I think it is a good idea to have them all. It does not scale to the world. Thanks, GerardM (talk) 14:36, 11 October 2020 (UTC)
    • I think we did at some point (maybe 5 years back). The question is if it's still up-to-date. --- Jura 14:46, 11 October 2020 (UTC)
      • It looks like they are still there - of the 408986 instances of street (Q79007) worldwide, 230272 are in the Netherlands. There are more items with National Heritage List for England number (P1216) (399135, out of 1925392 items with heritage designation (P1435)). Although England, Scotland and Wales have a lot of sites with national recognition, this type of data is not only added for Britain - there are 63989 items with Rijksmonument ID (P359), for example. One reason for the number of historic sites is Wiki Loves Monuments (Q1353202); although most are unlikely to be the subject of Wikipedia articles, they are (or will eventually be) included in lists. Peter James (talk) 15:48, 11 October 2020 (UTC)
        • Well, new streets can get built. Ideally we would also have all name changes. BTW, last month we got most (if not all) streets in Monaco. --- Jura 15:55, 11 October 2020 (UTC)

In general, yes, all streets are notable per WD:N #2: “It refers to an instance of a clearly identifiable conceptual or material entity. The entity must be notable, in the sense that it can be described using serious and publicly available references.” The first condition might be problematic in some cases. In some parts of the world, the second condition might be a challenge. --Emu (talk) 15:41, 11 October 2020 (UTC)

"all the streets and houses of Swale, Kent are included in Wikidata". No. That's a complete misrepresentation. Swale is a local government district with borough status in Kent, with an area of 144.2 sq mi (373.4 km2) and a population > 150,000. Wikidata has 141 items for listed buildings in the borough. There's little point in having a discussion if it starts from such a fatuous & bogus premise. Number of streets in Swale on wikidata? Zero. Number of roads in Swale on wikidata? Zero. --Tagishsimon (talk) 22:29, 11 October 2020 (UTC)

• I was too wondering about if it was a good idea to import french database // or on streets. Almost all french streets have a FANTOIR code (P3182) and are nationally referenced. That said... I did not go further on the idea because there is too many streets, too much work to reconcile between already existing streets inside WD and the would-be new ones. :) Bouzinac ^{💬●✒️●💛} 19:15, 14 October 2020 (UTC)

Q92526

Cannot add plwiki. Page locked. Help please. 83.23.136.140 16:21, 14 October 2020 (UTC)

 Done --HarryNº2 (talk) 16:52, 14 October 2020 (UTC)

first criticality (Q100349048)

I just made this item because the validation checker was telling my I could not put criticality (Q17008131) as a significant event (P793). But I am a bit confused by the "subclass of" stuff. Would be nice if somebody experienced could check it and its use in Belarusian nuclear power plant (Q2639309). Thanks. --Ysangkok (talk) 16:45, 14 October 2020 (UTC)

John McCain: senator * 10

Last time we discussed this, the preference of some seems to be the use multiple statements with position held (P39) = United States senator (Q4416090) on its item e.g. at [11], see Wikidata:Project_chat/Archive/2018/12#John_McCain_(Q10390)_and_position_held_(P39). While I still think this is useless, at least it had the advantage of being consistent for that position (US senator).

In the meantime, a newbie user changed one of these statements to use Q98077491 instead. See Q10853588#P39. I don't think this was discussed or proposed anywhere prior to these edits. The clean solution seems to be to delete Q98077491 and restore Q4416090 there. This was proposed at Wikidata:Requests_for_deletions/Archive/2020/08/20#Q98077491, but after the debate carefully avoided to address the issue with Q98077491, admin DannyS712 (talk • contribs • logs) closed the discussed on RfD and suggested to discuss the deletion first elsewhere. Thus this post. I will leave a note on the relevant WikiProject.

An explanation at Wikidata:Project_chat/Archive/2020/08#Special_subclass_with_"identity_of_subject_in_context"_(P4649)_qualifier concludes that the qualifier identity of subject in context (P4649) isn't used correctly on Q98082336#P31, so Q98082336 is malformed anyways. To clean this up, I suggest we restore Kamala Harris etc, to the John McCain format and then delete Q98082336 and (any similar malformed and unused items). --- Jura 04:44, 2 October 2020 (UTC)

Usually we use such specific items as values of parliamentary term (P2937), which is in turn a qualifier.--GZWDer (talk) 05:06, 2 October 2020 (UTC)

I think that would be 116th United States Congress (Q28227688), not Q98077491.

Maybe a simpler way would be to use parliamentary term (P2937) directly on items .. so position statements remain manageable. --- Jura 05:09, 2 October 2020 (UTC)

The practical reason for preferring the route chosen was that Wikidata lists can't filter on qualifiers. A secondary consideration is described below, it allows a clear separation of the models being used so that code based on the older model is uneffected by the more complete time series model. Gettinwikiwidit (talk) 07:05, 11 October 2020 (UTC)

@Gettinwikiwidit: are you still planning to finish making all US Senator P39s consistent? If so, do you have an expected timeframe or list of steps to take, or can you provide any details on anything that's blocking that? I agree that we should remove the identity of subject in context (P4649) qualifiers on the instance of (P31) statements, but I don't believe that those statements being poorly modelled in that regard means that the items as a whole aren't valid (especially considering the circumstances under which those qualifiers were added.) It is already well established that such term-specific items can be useful, and if someone (whether Gettinwikiwidit or someone else) is using these as a way to get complete and consistent information on all historic Senators, then continually erecting roadblocks to that seems less than helpful. If this has stalled, however, and no else is going to finish off this process, then I agree that reverting back to a position held (P39):United States senator (Q4416090) approach is probably best. --Oravrattas (talk) 05:19, 2 October 2020 (UTC)

It's not welcome to do such changes in uncoordinated way. --- Jura 05:24, 2 October 2020 (UTC)

It's not welcome to harass and harangue people who are trying to improve the data. Please be more kind and welcoming. --Oravrattas (talk) 05:27, 2 October 2020 (UTC)

I don't think your comments, advice to the new editor, self-repetitions and accusations in this matter are helpful. It's still a mistery to me why WMF paid your then employer to improve the data model which lead to a result that doesn't seem to be working. --- Jura 05:34, 2 October 2020 (UTC)

@Jura1:, it's not clear what you mean by "coordinated". You've never made clear your interest or your concerns, so in my mind you're not interested in contributing meaningfully to this project. @Oravrattas:, I was planning on finishing this up. I had been in contact with the Assistant Senate Historian to clarify some discrepancies between their data and what was already in Wikipedia and had been waiting on a reply. I'm now inclined to simply upload the data in the state it is in and simply edit it later pending that discussion. As I've mentioned here and elsewhere I'm happy to discuss my work. The data and a short description of it can be found here: wikidata project Gettinwikiwidit (talk) 20:14, 2 October 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘

• Uncoordinated, means that this wasn't discussed or proposed onwiki to Wikidata contributors before. Furthermore, it explicitly breaks the existing, previously discussed structure for this position. Do not expect people to read posts on other websites.

Please re-read my cleanup proposal above. If there are any specific points that aren't clear to you, please say so explicitly (e.g. please help me correctly format labels, please help me input this reference for that statement, despite reading Help:Label and Help:Sources, I didn't manage to do so in this or that edit). --- Jura 09:21, 2 October 2020 (UTC)

@Jura1: Again, you have not engaged with the discussion of the data at all beyond seeking to rollback changes. Focusing on a "clean up" is more of the same. The data will be consistent and complete once my project is complete and the advice will be irrelevant at that point. If you want to discuss what the data should look like then please frame the discussion that way. Gettinwikiwidit (talk) 06:40, 3 October 2020 (UTC)

• @Gettinwikiwidit: It's great that you're going to continue with this. I agree that it would probably be best to go ahead with getting the data as complete as possible for now, and then make any corrections later based on comments from the Assistant Senate Historian. Having the P39 data for all Senators entered in a consistent manner should be the primary goal. Any cleanup of that should relatively straightforward to automate later if there is consensus for that. --Oravrattas (talk) 13:18, 2 October 2020 (UTC)

• I don't see any concrete proposal or consensus for a change. Accordingly, I think we should move ahead with the cleanup as outlined. --- Jura 06:24, 3 October 2020 (UTC)

• • @Jura1: this is clear-cut bullying and harassment of a new user who is trying to improve our data substantially. You have no consensus whatsoever for your approach. Please stop. --Oravrattas (talk) 07:18, 3 October 2020 (UTC)
    • Please refrain from your accusations. Your activity here has already brought sufficient problems and lead the foundation to spend money something that hasn't worked out. If you can't help the new user outline how they should proceed, you are not helping. We need to move ahead and clean this up. --- Jura 07:27, 3 October 2020 (UTC)

• It seems that the recommended proposal hasn't been brought forward. Accordingly, I don't see another option than clean up as proposed. --- Jura 18:17, 10 October 2020 (UTC)

I don't understand the urgency here. Can you please explain? Gettinwikiwidit (talk) 07:39, 11 October 2020 (UTC)

• • @Gettinwikiwidit: any updates? --Oravrattas (talk) 20:55, 10 October 2020 (UTC)

I see a few options:

1. Choose between either (multiple position held (P39)United States senator (Q4416090) with start-end, parliamentary term, and whatever other qualifiers, as at John McCain P39) or (position held (P39)Q98077491, position held (P39)Q98081012, etc. without parliamentary term but still with all the other qualifiers, as at Kamala Harris P39).
2. Use position held (P39) with both United States senator (Q4416090) and Q98077491.
3. Replace the position Q98077491 with a group / collection ⟨senators of the 116th US Congress⟩, and make the senators member of (P463) this group of humans (Q16334295). Keep ⟨United States senator (Q4416090) with qualifiers⟩ as the positions held.
4. Use 116th United States Congress (Q28227688) to do double duty as a legislative term and a group of humans, and make the senators members of that (which means it would encompass both reps and senators).
5. [added late, after re-reading discussion above] Jura's suggestion is intriguing. That would have, for example Kamala Harris (Q10853588)parliamentary term (P2937)116th United States Congress (Q28227688)some qualifierUnited States senator (Q4416090). Would you need the qualifier to distinguish senators from house rep's?

Option 1 (either-or) means those who support one approach or the other will end up unhappy. Option 2 (both positions) feels dirty: we're representing the same “thing” (position held) in two different ways. Though is there any principle that says this is not okay? Doubling the positions held for someone like McCain would be very unwieldy. Option 4 (double duty) could work, but it may also create more problems than it solves. (There is a general question of using multiple instance of (P31) to achieve in Wikidata what object-oriented programmers would call multiple inheritance, but that's beyond the scope of this discussion.) Option 5 would allow contiguous terms to be collapsed into a single position-held, at the expense of splitting information between position-held and parliamentary-term. That probably deserves further examination. Option 3 (change position to group) allows data consumers to easily get, say, “all senators of 116th Congress” without untangling qualifiers, and doesn’t interfere with existing (or future) approaches to modelling position-held. What do you all think of #3? (As I write on it, I'm developing the sense that #3 seems preferable to 1,2,4. Is there a down-side that I’m missing? Would it be an acceptable compromise?) Pelagic (talk) 02:00, 11 October 2020 (UTC)

The problem I see with any solution which relies on splitting statements based on the particular Congresses in which they served (and do correct me if the point I'm bringing up has in fact been brought up before) is that a senator's term is six years in length, meaning that someone elected in 2014 (and sworn in on January 3rd, 2015) does not have to re-swear in on January 3rd, 2017 or January 3rd, 2019 (since they were not up for re-election in 2016 and 2018, and thus absent other intervening factors there is direct continuity over that six-year period which does not exist for representatives in the House). If we are to split up P39 "United States senator" based on any particular temporal criterion, then the most natural one (and the one which corresponds directly to how the Senate actually functions) is to use, rather than individual two-year-length Congresses, items which combine different classes of Senators and the Congresses which under normal circumstances their terms of office would begin.

For example, Dick Durbin (Q434804) could have P39 "United States Senator (Class 2, starting at the 105th United States Congress)" and similar values for the 108th, 111th, and 114th United States Congresses. Note that these would not be the same as the respective items for the Congresses which begin the six-year senatorial terms that the proposed P39 values represent!. As another example, Johnny Isakson (Q130024) and Kelly Loeffler (Q76570207) would both have P39 "United States Senator (Class 3, starting at the 115th United States Congress)" with different start time (P580)/end time (P582) qualifiers under this scheme. Mahir256 (talk) 03:00, 11 October 2020 (UTC)

@Pelagic: @Mahir256: I advocate for option 2 listed above. Effectively we're not representing the same thing two different ways, but rather capturing two different aspects of the same occurrence. One covers aspects of the fact that someone held a position and the other covers the actual time series of events. A position is not held or not in the absolute, but held relative to a given time. Thus position held (P39) will always fall short to represent the truth. You will always need multiple statements because holding or the seat or not will change over time and indeed there have been senators who have held non-consecutive terms. However, the argument above seems to presuppose that the old method of using position held (P39) has value. I have no reason to argue that and will leave it to others to layout its merits. Therefore I suggest leaving both in. It's easy to distinguish between the two in queries and thus there's no reason for one model to interfere with the other.

I will say, that I don't understand the practical benefits of the other suggestions. Ideally we'd be most concerned about how the data is practically used. We are "modeling" data, not defining it. A model is only as good as it's uses. One argument for the position held (P39)Q98077491 model is that it's what is used by the government website itself. For example, search for John McCain here: https://bioguideretro.congress.gov/. That also points out another benefit whose merits or demerits can be discussed. It's easy to import the data from authoritative sources.

As for the "most natural" subdivision, it seems to me that this is also dependent on the imagined uses of this data. Having worked with time series before, it's generally better to have a common quantization unit across heterogeneous data. I.e. it's better two work with a least common divisor than to try to compare compound numbers. This allows you to more easily ask questions like "what was the composition of the Senate at time X?" which to my mind are far more likely to be asked then who was the fifteenth senator to have served in the Senate to represent the Class III seat from Alabama. The ordinal relative to the six year term is less relevant and makes it hard to ask questions across senate classes. I was imagining using start time as suggested above. Two senators sitting in the Senate at the same time weren't necessarily elected at the same time. I was also thinking of representing the class the senator served in as well. Using either this or start time you could ask questions like "Who is up for election this year?" pretty easily.

If people care to browse the actual data, I have it here: https://github.com/gitonthescene/wikidata/tree/master/every_politician/United_States_of_America/Senate. Please let me know if you have questions. This is the complete history of Senate seats held. Currently the data available in wikidata is pretty spotty.

Gettinwikiwidit (talk) 04:08, 11 October 2020 (UTC)

• The main problem with using Q98077491 is that actually a database artifact (it only lives at Wikidata and isn't even correctly formatted; despite numerous requests, no reference supporting its claims were added to the item). As such, I don't think it should be used in any solution. We don't want to end up with additional Q5 items of "Albert Einstein (physics)", etc, merely because it suits some discontinued database re-user. --- Jura 06:30, 11 October 2020 (UTC)

How does where the data comes from affect whether the model can be usefully employed? Your comments continue to be geared toward the people and methods of adding the data rather than the usefulness of the data itself. This doesn't seem productive. Gettinwikiwidit (talk) 07:00, 11 October 2020 (UTC)

The question is what fact you want to model. An item for a person "Albert Einstein (physics)" in addition to another item for a person "Albert Einstein" may well be useful for some application. It just doesn't fit into Wikidata. Besides, there are not two such persons. --- Jura 07:20, 11 October 2020 (UTC)

I don't think anyone but you questions what we're modeling here. I'll press on. Gettinwikiwidit (talk) 09:50, 11 October 2020 (UTC)

• This leaves us essentially with the question of using one or several statements with United States senator (Q4416090). Personally, I'd prefer one, but it seems that people favor several for this position. Even though start and end dates allow to compute which parliamentary terms people participated, option #5 above would add them with P2937 as main statements. --- Jura 06:30, 11 October 2020 (UTC)

I think there are a couple of distinct questions here that are being conflated. In general how we model the memberships of national legislatures is fairly well established by now, and documented at Wikidata:WikiProject every politician/Political data model. We should be very hesitant about introducing an entirely new model for a country in a way that the large numbers of queries, reports, tools, etc that already know how to work with that model won't understand. The issue is that the model is not entirely uniform: there are several different choices that can be made within it. We currently have quite poor historic data for the US Senate, and an editor who wants to fix that. My opinion is that, so long as they are following one of the established models, they should simply be allowed to get on and do that. Once the data is complete, and consistently modelled, it will be relatively simple for us to migrate it to a different model if we decide that that would be better. The model that Gettinwikiwidit wants to use here probably isn't the one that I would use, but I don't see that as a big problem. If they find it easier to get the data imported using that model, then we're in a much better position than we currently are, which to me is much more important.

The data model Gettinwikiwidit plans to use is essentially identical to the model used by Wikidata:WikiProject British Politicians, which I believe is is by far the most impressive example we have yet of historic data across any country. I don't want to speak for Andrew Gray who is responsible for the vast majority of that amazing work, but my understanding is that introducing this model made that work substantially easier, even if it's not necessarily the best long term approach, and that at some point it might be migrated to a slightly different version. It has also been chosen as the preferred model by groups working on entering such data in about half a dozen other countries, for, I believe, similar reasons.

This is not a case of two people both wanting to enter all the historic US Senate data, but not being able to agree on which model to use. This is one person wanting to enter the data, and other people arguing about what way they want the work to be done instead. This is hugely un-welcoming to a new editor who is trying to help out. Gettinwikiwidit should be allowed to enter the data according to a documented model, not have constant blocks put in the way of that. This seems to be firmly in Perfect is the enemy of good territory. "Better a diamond with a flaw than a pebble without", etc. --Oravrattas (talk) 06:44, 11 October 2020 (UTC)

The problem is that only only "United States Senator" is a position. The other is decidedly not. It is fundamentally wrong, just explain how depending on what time period, a senator is not a senator? Thanks, GerardM (talk) 11:25, 11 October 2020 (UTC)

I have not followed this entire debate closely, so please pardon me if this is an irrelevant or already-discarded alternative, but instead of problematic sttements like

• position held (P39)Q98077491

would it achieve essentially the same result in a more palatable way to say

• member of (P463)116th United States Congress (Q28227688)

--Scs (talk) 13:00, 11 October 2020 (UTC)

@Scs: that part is already well established. In some countries we used to model this using member of (P463), but it was decided about 5 years ago now that position held (P39) was better, and have standardised on that across hundreds of legislatures. The debate here isn't about position held (P39) or not, it's about Approach 4 vs Approach 5 of the documented data model, both of which are in widespread use, and are entirely compatible in terms of how to query etc. But more importantly than being about choosing one, it's also about who gets to decide. If I were importing this data, I'd probably go for Approach 4, but I'm not the one doing that work — Gettinwikiwidit is — and all this arguing is just getting in the way of us actually having significantly better data than we currently do. If Gettinwikiwidit is going to do the work of creating a full history of the US Senate, and is going to do that according to already well-documented model, that's something we should be celebrating and encouraging, not nitpicking it to death. --Oravrattas (talk) 13:28, 11 October 2020 (UTC)

• @Scs: sounds like a valid approach. The approach with parliamentary term (P2937) as main statement could approach that, it could be combined with the main datamodel (a single senator statement).
  BTW please note that I have been asked not to comment on Oravrattas/Gettinwikiwidit accusations, so please note that I wont do that. --- Jura 13:42, 11 October 2020 (UTC)

I support us getting this data added, and I am very disappointed that Jura has started trying to interfere with good-faith contributions from someone who is trying to improve it. @Jura1:, please remember this discussion in 2019 which led to editing restrictions being placed on you over political data - it is clear that the community feels this sort of behaviour is not acceptable. Everyone knows by now that you dislike the existing political data model, but challenging it every single time someone starts improving data, opening up pointless deletion requests, and claiming it's "malformed" or "useless" isn't appropriate.

In terms of the specific suggestions, above, I agree entirely with Oravrattas. I think options #3 (P463), #4 (members linked from term item) and #5 (P2937) are probably unsuitable in that they would create a whole new data model just for US Senators - we have a huge amount of political data in place using P39 based claims. Option #2 (overlapping P39) is confusing as, again, we don't really do this for any other country or parliament. The fact that it exists on some items is an artifact of this data having not been properly and systematically worked on. I feel that if we have overlapping claims like this it will just confuse reusers of the data.

In terms of option #1, I have no strong feelings on using the claim based on "senator" versus "senator (115th)" etc. For the UK we went with numbered terms but this is in part because of the complexity of historical data. This is less of a problem for the US, plus they have the complication of one elected six-year term covering three distinct parliamentary terms which might make numbered items by term less appropriate. I would be happy with whatever method is seen as most appropriate by the subject experts, and happy to leave figuring that out to the person who is actually working on it and contributing data - Gettinwikiwidit. They are clearly in touch with the experts and have a plan to get this completed, and I'm very happy to see that. Andrew Gray (talk) 13:44, 11 October 2020 (UTC)

• @Andrew Gray: maybe you could explain us how the label of "United States senator (116th Congress)" and its P31-qualifier P4649 is well formed in terms of Help:Label and Property:P4649? I will refrain to comment on your accusations. --- Jura 13:58, 11 October 2020 (UTC)
• @Andrew Gray: is it correct that you usually add data in the overlapping format presented above as #2? --- Jura 14:20, 11 October 2020 (UTC)

Unless I have misunderstood #2, it seems to suggest having one P39:senator:1980-2004, and additional P39s covering 1980-2, 1982-4, etc (or 1980-86, 1986-92, etc). I would avoid intentionally keeping data with overlapping date qualifiers unless there was a specific reason for it (eg someone holding two seats simultaneously).

There is a bit of duplicated UK data out there, though - some pre-1832 items still have an old-style Member of Parliament of Great Britain (Q18015642) (with no date qualifiers) and a new-style member of the 18th Parliament of Great Britain (Q94911433) (with date qualifers), but the plan is for the Member of Parliament of Great Britain (Q18015642) claim to be removed to be consistent with the overall data model. I have had to focus on other work recently so didn't get around to doing this after the last big import - now that you've reminded me I'll put a script together tonight to clean those up and avoid any confusion. Andrew Gray (talk) 14:54, 11 October 2020 (UTC)

You might want to re-read what you commented on and check what you are contributing yourself. Personally, I find it highly confusing that you write one thing and do another. It's really disappointing, especially as some might have been led to believe that your model was a well-formed, stable model to follow. --- Jura 15:06, 11 October 2020 (UTC)

I do not claim to be perfect. The model is well-formed and stable when the data is complete. It cannot all be added instantly, and for the period we are discussing I have only recently started importing this material on a systematic basis, there is a huge amount of manual processing involved, so of course there will be gaps and imperfections. Complaining that imports are not perfect overnight as meaning "you write one thing and do another" is just silly and offensive. Andrew Gray (talk) 15:54, 11 October 2020 (UTC)

I don't recall commenting on your imports. --- Jura 15:58, 11 October 2020 (UTC)

I remain frustrated that this conversation is not centered on the usefulness of one model or another, but always drifts back to one person or another's gut instinct about what is "right". Ideally, we'd be more focused on what sorts of queries are possible or not as well as what sort of queries are more likely or not. Having a good feeling or not isn't, at the end of the day, what's most important, least of all from people who haven't demonstrated any need to access the data. I chose the model I did because it seemed to fit existing data such as it was at the time and because it fit into a framework which presented the data in a useful manner. The choice of Q98077491 was driven by the fact that Wikidata list templates don't currently filter on statement qualifiers. I.e. it was a practical choice taken to make something happen. The spotty state and inconsistent state of the data currently has been this way for two years since the last person tried fleshing it out. There were complaints then, but they weren't followed up with action and thus left the data spotty for two years. I don't understand the point of arguing for a model being "right" when it's not followed up with action to realize it. Moreover, as Oravrattas mentions above, *any* well structured data can be *re-organized*. Gettinwikiwidit (talk) 20:35, 11 October 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ @Gettinwikiwidit: For the avoidance of doubt, I am happy to support the structure you're using, and very glad you are working on it :-). As you say, we can always change things around later if appropriate, once the data's organise and imported - but that's the critical bit. Andrew Gray (talk) 21:22, 11 October 2020 (UTC)

Since apparently some list template (which one?) is the reason why the Q98077491 model is being targeted for, can we please quickly have a look whether there are other templates or data users relying on the status quo using United States senator (Q4416090)? It might be easier to fix the problematic template. Technically I still cannot see any compelling advantage of either model over the other, but the status quo seems much more used as far as I can see; the only noticable example of Gettinwikiwidit's model are UK members of parliament apparently. —MisterSynergy (talk) 21:35, 11 October 2020 (UTC)

I'm happy to check, but I'm not sure how I would. I changed the one place I found following my nose. FWIW, It was the use of the Wikidata list tempate here here Gettinwikiwidit (talk) 22:51, 11 October 2020 (UTC)

You can do this with the established model as well, see User:MisterSynergy/ListeriaTests/115th United States Congress for example. ListeriaBot allows you to process variables from the SPARQL query directly. —MisterSynergy (talk) 11:59, 12 October 2020 (UTC)

@MisterSynergy: Oh, I see. I didn't realize that columns could be variables in the query. TBH, this was my main motivation for using Q98077491. The only remaining argument I see for using it is that it allows a clearer distinction from the old model which preferred a single entry per senator. I'm happy to change it back to only using the parliamentary term (P2937) qualifier with United States senator (Q4416090) if there is a consensus. In fact doing it that way should make adding new congresses easier since there would be fewer moving parts. I'm still not sure how to track down current users of United States senator (Q4416090), though. Thanks for the help! Gettinwikiwidit (talk) 13:34, 12 October 2020 (UTC)

@Gettinwikiwidit, MisterSynergy: Possibly worth noting here that that query won't quite work in the case where someone has more than one P39 statement in the relevant period, as Listeria will silently discard additional rows for the same ?item. However, that shouldn't happen very often with US Senators — at least not compared to other countries where politicians change parties frequently, or temporarily move from the legislature to the Cabinet, etc. So I'd definitely support switching everything to a position held (P39) of United States senator (Q4416090) with a parliamentary term (P2937) qualifier here. (NB: I don't know how you're planning on actually importing the data, but you should be aware that various import tools, such as QuickStatements, also won't work with this model. I'm happy to help out with alternative import methods, however, if that would otherwise be a blocker.) --Oravrattas (talk) 14:09, 12 October 2020 (UTC)

Yes, currently it would not list such senators more than once. However, one could aggregate the qualifiers with GROUP_CONCAT() in case this happens more often, which is a bit hacky.
That said, since here are plenty of lists to maintain, one might also want to consider making a bot script for this very purpose. ListeriaBot is a nice and versatile tool that is useful in many situations, but sometimes a specific tool can be better for a particular job. It is not overly complicated to query with a script, write tables to a wiki page, and run this script regularly for updates. —MisterSynergy (talk) 16:06, 12 October 2020 (UTC)

Indeed. PositionHolderHistory already does exactly that for the sorts of positions that only one person usually holds at a time (Presidents, Ministers, Ambassadors, etc), and I hope to do likewise for legislative positions soon. --Oravrattas (talk) 19:55, 12 October 2020 (UTC)

Well, tracking use is already difficult within Wikimedia. That’s why we want to be very cautious with proposed changes such as this one.
I don't think that you need to ask for permission if you want to complete the missing claims with the currently used data model (using United States senator (Q4416090) with parliamentary term (P2937) qualifiers). The other one using Q98077491 was a bit quirky with no clear benefit, which is why you got some opposition. —MisterSynergy (talk) 16:06, 12 October 2020 (UTC)

The perceived benefit at the time was to make this list generator work, but that never got discussed because of the nature of the conversation. Again, I think if we try to keep conversations focused on the usefulness of a model we could have gotten here much sooner. Thanks again. Gettinwikiwidit (talk) 21:24, 12 October 2020 (UTC)

@MisterSynergy: As well as the UK (both nationally, and for the legislatures of Scotland, Wales, and Northern Ireland), this model is also used in India, Pakistan, Ghana, Estonia, Iceland, Hong Kong, Malta, and some sub-national legislatures, such as Hamburg. (There are probably also others that I'm not aware of.) The model was designed such that pretty much all standard queries will work with both versions, as long as they include a P279 check, as (per much previous discussion here) should be a standard feature of most queries due to how subclasses work in Wikidata. --Oravrattas (talk) 06:49, 12 October 2020 (UTC)

As far as I'm aware one of the other reasons why some people prefer this model is that it is possible to import data via QuickStatements when using it, which is largely impossible with the other version (as if someone held the same position on more than one occasion, QuickStatements will add all the qualifiers for the second onto the first instead.) There are, of course, other methods of bulk data entry other than QuickStatements, but most of the ones that can create two P39 statements with the same target item, but different qualifiers, require the user to be significantly more tech-savvy. --Oravrattas (talk) 06:54, 12 October 2020 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘

• @Oravrattas: @MisterSynergy: @Andrew Gray: Okay, I'm updating the full Senate record from the data mentioned above. I have not yet deleted previous statements. I have saved off all the Senator information as it existed before this upload. Thus, there is still clean up to do but I'm happy to discuss how best to do this. We should probably discuss it in a different thread. I'm open to suggestions of where that conversation should take place. Gettinwikiwidit (talk) 09:23, 14 October 2020 (UTC)
  • After the discussion in this thread a couple of days ago I thought you are going to switch to the other model using United States senator (Q4416090) as main values, rather than position-by-term items which caused this entire issue. Is this not the case?
    Anyways, here are three queries to monitor the coverage of P39 claims for US senators by parliamentary term: model using Q4416090 as main values, model using position-by-term items, and P39 statements without any P2937 qualifier. Eventually everything should be migrated to only one model. —MisterSynergy (talk) 12:30, 14 October 2020 (UTC)

Sorry. That was exactly my plan, but my I hadn't saved my change to the schema in OpenRefine. I made another error too because I assumed the upload would understand records and not rows. I'm trying to revert that group now and will do a new one with these issues fixed. Also my test case was too narrow to recognize that I had these errors. Sorry for the trouble. ( FWIW, I'm having issues with the undo in EditGroups, but it feels like the server might be overloaded. I'll try to complete it tomorrow. ) Gettinwikiwidit (talk) 12:47, 14 October 2020 (UTC)

Please be careful with the batch editing. Start with small batches, and watch your contributions continuously while editing in the beginning. We’ve already seen plenty of other beginners screwing up badly using these powerful edit automation tools. It is admittedly a relatively complex process to use them. —MisterSynergy (talk) 13:03, 14 October 2020 (UTC)

@MisterSynergy: Fixed. Clean up still pending. I'm eager to have feedback on how best to proceed. Gettinwikiwidit (talk) 22:10, 14 October 2020 (UTC)

I don't have full overview, but based on the queries linked earlier today by me, I'd say that there are still some remaining position-by-term claims left which need to be migrated or removed. Then there seems to be an excess of United States senator (Q4416090) claims particularly for the more recent parliamentary terms, so some statements are now duplicated. It needs to be figured out which ones they are (probably with WDQS), and whether they can simply be removed or something needs to be kept or migrated to the duplicated statement beforehand. The removal can technically be done with QuickStatements as described here using the statement-ID method, but this should carefully be evaluated in advance. Finally the positionholder claims without a parliamentary-term qualifier need to be dealt with; there are roughly 2000 of them, and I think a discussion with involved users would be helpful before they can be touched at all. —MisterSynergy (talk) 22:35, 14 October 2020 (UTC)

Yes, this is why I saved off the full state of the senate data beforehand to have the statement ids available. It's not clear to me who qualifies as an involved user. If there is a better place to make an open call for comment, please let me know. If I don't see any responses, I'll assume there is no interest and will follow my best judgement. Gettinwikiwidit (talk) 22:44, 14 October 2020 (UTC)

Missing collective language code aus for monolingual text

Hi, all! I was trying to add native (Australian Aboriginal languages (Q205143)) name (P2561) for Acacia bidwillii (Q15289416), but the collective language code aus (ISO 639-2 and -5) is “not a known language code” for Wikidata. Historical sources often don’t have an endonym for the language, only a region where it was spoken. I ended up flipping a coin between mis and und.

I just now found Help:Monolingual text languages. Seems I should normally do a Phab ticket without needing discussion/consensus here first? Or is this a different situation, being a collective code?

Pelagic (talk) 11:28, 10 October 2020 (UTC)

• Normally, you just ask for it in phab (see User:Lea_Lacroix_(WMDE)/List_of_lists_of_languages#General_ideas). You might need to stated that the exact language isn't known (so "aus" is preferable over "mis", but not "pjt"). --- Jura 11:33, 10 October 2020 (UTC)

There is no such thing as a collective language. Both ISO 639-2 and 5 are problematic. Thanks, GerardM (talk) 12:46, 10 October 2020 (UTC)

Indeed, GerardM. Though und isn’t a language either. The question then is whether we gain anything by using more-specific alternatives to und. It might make querying more complicated to look for (language codes) and (und + qualifiers), but at this stage based on your answer I’m inclined to let the qualifiers do the work. Unless someone has a counter-argument in favour of collective codes. Thanks Gerard and Jura for the feedback. Pelagic (talk) 08:36, 15 October 2020 (UTC)

@Pelagic: maybe is more of a collective code for languages, than a code for collective languages. I think the standard calls it macrolanguage code. "aus" is standard, so I don't see why we wouldn't want to use it to identify. --- Jura 08:45, 15 October 2020 (UTC)

So it is a problematic standard and you do not see why we should not use it. We should not use it because it does not identify the string properly... Everything that fits into such a code does make for rather ugly inconsistencies and, there is no solution for that. Thanks, GerardM (talk) 09:01, 15 October 2020 (UTC)

What alternative do you suggest? --- Jura 09:04, 15 October 2020 (UTC)

There is no one size that fits all. Thanks, GerardM (talk) 09:14, 15 October 2020 (UTC)

Well, just the use case discussed here (Q15289416#P2561). --- Jura 09:32, 15 October 2020 (UTC)

Checkin and checkout times

Before making any proposal I would ask how to handle check-in and check-out times for accommodation facilities or similar. They should be data objects using Q-Ids like 10:00 (Q55811483) representing times/hours. Of course, strings shouldn't be used as proposed four years ago. My question is: shall we use a general property (and which) combined with a qualifier or shall I propose two new properties? --RolandUnger (talk) 06:40, 14 October 2020 (UTC)

What check-in/out times are you talking about? Very difficult to modelize. For instance, departures from Orly Airport (Q223416) varies according to the destination and the airline… before take-off, not the same time towards Metropolitan France (Q212429) // Europe (Q46) // Overseas France (Q203396) // rest of the world (Q61029267)… Bouzinac ^{💬●✒️●💛} 07:02, 14 October 2020 (UTC)

I mainly think about times for hotels and other accommodation facilities handled as a point in time. They are therefore similar to opening hours of establishments. I do not think yet on duration before departures at airports. But maybe we can combine this for checkin (but we need a lot of Q-Ids for several periods). In this case we need additional qualifiers for destinations and airlines. --RolandUnger (talk) 07:28, 14 October 2020 (UTC)

@RolandUnger: This seems closely related to opening time (P8626) and closing time (P8627) which were recently created. However it also seems distinct enough that new properties would be useful. And no, I don't think this should be applied to airline departures, that is a completely different data domain. ArthurPSmith (talk) 14:46, 14 October 2020 (UTC)

@ArthurPSmith: Thanks for your answer. --RolandUnger (talk) 06:02, 15 October 2020 (UTC)

Adding Royal Horticultural Society url to items on plants

Many species (Q7432) of garden and ornamental plants have a page with a stable url on the Royal Horticultural Society (Q1032739) website that would be nice to add to the Template:Taxonbar (Q22741012), which is used on many Wikipedias. For example, Rosa xanthina (Q3442159) has the following url: https://www.rhs.org.uk/Plants/121050/i-Rosa-xanthina-i-Lindl/Details. All RHS pages on plants take this form. I proposed this on the enWikiProject Plants talk page about six weeks ago, and editors there thought it would be a good idea, with no dissents. There I stated, "It seems preferable that only existing Wikidata items on species should be given the links, since I am sure there are examples of RHS retaining old horticultural names, and there are certainly gobs of cultivars and hybrids that don't have Wikidata items. There are probably less than 10,000 species listed by the RHS. Importantly, they are not always in agreement with Kew, providing some extra value to their listing." Unfortunately I cannot see a way to proceed due to my Wikidata illiteracy. Can someone assist? Abductive (talk) 04:50, 13 October 2020 (UTC)

Without special property for it you can always say exact match (P2888) with the URL. Not that it matters but, what service do these pages provide? On first glance I couldn't find anything new or useful. Is this click bait? --SCIdude (talk) 14:01, 13 October 2020 (UTC)

They are often quite different, and include more (and different) common names, synonyms, and gardening information including the very important Award of Garden Merit (Q3501416). Abductive (talk) 19:49, 15 October 2020 (UTC)

The standard way to forward is a property proposal to create a property for Royal Horticultural Society IDs. ChristianKl ❪✉❫ 11:06, 15 October 2020 (UTC)

I tried and I haven't got a clue. Abductive (talk) 19:49, 15 October 2020 (UTC)

How to model a financial loan on Wikidata?

I'm importing some data from the European Investment bank about projects they have have funded (e.g. https://www.eib.org/en/projects/pipelines/all/20160014).

I would like to add available data about the amount of money that has been lent by the European Investment Bank, and potentially other organisations that may have lent money for a project.

Does anyone have any experience of modelling loans on Wikidata, or know of examples I can look at?

My initial thought was that there should be a new property for "creditor" (as well as an inverse property for "debtor"). Then I could add something like "creditor" = "European Investment Bank" to the item for the project. With this model, the actual amount owed would then be added as a qualifier, probably needing yet another new property for "loan amount" or similar (other existing properties like interest rate (P5899) could also be used as qualifiers). The new creditor property would also be similar to the existing investor (P1951) property, but for recording a loan instead of an investment. NavinoEvans (talk) 12:49, 15 October 2020 (UTC)

investor (P1951) is suitable for both equity and non-equity investments (we have more specific owned by (P127) with proportion (P1107) qualifier for equity investments), not sure if new creditor property is needed. quantity (P1114) may be suitable for a loan amount. I am pinging Wikidata:WikiProject Companies as well.

quantity (P1114) is intended for integer counts of something (unitless) so probably not appropriate for an amount of currency. We don't have a generic "amount" property, I've long thought we should... ArthurPSmith (talk) 16:25, 15 October 2020 (UTC)

Thanks very much for the input both, that's really handy to know. Yes it crossed my mind that an "amount" property would be a great help here (and in many other situations!). Definitely units are essential in this case to show the currency. I'll wait a bit to see what other ideas come in here, but will propose an amount property shortly I think.

I suggest looking at this Wikibase instance funded by the European Commission which might be of interest to you: https://linkedopendata.eu/. Among other things, they have uploaded 705095 EU-financed projects. Johanricher (talk) 17:30, 15 October 2020 (UTC)

This is amazing! I can't believe I'd missed this one, thanks a lot :) NavinoEvans (talk) 18:04, 15 October 2020 (UTC)

Kopiersperre Jklamo ArthurPSmith S.K. Givegivetake fnielsen rjlabs ChristianKl Vladimir Alexiev Parikan User:Cardinha00 MB-one User:Simonmarch User:Jneubert Mathieudu68 User:Kippelboy User:Datawiki30 User:PKM User:RollTide882071 Andber08 Sidpark SilentSpike Susanna Ånäs (Susannaanas) User:Johanricher User:Celead User:Finnusertop cdo256 Mathieu Kappler RShigapov User:So9q User:1-Byte pmt Rtnf econterms Dollarsign8 User:Izolight maiki c960657 User:Automotom applsdev Bubalina Fordaemdur

Notified participants of WikiProject Companies

24-hours online meetup for the Wikidata birthday

Hello all,

As you may know, every year around October 29th we're celebrating Wikidata's birthday with various events, presents and wishes. This year, you can find a calendar of various events, most of them taking place online. You can still organize one, find more information here. You can also prepare a birthday present and add it to this page.

The year 2020 is special in many ways, for example because most of the international events where we normally gather, meet and have informal time together have been cancelled. I definitely miss these events and the possibility to chat with some of you, discuss the projects you're working on and discover what's new on Wikprojects.

For this reason we decided to try something new: a 24-hours long online meetup, where you can join and leave whenever you want, depending on your time zone, and chat with each other about what is making you enthusiastic about Wikidata. It will start on October 28th at 17:00 GMT (18:00 in Central Europe) and will end on October 29th at the same time. This meetup will be informal, self-organized, and welcoming everyone, from newcomers to experienced users. It is most importantly an experiment: we never tried something like this in the past, and we're still figuring out about the details.

There's no formal program during the meetup, and people are invited to chat about anything related to the main theme. However, we're inviting people to sign up as a facilitator for a one-hour slot: you can find a second person to co-facilitate with you, sign up in the schedule, and add an optional topic to the slot. This opportunity is great if there's a topic you'd like to discuss with the community.

As we're an international crowd, the discussions will mostly take place in English. However, it is possible to include other languages: for example, a slot in French has already been added. Please also note that the meetup is covered by the code of conduct for technical spaces.

So, feel free to note down the date in your calendars, don't hesitate to sign up to facilitate a slot! The link to the event (we will use the software BigBlueButton) will be added shortly before the date.

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

See you there! Lea Lacroix (WMDE) (talk) 15:24, 15 October 2020 (UTC)

When will this project get a working ClueBot?

From March to today (when it was reported to en-WP's administrator noticeboard), our project's description of the New York Yankees (Q213417) was baseball team and Major League Baseball franchise in the Bronx, New York, United States, buttheads, cheaters who have won too many World Series. I'll state the issue bluntly: the incompetence of this project with handling vandalism is an utter embarrassment, and it provides a 100% valid rationale for the Wikimedia projects that decide not to integrate with us.

A few of us were chatting about this issue on the Wikidata Discord chat a little while back, and it seems that there really aren't any active bots that have been tuned for operation on Wikidata. Does anyone want to work on this problem? It seems like it ought to be literally the #1 priority for this project (improving the references system being the only other real contender), yet I can't find any other discussion on it and my query back in May died after a single reply and remains the most recent post on the vandalism talk page. {{u|Sdkb}} ^talk 06:59, 7 October 2020 (UTC)

• If you think it's needed, add it to Wikidata:Bot requests. --- Jura 07:35, 7 October 2020 (UTC)

  @Jura1: I'm unfortunately not technical enough to know how to create a well-formed bot request, other than to know that we need something. {{u|Sdkb}} ^talk 08:02, 7 October 2020 (UTC)

  • No technical knowledge is needed to make such requests. --- Jura 08:05, 7 October 2020 (UTC)

• It is quite embarrassing for Wikidata that vandalism is not detected. I think this an aspect of the general data quality problems of Wikidata. Right now Wikidata does not seem to have the number of editors it needs, given the scope of the project, to address problems. There are long to-do lists, like the constraint violation reports or the unpatrolled recent changes, but they grow faster than they are resolved.
  If there is serious interest in such a bot I can help with drafting the requirements. There are different kinds of vandalism in Wikidata, changing labels or descriptions and vandalizing statements. Which kind of vandalism do you think is doing the most damage right now? Pyfisch (talk) 16:23, 7 October 2020 (UTC)
• I dunno, seems like an accurate description to me ... (Joking, I was joking!). But it is not clear to me what sort of "bot" could help with the many varied kinds of vandalism possible here. We have Wikidata:WikiProject Counter-Vandalism with a variety of tools for humans to patrol recent changes that need it, but the number of those seems to be much higher than what we can handle. And yet not insurmountably high - the numbers are on the order of 6000 per day. I patrol my watchlist items and it doesn't take long, so 100/day per helper seems doable. Can we recruit/organize a group of 60 or so to handle this somehow? ArthurPSmith (talk) 17:56, 7 October 2020 (UTC)
  • It is not that simple. Part of the difficulty with patrolling and countervandalism activities here at Wikidata is the inherent multilingual nature of the project. For terms (labels/descriptions/aliases) vandalism, you basically need to understand the language in which the changes where made; for statements, you quite often need to have a look into sources in languages you do not understand. I do not think that the problem can be solved by simply gathering 60 editors who do 100 patrols a day each. I also think that this language diversity is the main obstacle for a countervandalism bot, because from a purely technical perspective it should be much simpler to check Wikidata edits than changes to unstructured Wikipedia pages. —MisterSynergy (talk) 18:27, 7 October 2020 (UTC)
  • I think the most effective approach to instances where we don't have enough editors is not to just ask for them but to make the project more welcoming to them so that they come. Wikidata inherently has some elements that are harder to learn for humans, but better documentation and instructions could go a long way. I've been engaged in a long and hard battle to improve newcomer resources and improve the user interface at en-WP, and perhaps someone should take on a similar effort here. {{u|Sdkb}} ^talk 06:52, 8 October 2020 (UTC)

    Also, I just checked Wikidata talk:WikiProject Counter-Vandalism, and the most recent post there is from...2019. {{u|Sdkb}} ^talk 17:09, 8 October 2020 (UTC)

• I do not know exactly how ClueBot works, but from what I understand there are two issues: (i) there is no context on Wikidata, vandalism is when a description is changed to another description, or a value of a property to another value, but there no such thing as for example removing a part of the sentence including a period at the end, I guess this is the kind of things what ClueBot is trained for. If in the English Wikipedia someone changes the name of mayor of Pryluky, Chernihiv Oblast, Ukraine, no bot would be ever able to catch this, unless the name of the new mayor is Donald Trump or Michael Jackson. This is kind of vandalism were are predominantly dealing with here, and it is not well suitable to be caught by the bot; (ii) whereas it would be great if we could have some progress in vandalism in English, we also have other major languages with their share of vandalism, and ClueBot for these languages has never been developed.--Ymblanter (talk) 19:26, 7 October 2020 (UTC)
• Writing a good anti-vandal for wikidata would be hard. I think it's doable but it'd be many many hours of work. For example, I think I have the knowledge to do it but who has the time or money for infrastructure to do it right? Not me. BrokenSegue (talk) 19:31, 7 October 2020 (UTC)

  @BrokenSegue: Have you or others considered meta:Grants:Project? Sophivorus recently used it to get compensation for working on excerpts at en-WP and might be able to provide feedback on the process. {{u|Sdkb}} ^talk 06:47, 8 October 2020 (UTC)

  @Sdkb: practically the people qualified to do this will likely have high-paying jobs that eat up much of their time and pay better than a grant. I have considered applying for a grant maybe after I retire. Maybe students/researchers are our best bet for now. BrokenSegue (talk) 16:50, 9 October 2020 (UTC)

Three thoughts here, one on ranking, one on the constraints and one on the minor wikis.

1. I think it is time to restrict ranking on this wiki, so that only autoconfirmed users can mark statements as preferred. That way, an group of people could fix the statements and mark them, instead of wondering what the other people in the group just fixed. Also, it is possible to write an Lua import module on wikipedia that ignores any statements which are not marked as preferred. So, this has the potential to change the view on wikidata quality on wikipedia a bit quicker.

2. About the constraints it might be just time to make it impossible for people to make some constraint violations. It is a matter of newbie friendlyness versus keeping things orderly. If there are any bots who make constraint violations, then they should be fixed, ASAP.

3.On minor wikis, I just think that the Wikidata vandalism dashboard is not well enough known for users from those minor wikis to use. Those minor wikis after all have fewer users, so they are not going to go over an list over all of the changes on major languages.--Snaevar (talk) 21:48, 15 October 2020 (UTC)

• Out of curiosity, is there anybody to respond on my remark from the query: I am sure we had one. But I don't know what the name was and if it's still working.? It might have been connected to the ORES service. --Matěj Suchánek (talk) 07:49, 8 October 2020 (UTC)

I also remember us having one. Maybe, it was discontinued because of too much false positives? According to https://www.wikidata.org/wiki/Wikidata:ORES the ORES scores seem at the moment only to be used for RecentChanges/Watchlists to highlite bad edits. Given that Wikidata has less context then Wikipedia for individual edits that might be the best we can do. To the extend that ORES needs to be improved for Wikidata, there seem to be pages where patterns that are missing can be reported. ChristianKl ❪✉❫ 23:50, 11 October 2020 (UTC)

Goodreads vs Goodreads

What is the difference between Goodreads version/edition ID (P2969) and Goodreads work ID (P8383)? And why are they showing up on the same data item together? --EncycloPetey (talk) 15:51, 14 October 2020 (UTC)

My understanding is that both properties appearing on the same item is an error (and if not caught by a constraint should be caught by one). Is there something about the English description of P8383 that does not make the difference clear? Mahir256 (talk) 16:01, 14 October 2020 (UTC)

Sorry, I missed the English desciption because I went directly to the Talk page looking for information. Putting it in the English description is a weird choice for placing the explanation of the property's usage. I have renamed the edition property in English to make it clearer. --EncycloPetey (talk) 00:24, 15 October 2020 (UTC)

Further, I see that the example usage for Goodreads version/edition ID (P2969) has both properties on the same page. So even the example is wrong. --EncycloPetey (talk) 00:27, 15 October 2020 (UTC)

I really see no harm in having both but yes the difference is between the work and the edition. Many people attach ISBNs to "works" not realizing that those are really relevant to editions but it's also kinda annoying to have to make a new item just to associate an ISBN with a book. BrokenSegue (talk) 01:01, 15 October 2020 (UTC)

Annoying but necessary, since each edition can potentially have a different editor, publisher, date of publication, place of publication, ISBN, translator, illustrator, cover artist, etc. And none of that data can be entered until someone creates the data item for the edition. It's not about "harm"; the property description itself says that they shouldn't be on the same data item. --EncycloPetey (talk) 15:23, 15 October 2020 (UTC)

I would rather a user add a matching ISBN to a "literary work" than do nothing even though it's not technically correct. BrokenSegue (talk) 17:40, 15 October 2020 (UTC)

There was a suggestion in a thread at Wikiproject Books to create edition items only for the special-but-common case where there is only one edition, or you want to only add the one. The data on editons tends to be a superset of the work item's anyway, since they include author, language, title, etc. in addtion to edition-specific data such as ISBNs and number of pages.--Matthias Winkelmann (talk) 21:37, 15 October 2020 (UTC)

The language and title are edition-specific items, not work items. Works can be translated into other languages, and the title can change with editions even in the same language (e.g. the first Harry Potter book), so language and title are edition-specific bits of data as well.

Note that this discussion started because of dual Goodreads properties appearing on data items for works that have multiple editions and for which data items exist for multiple editions. So the issue of just-one-edition is not relevant here. --EncycloPetey (talk) 21:44, 15 October 2020 (UTC)

I went ahead and created a conflict constraint in both properties. NMaia (talk) 01:45, 16 October 2020 (UTC)

Mudinepalli

• Mudinepalli (Q3418849)
• Mudinepalli (Q55641153)

Items probably should be merged... same place in India. Wbm1058 (talk) 22:18, 15 October 2020 (UTC)

 Done NMaia (talk) 01:51, 16 October 2020 (UTC)

Wikidata inaccessible from mobile Wikipedia

Anyone ever tried to get to an article's Wikidata item in the mobile version of Wikipedia?

As you probably noticed, at https://en.m.wikipedia.org/wiki/Wikidata , there is no link to Q2013. I still find that strange.

Even https://en.m.wikipedia.org/wiki/Wikidata#/languages (a list built with data provided by Wikidata), has no link to Q2013.

It seems to me that the mobile version isn't following Wikidata:Data_access#Best_practices_to_follow. --- Jura 07:13, 10 October 2020 (UTC)

Press "More"/"vertical triple dots" on the bar under the page name of the article -> "Wikidata item". -Premeditated (talk) 09:21, 10 October 2020 (UTC)

• Is that the top left corner? I only have "Home", "Random", "Nearby", "Log in", "Settings", "Donate" there.
  Maybe you are logged-in? Wikidata:Data_access#Best_practices_to_follow isn't for logged-in users only. --- Jura 09:32, 10 October 2020 (UTC)

(ec) Hi, Jura, is this to do with everybody getting logged out again? If I'm logged out I get the AMC-style toolbar with just three buttons: languages, watch, edit. If I log in but disable mw:AMC, I see the history button also. Only if I re-enable AMC do I see the [⋮] button with drop-down page tools: permalink, Wikidata link, etc. —Pelagic (talk) 09:49, 10 October 2020 (UTC)

• No, this just about what mobile Wikipedia users get in general. What logged-in users get doesn't really matter. --- Jura 09:55, 10 October 2020 (UTC)
  • Sorry, I saw that you meant logged-out after I hastily committed my edit-conflict text. I would argue that a lot of logged-in mobile users would not have opted in to "Advanced Mobile" mode, so that’s also a problem (but I don’t have numbers on AMC uptake). Also, for what it’s worth, I haven’t found any Page Tools (including Wikidata link) in the iOS mobile app either. Pelagic (talk) 10:29, 10 October 2020 (UTC)

• To see what it represents, maybe we can compare pageviews on "English Wikipedia" in desktop (1,632,538) with the same in mobile view (1,443,922). Mobile app would be 18,439 so that seems somewhat irrelevant. So it seems that almost half of the users don't get a chance to browse the Wikidata summary of Wikipedia articles. --- Jura 11:15, 10 October 2020 (UTC)

I believe this interface is mw:Reading/Web/Advanced_mobile_contributions, developed by WMF. Feel free to create a ticket on Phabricator with the tag "Advanced Mobile Contributions". I reached out to the team and it seems that no major change is planned before summer 2021, but it's worth mentioning the issue on Phab nevertheless. Lea Lacroix (WMDE) (talk) 16:53, 13 October 2020 (UTC)

• @Lea Lacroix (WMDE): I think that is just for logged-in users. The question is here how to address the requirements of Wikidata:Data_access#Best_practices_to_follow by WMF itself for logged out users. --- Jura 05:52, 16 October 2020 (UTC)

Species... Found in

Hi all. Can someone point me to any discussions on creating a new geo property which could be placed on species eg under Ulmus glabra (Q147498) we would have a property called 'This species can be found in...' Germany (Q183) etc. Visually, we do have taxon range map image (P181) of course, but this new detailed info could be taken into an API with info such as 'Moths which can be found in this country include...', as well as creating new maps for Wikipedia etc. A second possible property could be 'This species is native to...' Thanks. Llywelyn2000 (talk) 10:22, 16 October 2020 (UTC)

@Llyweyn2000: Honest question: aren't the endemic to (P183) and invasive to (P5588) properties already useful in this regard? Probably no need for another new property, I think? -- Btcprox (talk) 12:01, 16 October 2020 (UTC)

Brilliant! These two are new for me. I'll try a few and come back. Perhaps I'll come back regarding 'Native to....' at a later date, but this solves my immediate problem. Many thanks! Llywelyn2000 (talk) 14:31, 16 October 2020 (UTC)

Qualifiers to differentiate

Colonia Iulia Urbs Triumphalis Tarraco (Q1501940) is both roman colonia (Q756780) and archaeological site (Q839954). Thus, Tarraco belongs to Ancient Rome (Q1747689) and Spain (Q29) for country (P17). I have shown this with statement is subject of (P805) and done the same for located in the administrative territorial entity (P131). However, a warning marks the values of the second property. Are the qualifiers correct? How should it be done? Thank you. --Romulanus (talk) 14:16, 16 October 2020 (UTC)

Icons for identifiers

I've started adding favicons as icon (P2910) values for various properties, primarily for external identifiers. I think it'd be neat to use these to stylize external links. You can see all the icons added so far with this SPARQL query. If anyone wants to help in my icon-adding quest, it'd be much appreciated! Cheers, IagoQnsi (talk) 14:48, 16 October 2020 (UTC)

Here's a query I've been using to find properties to add icons to. –IagoQnsi (talk) 19:22, 16 October 2020 (UTC)

Q100410681 can be deleted

I created Q100410681 when I should not as Q1877545 existed. I have made Q1877545 identical to Q100410681, so Q100410681 can be deleted or joined.Smiley.toerist (talk) 13:02, 17 October 2020 (UTC)

I merged both data sets. --HarryNº2 (talk) 17:42, 17 October 2020 (UTC)

 Done --HarryNº2 (talk) 17:42, 17 October 2020 (UTC)

GEOPEDIA, linking Wikipedia and QGis

As a geographer I have devoted an important amount of time gathering, classifying, and selecting information from different and desegregated sources. That's why an open geographical database would offer a great opportunity, in the age of global markets, for governments, industries, and emerging companies. Due to the necessity of this platform is important to build an interactive map. Wikipedia and Wikidata have a great amount of information that could be shown through cartography, and cartography can be built with the open, free platform QGis. Cooperation between both platforms, QGis and Wikipedia, would boost them and would make easier access to information.  – The preceding unsigned comment was added by Dl.ojeda2008 (talk • contribs).

It likely obvious to you, but I am not sure what you mean by "QGis", I assume you are referring to QGIS? Right now OpenStreetMap already links to Wikidata and vice versa and Wikidata incorporates geographic data from a variety of sources. It is possible to create interactive maps like this one from with information from Wikidata. --Pyfisch (talk) 18:36, 17 October 2020 (UTC)

Sitelink to redirect or intentional sitelink to redirect

Which "badge" should be used on interwiki? How redireect maybe unintentional? Eurohunter (talk) 22:05, 8 October 2020 (UTC)

If you add it intentionally, add the intentional sitelink to redirect (Q70894304) badge. The other one, sitelink to redirect (Q70893996), is merely a marker that should probably only be used by bots to mark redirects which are not marked as such at all. There are plenty of reasons why an "unintentional" redirect is connected to in item, e.g. when a full Wikipedia article is transformed to a redirect for whatever reason and nobody changed it on Wikidata. —MisterSynergy (talk) 22:25, 8 October 2020 (UTC)

@MisterSynergy: But then, how can such a linked redirect page be "intentional"? You still didn't explained it carefully, which may lead to some possible different behaviours on this topic: "Hey I think this item must have a link in my language, but there's no such a page in my language Wikipedia, so I created it as a "broken" redirect and linked it, then I fixed it, is this intentional or not?" "Hey I can't directly link the correct named page in my language to an item, can I temporary break it and link it, I promise to restore it after linking is successful, so this is intentional, right?" "Hey someone (or their sockpuppets) created a vandalism contents only page and then linked it to Wikidata, while it can just be speedy deleted, this cound really be unfair for me because the title may still useful, should I clean it up and change to a redirect, and try to say I'm linking an intentional redirect?"... --Liuxinyu970226 (talk) 04:54, 9 October 2020 (UTC)

The situation regarding redirects is somewhat changing, but unfortunately very slowly. The corresponding RfC at Wikidata:Requests for comment/Allow the creation of links to redirects in Wikidata is still not completely implemented, as far as I understand; however, this nasty hack which you describe should not be necessary forever, so that some redirects can be considered valid in Wikidata. They are irrelevant for notability, and for sure a lot of redirects in Wikipedias should not be linked to Wikidata at all; however, redirects as sitelinks are a thing and these relatively new badges are meant to keep track of them. You can learn more about their introduction in the phabriactor ticket phab:T235420. —MisterSynergy (talk) 10:49, 9 October 2020 (UTC)

I started a draft on a possible policy to deal with the redirects: https://www.wikidata.org/wiki/User:ChristianKl/Draft:Help:SitelinksToRedirects . I'm happy for any comment and contribution to it before starting a RfC for it. ChristianKl ❪✉❫ 16:48, 9 October 2020 (UTC)

@ChristianKl: We already have some guidance at WD:N:

"If available, a template corresponding to Template:Soft redirect with Wikidata item (Q16956589) should also be added to the redirect page [on the wiki in question], to show that the Wikidata sitelink is intentional, and not eg the unintended left-over effect of a Wikipedia article merge that should also have been reflected on Wikidata. Sitelinks to redirects should typically not be created unless (a) there is a substantial section about the subject on the target page of the redirect, reflecting all or most of the information in the Wikidata item; and (b) there is good reason not to merge the two corresponding Wikidata items."

Jheald (talk) 21:34, 10 October 2020 (UTC)

That section is not the result of a consensus finding process and having it in the references section is a bit strange. It seemed okay in the meantime when the situation along sidelinks was unclear after the RfC, having a new RfC that actually species a clear policy seems better. I do grant that it makes sense to mention Template:Soft redirect with Wikidata item (Q16956589) in the page. Given that we now have the badges, it seems to me that the badges should be the primary way to mark intentionality. ChristianKl ❪✉❫ 23:56, 10 October 2020 (UTC)

@Liuxinyu970226: @MisterSynergy: @Jheald: @ChristianKl: People are removing redirects for no reason what u can see at Xbox Series S (Q98967383). How I can add there badges without removing them from item then blanking them on Wikipedia for "fake page" then restoring them because it is impossible to connect redirects directly. Eurohunter (talk) 12:26, 17 October 2020 (UTC)

This is unfortunately not yet possible, as WMDE needs to modify the configuration regarding sitelinks somehow. Until this happens, there is no other possibility than to remove the redirect on Wikipedia temproarily, add the sitelink with badge, and activate the redirect on Wikipedia again. ---MisterSynergy (talk) 22:42, 17 October 2020 (UTC)

@Eurohunter, MisterSynergy: Done for de, en and es, plus added zh, but Portuguese not done as warned me as "Aviso: esta operação foi identificada de forma automática como prejudicial por remover uma quantidade significativa de conteúdo. Se está revertendo um vandalismo no artigo, por favor, desenvolva um pouco mais o assunto a ser tratado pelo artigo, e então pode pressionar novamente o botão "Gravar página". Neste caso, você pode também nos avisar aqui, para que tal mensagem não apareça novamente. Uma breve descrição da regra de abuso com a qual a sua acção coincidiu é: Branqueamento de página". --Liuxinyu970226 (talk) 23:08, 17 October 2020 (UTC)

RFC: Better Managing Bot Load

In light of the recent issues involving bot load on wikidata I threw together a technical proposal to try to better manage the limited edits per minute available to us.

I'd appreciate it if bot operators/devs commented on my proposa RFC: Wikibase API Proxy.

BrokenSegue (talk) 03:02, 16 October 2020 (UTC)

@BrokenSegue: For RFCs we have Wikidata:Requests for comment. --Succu (talk) 21:08, 17 October 2020 (UTC)

true, but that page says "You are more than welcome to open a new RFC process to get opinions over a topic, but that should be done after a long discussion via the other channels." no idea what other channels they mean and also that page seems to get lower traffic than here. I'll use it in the future though. BrokenSegue (talk) 22:40, 17 October 2020 (UTC)

Drapers' Gardens: one area and two different buildings

I posted the following in the talk page for Drapers' Gardens but think that this "Interwiki conflicts" page is probably more appropriate.

I've split the Drapers' Gardens Wikidata item into two, because the existing item was a mishmash of facts about two different things: the Drapers' Gardens tower designed by Seifert and built in the sixties, and the Drapers' Gardens office block designed by Foggo Associates and built in 2009. Here is the item specific to the 2009 building.

This Wikipedia article covers Drapers' Gardens as an area, covering the history of the area and the two different office developments built on it. At the moment, its "Wikidata item" link refers to the Seifert tower.

It might be better to split this Wikipedia article into three: one for the area and its history; one for the Seifert tower; and one for the modern office block. Alternatively, create a Wikidata entry which represents the Drapers' Gardens area, rather than a specific building, and have this Wikipedia article refer to that.

Can those with experience in this sort of situation please advise on how to represent Drapers' Gardens when it is three things: an area in City of London, a building built in that area and then demolised, and also a newer building built in that same area? --Bobulous (talk) 14:05, 17 October 2020 (UTC)

The usual procedure is to create an item and make it instance of (P31)-->Wikipedia overview article (Q20136634) with main subject (P921)-->all items the article is about. Just like a scientific journal review article. --SCIdude (talk) 15:47, 18 October 2020 (UTC)

Merging request

I mistakenly duplicated object's site Q94579544 with creating Q100546713. Francesco 13 (talk) 13:59, 18 October 2020 (UTC)

@Francesco 13: Hi and welcome to wikidata! You can merge items yourself: See help:merge. --Haansn08 (talk) 14:24, 18 October 2020 (UTC)

Thank you! Francesco 13 (talk) 16:34, 18 October 2020 (UTC)

Removal of ethnic group without sources

Hi, I removed ethnic group (P172) from a small sample of living people where the claim had no source at all. As ethnic group (P172) is a controversial property and it should only be used for people who claim the given ethnic group themselves or it is widely agreed upon by scholars. Please discuss at Wikidata:Edit groups/QSv2T/1603028249716 on how you want to proceed with these claims lacking sources. --Pyfisch (talk) 15:06, 18 October 2020 (UTC)

Idea for item update

You found a good idea to add product or material produced or service provided (P1056) on Nokia (Q1418)? --62.18.86.255 21:37, 18 October 2020 (UTC)

Please help me restore

von Knorring (Q20726377) was the surname used by multiple noble families in Germany and Sweden and it was improperly merged with Knorring (Q2362536), which was one of the noble families that use that very surname. I demerged but it will not let me restore von Knorring (Q20726377) to the premerge state. Can someone help. --RAN (talk) 01:47, 19 October 2020 (UTC)

OK, I think I fixed it, the Wikipedia links and categories were all jumbled between the Swedish noble family, and the German noble family, and the surname, and the disambiguation page. That will still require someone else looking at it, but I was able to make the restorations to von Knorring (Q20726377). A bad merge can really screw things up. --RAN (talk) 01:54, 19 October 2020 (UTC)

Two items to fusionate

Hi, I found 2 elements (Festividad Día Mundial de la Pereza and World Day of Laziness) that talk about the same theme, the World Day of Laziness in Itagüí (colombian municipality).

For that reason, I propose his fusion.

I apologize if my english can't have a really good level. Sr. Knowthing (talk) 02:54, 18 October 2020 (UTC)

Done, but it's not really a television channel, is it? Ghouston (talk) 07:05, 18 October 2020 (UTC)

I switched it to "festival". --RAN (talk) 04:53, 19 October 2020 (UTC)

Wikidata weekly summary #438

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

• Discussions
  • Other: Proposal to spin-off non-Earth coordinates from property P625 (coordinates location), currently only 7000 of 8 million values: see Property proposal/planetary coordinates

• Events
  • Upcoming: live SPARQL queries on Twitch and in French by Vigneron, October 20 at 18:00 CEST
  • Upcoming: Next Linked Data for Libraries LD4 Wikidata Affinity Group call: hands-on work with Wikidata on creating items for authors of COVID 19 papers and brainstorming for future hands-on work, 20 October. Agenda
  • Upcoming: Wikidata Lab XXV: Interoperabilidade e acesso aos dados, in Portuguese by Ederporto (Wiki Movimento Brasil) - October 22 17:00 UTC (14:00 BRT)
  • WikiCite 2020: Virtual Conference happening 26-28 October 2020. Many live-stream sessions across several languages and timezones. LWyatt (WMF) (talk)

• Press, articles, blog posts, videos
  • Tisane Labs announce new feature in Tisane API, allowing tagging and extraction of Wikidata entities: blog post
  • Video: LIVE Wikidata editing at HACK4OPENGLAM - YouTube
  • Video: Wikidata Lexeme BY Mahir256 - YouTube
  • Video: Introduction to Wikidata (in Swedish) - YouTube
  • Video: Introduction to SPARQL (in Swedish) - YouTube
  • Video: Bibliographic data on Wikidata (in Swedish) - YouTube

• Tool of the week
  • Prepbio tool uses Wikidata to create the beginnings of a Wikipedia stub biography article. (2020-10-19)

• Did you know?
  • Newest properties:
    • General datatypes: ritual object
    • External identifiers: Scottish Brick History Brick & Tileworks ID, Spotify playlist ID, Artist ID of the department of Prints and Drawings of the Louvre, NBIC taxon ID, Danish educational institution number, CTAN package ID, Dresden Street ID, Emojipedia ID, FIBA Hall of Fame ID, Stadtwiki Dresden article
  • New property proposals to review:
    • General datatypes: Unicode name, branch or affiliate, economy of topic, checkin and checkout times (v.2), maximum current, Skipsrevyen ID, route operated, power plant unit, annual average daily traffic, Humanitarian OpenStreetMap Tasking Manager project ID, isotopically modified form of, Numérisé par
    • External identifiers: Kramerius of Moravian Library UUID, Hungarian Water Polo Federation player ID, Irish Statute Book ID, Mir@bel publisher ID, CCFr library ID, Decorati onorificenze repubblica, BHL name ID, AWMDB person ID, AVN person ID, American Heritage ID, xHamster ID, Great Places ID, Famepedia ID, Nachlässe in Austria, Archivio Storico Ricordi opera ID, PAN member, DC Historic Sites ID, Literary Encyclopedia (1929-1939), Concise Literary Encyclopedia, aagm people ID, aagm sites ID, Yandex Zen ID, GeoGuessr ID, Polski Słownik Judaistyczny ID, SMB-digital ID, identifiant station Vélib' Métropole
  • Query examples:
    • Gallery of 3D models of structures mapped in OpenStreetMap that are available under Creative Commons licenses at Wikimedia Commons - OSM/Wikidata query (sophox.org)
    • Item named after item named after item named after item named after item named after item
    • UK parliament members' opposition posts, start dates and end dates where Wikidata has a Parliament ID for that member (Source)
    • Count of Irish Statutory Instruments, 1948-2020 (Source)
    • Movie genre trends (Source)
    • Scientific papers explaining backward contact tracing (Source)
    • List of sovereign states in en, fr, de, ru (Source)
    • Locations depicted in paintings (Source)
    • The most commonly depicted locations (Source)

• Development
  • Blocked users trying to create new Items no longer make us skip QIDs (phab:T232620)
  • So far we don't have JSON dumps for Lexemes. We're working on regularly generating them as well now (phab:T264883)
  • Continuing work on getting the Query Builder to generate the first query. Almost there.
  • Started the feedback round for the REST API. Please provide your input if you are currently using the action API or plan to develop against Wikidata's API in the future.

You can see all open tickets related to Wikidata here. If you want to help, you can also have a look at the tasks needing a volunteer.

• Monthly Tasks
  • Add labels, in your own language(s), for the new properties listed above.
  • Comment on property proposals: all open proposals
  • Contribute to a Showcase item.
  • Help translate or proofread the interface and documentation pages, in your own language!
  • Help merge identical items across Wikimedia projects.
  • Help write the next summary!

Read the full report · Unsubscribe · Mohammed Sadat (WMDE) 15:55, 19 October 2020 (UTC)

Wikidata query suddenly broke?

Does anyone happen to have an idea of why this query fails? It used to work, as far as I know. It was being used for WikiProject Video Games' list of new games, but it hasn't been able to update the list since October 8th because the query is broken. Nicereddy (talk) 22:05, 18 October 2020 (UTC)

it works if you replace the " BIND(IRI(REPLACE(?id, '^(.+)$', ?formatterUrl)) AS ?url)" bit with "BIND('foo' as ?url)" so I think the issue is there. No idea why it would suddenly fail because of this. BrokenSegue (talk) 22:33, 18 October 2020 (UTC)

It got solved here - https://www.wikidata.org/wiki/Wikidata:Contact_the_development_team#SPARQL_Query_stopped_working_October_8th/9th --Tagishsimon (talk) 20:03, 20 October 2020 (UTC)

Request for comment: email address (P968)

Hello all,

I just want to draw a slightly broader attention to the issue raised here. In short, all emails have to be entered with the prefix "mailto:". This causes several problems: When entering an email on wikidata:

1. There is no prompt indicating that "mailto:" should be included

2. The error if the "mailto:" is omitted is Could not save due to an error. This URL misses a scheme like "https://":

3. It is counterintuitive to most users to use a prefix that's usually invisible

4. It is different from other identifiers, such as ORCID iD (P496) which don't require an explicit prefix

Additionally, when using external tools (e.g. quickstatements, or the API clients in python and R):

5. It causes errors with no warning messages

For balance, downsides of changing:

1. Infoboxes in wikis that list email addresses would have to be checked and corrected

2. External tools that draw from wikidata and rely upon that prefix may break

So, I think that if it is to change, the sooner the better to lessen these drawbacks and maximise the benefits. T.Shafee(evo&evo) (talk) 11:39, 20 October 2020 (UTC)

Auto-add interwiki for templates

Hi. I found that templates are often copied from one wiki into another (with localization) and often the template name stays the same. For example there are around 200 templates in this category uk:Категорія:Шаблони:Підсторінки шаблону S-par that have equivalents in English wiki, but interwiki are not set. I suspect there are many many more of them and maybe there's an existing tool to do it. Smth like in Wikidata:Tools/Edit items. Or maybe tool that allows to set a lot of interwiki at once (import from csv, excel?). --Kanzat (talk) 01:03, 21 October 2020 (UTC)

Timetables need a period validity property

I have problems coupling this timetable extract to the year 1933. For any public transport timetable the periode of validity is an essential property, certainly for historic timetables. Even for the date of publication I get an error.Smiley.toerist (talk) 10:33, 21 October 2020 (UTC)

Wrong postal code for Indiana (Q1415#P281)

The field "postal code" states "15701" which is the ZIP code for Indiana, Pennsylvania. I would like to change it to "IN" but the page doesn't have an edit button.  – The preceding unsigned comment was added by [[User:|?]] ([[User talk:|talk]] • contribs).

Maybe @Abián: who protected the page wants to help. --- Jura 07:36, 22 October 2020 (UTC)

Request to move Roffe to Roffe (surname)

The Category:Roffe (surname) exists in Commons but does not link to the related Wikidata Roffe entry https://www.wikidata.org/wiki/Q28123821 I assume a move as requested would fix this Paul foord (talk) 02:48, 22 October 2020 (UTC)

No, it's just a missing sitelink, which I added. Ghouston (talk) 04:20, 22 October 2020 (UTC)

• If you are looking for similar names also, here is Roffé (Q100712314) and Roffés (Q100712440) --- Jura 06:26, 22 October 2020 (UTC)

Please volunteer to help

At Talk:Q785304 there is a project to fix the chart of the Mayors of New York City, they are ordered by date of start_time and get concatenated by a bot. Once ordered the errors and omissions become obvious. You can see the most recent 10 or so are complete. As always with these officeholder charts we may get some leftovers, because some may be interim holders that are not listed in the official account and assigned a number. They can appear in the chart but not be assigned a mayoral number. --RAN (talk) 07:09, 22 October 2020 (UTC)

• There used to be a bot that would import start and end dates from Wikipedia. Maybe it can be re-activated? --- Jura 07:31, 22 October 2020 (UTC)

Important: maintenance operation on October 27

Read this message in another language • Please help translate to your language

The Wikimedia Foundation tests the switch between its first and secondary data centers. This will make sure that Wikipedia and the other Wikimedia wikis can stay online even after a disaster. To make sure everything is working, the Wikimedia Technology department needs to do a planned test. This test will show if they can reliably switch from one data centre to the other. It requires many teams to prepare for the test and to be available to fix any unexpected problems.

They will switch all traffic back to the primary data center on Tuesday, October 27 2020.

Unfortunately, because of some limitations in MediaWiki, all editing must stop while the switch is made. We apologize for this disruption, and we are working to minimize it in the future.

You will be able to read, but not edit, all wikis for a short period of time.

• You will not be able to edit for up to an hour on Tuesday, October 27. The test will start at 14:00 UTC (14:00 WET, 15:00 CET, 10:00 EDT, 19:30 IST, 07:00 PDT, 23:00 JST, and in New Zealand at 03:00 NZDT on Wednesday October 28).
• If you try to edit or save during these times, you will see an error message. We hope that no edits will be lost during these minutes, but we can't guarantee it. If you see the error message, then please wait until everything is back to normal. Then you should be able to save your edit. But, we recommend that you make a copy of your changes first, just in case.

Other effects:

• Background jobs will be slower and some may be dropped. Red links might not be updated as quickly as normal. If you create an article that is already linked somewhere else, the link will stay red longer than usual. Some long-running scripts will have to be stopped.
• There will be code freezes for the week of October 26, 2020. Non-essential code deployments will not happen.

This project may be postponed if necessary. You can read the schedule at wikitech.wikimedia.org. Any changes will be announced in the schedule. There will be more notifications about this. A banner will be displayed on all wikis 30 minutes before this operation happens. Please share this information with your community.

-- Trizek (WMF) (talk) 17:10, 21 October 2020 (UTC)

@Trizek (WMF):, Just wondering why this is happening right now? Does it have anything to do with COVID-19? Thanks in advance, Ottawahitech (talk) 20:26, 21 October 2020 (UTC)

This appears to be regularly scheduled test, that is done once every year: mw:Switch Datacenter --Pyfisch (talk) 21:04, 21 October 2020 (UTC)

Ottawahitech, like Pyfisch says (thanks!), it is a regular exercise. It is not related to what's going on in the news. Trizek (WMF) (talk) 11:19, 22 October 2020 (UTC)

Two 'convict' items

Can convict (Q13219330) and convict (Q989174) be merged? Which should remain? It seems to require Hungarian knowledge (which I don't have). There are some things that give me pause:

• The older item convict (Q989174) is less populated, while the newer convict (Q13219330) is more frequently edited to the extent that it is semi-protected. Does that mean the newer one should remain, not the old one?
• en:wikt:detainee says "elítélt" (Hungarian), the only sitelink of convict (Q989174), is a translation of 'detainee' (English). Does that mean the word is more comparable to prisoner (Q1862087) than to convict (Q13219330)? whym (talk) 09:49, 22 October 2020 (UTC)

QuickStatements

Hi, it often happen that I have to create several taxa items, I think, of course in the extand that it is possible, that QuickStatements could help me sometimes. What would be the code to create 4 items with:

Label: Qianguimon aflagellum; Description: species of crustacean; instance of (P31) taxon (Q16521); taxon name (P225) Qianguimon aflagellum; taxon rank (P105) species (Q7432); parent taxon (P171) Qianguimon (Q100605554); WoRMS-ID for taxa (P850) 1062895

Label: Qianguimon elongatum; Description: species of crustacean; instance of (P31) taxon (Q16521); taxon name (P225) Qianguimon elongatum; taxon rank (P105) species (Q7432); parent taxon (P171) Qianguimon (Q100605554); WoRMS-ID for taxa (P850) 1062897

Label: Qianguimon rongxianense; Description: species of crustacean; instance of (P31) taxon (Q16521); taxon name (P225) Qianguimon rongxianense; taxon rank (P105) species (Q7432); parent taxon (P171) Qianguimon (Q100605554); WoRMS-ID for taxa (P850) 1389612

Label: Qianguimon splendidum; Description: species of crustacean; instance of (P31) taxon (Q16521); taxon name (P225) Qianguimon splendidum; taxon rank (P105) species (Q7432); parent taxon (P171) Qianguimon (Q100605554); WoRMS-ID for taxa (P850) 1062898

• Note that in fact you can just give me the exemple for one item, and I will apply that to the others. Christian Ferrer (talk) 16:31, 20 October 2020 (UTC)

Presuming tab separated, then as below (maybe cut & paste from the edit screen since tabs don't render in HTML). --Tagishsimon (talk) 20:10, 20 October 2020 (UTC)

CREATE

LAST Len "Qianguimon aflagellum"

LAST Den "species of crustacean"

LAST P31 Q16521

LAST P225 "Qianguimon aflagellum"

LAST P105 Q7432

LAST P17 Q100605554

LAST P850 "1062895"

Or like this:

V1

CREATE||LAST|Len|"Qianguimon aflagellum"||LAST|Den|"species of crustacean"||LAST|P31|Q16521||LAST|P225|"Qianguimon aflagellum"||LAST|P105|Q7432||LAST|P171|Q100605554||LAST|P850|"1062895"

CSV

qid,Len,Den,P31,P225,P105,P171,P850

,"Qianguimon aflagellum","species of crustacean",Q16521,"""Qianguimon aflagellum""",Q7432,Q100605554,"1062895"

Have fun. --Premeditated (talk) 20:26, 20 October 2020 (UTC)

• Thanks you very much, I messed up a bit and I firstly created 4 empty items that I have merged. Finally I managed to use the tool correctly, e.g. Qianguimon elongatum (Q100700188). Thanks again, Christian Ferrer (talk) 21:17, 20 October 2020 (UTC)

WikibaseJS-cli template

Another way could be with Wikidata:Tools/WikibaseJS-cli, try:

wd generate-template --create-mode Q100700188 > Qianguimon.js

This generates a file:

You can then edit this into a template:

module.exports = function () { return { type: 'item', labels: { en: 'Qianguimon elongatum', nl: 'Qianguimon elongatum' }, descriptions: { en: 'species of crustacean', nl: 'taxon' }, aliases: {}, claims: { // instance of: taxon P31: 'Q16521', // taxon name P225: 'Qianguimon elongatum', // taxon rank: species P105: 'Q7432', // parent taxon: Qianguimon P171: 'Q100605554', // WoRMS-ID for taxa P850: '1062897' }, sitelinks: {} } }

module.exports = function (name, identifier) { return { type: 'item', labels: { en: name, nl: name }, descriptions: { en: 'species of crustacean', nl: 'taxon' }, aliases: {}, claims: { // instance of: taxon P31: 'Q16521', // taxon name P225: name, // taxon rank: species P105: 'Q7432', // parent taxon: Qianguimon P171: 'Q100605554', // WoRMS-ID for taxa P850: identifier }, sitelinks: {} } }

To generate similar ones, that would have been:

wd ce Qianguimon.js "Qianguimon rongxianense" 1389612

etc. --- Jura 17:50, 22 October 2020 (UTC)

Error de Wikidata en Commons

Ver aquí, no sé qué es lo que da el error. Jcfidy (talk) 06:21, 22 October 2020 (UTC)

• A bit of a mess. c:Category:Cocos nucifera gives the message "Error in Wikidata: wikidata item 'Cocos nucifera' (Q13187) has 'topic's main category' (P910)='Category:Coconuts' (Q9042343) so you should move the link to current wikicommons category from 'Cocos nucifera' (Q13187) to 'Category:Coconuts' (Q9042343)." Cocos nucifera (Q13187) is sitelinked to c:Category:Cocos nucifera. Template:Q9042343 is sitelinked to c:Category:Coconuts. - Jmabel (talk) 16:27, 22 October 2020 (UTC)

Wikidata's 8th birthday: online and onsite events

Wikidata is soon 8 years old

Hello all,

Wikidata's birthday is happening soon, on October 29th, and as every year, plenty of people and local groups got together to organize events for this occasion. Some of these events already happened, most of them will take place in the next few days. I'm pretty amazed by the diversity of the participants and the quality of the topics, and I think it is definitely worth to have a look, especially as most of these events will be accessible from remote!

Here's an overview of the upcoming events, you can find the links and details in the main calendar.

• Wikidata Cross-Domain Forum in Taipei, Taiwan and online, on October 24th
• Wikidata Eighth Birthday Celebration in Lagos, Nigeria, on October 24th: meetup, training and editing contest
• Birthday Celebration in Ulm, Germany, and online, on October 24th
• Wikidata Party in São Paulo, Brazil, and online, on October 24th: interactive presentations, activities and presents
• Several Wikidata and Wikibase-related presentations at the Italian WikiCon, on October 24-25, online
• Wikidata+OSM talk in Malayalam, online, on October 25th
• Wikidata+Wikipedia+OpenStreetMap workshop in Rennes, France, on October 25th
• WikiCite virtual conference, online, on October 26-28, with several sessions and multiple speakers in English, German, Portuguese and French
• Wikidata Education Week, online, on October 26-30
• Wikidata & Wikibase office hour, online, on October 27th: update from the development team and discussions
• 24-hours online meetup, on October 28-29: an open discussion with plenty of community members facilitating exchange around the topic "what makes you enthusiastic about Wikidata"
• Gene Wiki: how to synchronize and curate primary sources with and in Wikidata, online, on October 29th
• Birthday celebration in Rennes, France, on October 29th
• Birthday Celebration in Accra, Ghana, and online, on October 29th

For many of these events, you will be able to watch the content in replay afterwards, like for the WikidataLab videos that are available on Commons. Congratulations to all events organizers! It's awesome to see that despite the difficult context and lack of onsite opportunities, plenty of people dedicated time and energy to organize birthday celebrations <3

As usual, you can also prepare a birthday present and add it to this page, as well as birthday wishes. Cheers, Lea Lacroix (WMDE) (talk) 16:25, 22 October 2020 (UTC)

Where does the Yeti live?

Mystical creatures often stay in a region of their liking, for instance the legendary Yeren (Q2006961) lives in the moutains of Hubei (Q46862).

What property should I use to document that?

• residence (P551)
• location (P276)
• located in the administrative territorial entity (P131)
• some other property?

Thanks! Syced (talk) 00:31, 23 October 2020 (UTC)

endemic to (P183), perhaps, or taxon range map image (P181) if you have a map. Ghouston (talk) 01:51, 23 October 2020 (UTC)

Are we talk about an individual or the species of Yeti's? If it's an individual residence (P551) seems the nearest. If we are talking about the species then what Ghouston said. ChristianKl ❪✉❫ 20:37, 23 October 2020 (UTC)

mw:Help:Extension:WikibaseCirrusSearch or mw:Help:CirrusSearch

Which page would be better as an entry point for people searching on Wikidata?

Currently mw:Help:CirrusSearch is linked, but features used here are described on mw:Help:Extension:WikibaseCirrusSearch. There are probably one or two interface messages where we could update the link from one to the other. --- Jura 11:24, 23 October 2020 (UTC)

Comments and endorsements open for WikiCite plugin for Zotero grant proposal

I have posted a draft grant proposal to develop a WikiCite plugin for the open source reference management software Zotero. The idea is to add citations support to Zotero, retrieve citations data from WikiData, provide an easy way for users to fill in missing citation information and contribute it back to WikiData, and offer citation graph visualizations. I would very much appreciate it if you could post your questions/comments in the proposal discussion page. Please, endorse the proposal if you think it may be worth it. Thanks! --Diegodlh (talk)

I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. whym (talk) 11:43, 29 October 2020 (UTC)

Advice

Guys

so apparently me personally adding data about myself to Wikidata has been flagged as spam and self promotion?

Surely if i have a wikidata profile, and i add my social media links, or an article ive been featured in which is factual information, that cant be seen as spam.

How does that even work?  – The preceding unsigned comment was added by Craigcampbell0302 (talk • contribs).

I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. whym (talk) 11:43, 29 October 2020 (UTC)

Dean Court Hotel, York

• Dean Court Hotel (part) and attached railings, 1, 2 and 3 Duncombe Place (Q26549206)
• Dean Court Hotel (part) and railings attached to front steps, 4, 5 and 6, Duncombe Place (Q26549207)
• c:Category:Dean Court Hotel, York

The Dean Court Hotel in Duncombe Place, York, England, comprise two distinct groups of buildings: numbers 1, 2 and 3 Duncombe Place (Q26549206) and numbers 4, 5 and 6 Duncombe Place (Q26549207). These are two groups of what were houses that are now parts of the hotel. Each group is separately identified as being buildings of heritage interest and Wikidata has two separate entries for them. So far, so good. :)

Commons has one category for the Dean Court Hotel: c:Category:Dean Court Hotel, York. This category is included on (Q26549206) under 'other sites' but I am unable to include it on (Q26549207) as it is already being used. What is the solution here please? Thank you, --Malcolmxl5 (talk) 19:14, 17 October 2020 (UTC)

Make a new Wikidata item for the whole thing, and list the other items as parts? Ghouston (talk) 03:58, 19 October 2020 (UTC)

An item for the function (the hotel) and separate items for the buildings as parts, that could work (I’ve realised also there’s a third building that is part of the hotel: 21, High Petergate (Q26548945)). I shall have to write an en.wiki article on the hotel and buildings, I think. :) Thanks, --Malcolmxl5 (talk) 19:03, 20 October 2020 (UTC)

The relationship between the hotel and the buildings is probably occupant (P466). Ghouston (talk) 08:09, 23 October 2020 (UTC)

Yes, that would logical. --Malcolmxl5 (talk) 17:12, 24 October 2020 (UTC)

John Fraser vs John Fraser

Hello, I discovered there is some confusion between John Fraser (Q1700123) and John Fraser (Q94122683) but the situation is more complicated than usual so I am asking for second opinion. In my view, identifiers in John Fraser (Q1700123) (VIAF: 14256973) point to a writer born in *1932 rather than to an actor born in 1931. However, the interwiki links all point to the actor. The item storing correct actor's identifiers is John Fraser (Q94122683) (VIAF: 41145971492532331970). Should we keep John Fraser (Q1700123) for the actor and swap the identifiers, or move interwiki links to John Fraser (Q94122683)? Vojtěch Dostál (talk) 07:36, 23 October 2020 (UTC)

I'd keep John Fraser (Q1700123) as the actor, given the descriptions, sitelinks, birth date and occupation. From the history it seems that appropriate VIAF ids have been present at times, namely 20774607 (which is now a redirect) and 41145971492532331970, but people have removed them. I'd put them back, and deprecate anything that refers to somebody else. Ghouston (talk) 07:58, 23 October 2020 (UTC)

• See Help:Conflation_of_two_persons. I'd probably end up having John Fraser (Q1700123) deleted. --- Jura 0 8:19, 23 October 2020 (UTC)
  • @Jura1: Why? Both people are identifiable concepts.Vojtěch Dostál (talk) 09:58, 23 October 2020 (UTC)
    • Ah, because it's a conflation and as such this QID should be discontinued? Vojtěch Dostál (talk) 09:59, 23 October 2020 (UTC)
      • yes. People using it can't really be sure who it's about. --- Jura 10:04, 23 October 2020 (UTC)

OK, I created John Fraser (Q100741986) and John Fraser (Q100741993). Q1700123 is now redirected but I will nominate it for deletion once the links from other Wikidata items are fixed by bot.Vojtěch Dostál (talk) 13:00, 23 October 2020 (UTC)

• Rather than delete it I would convert it into an entry similar to Christopher Turner (Q5113349). In case third parties have are using the entry as an external link already. --RAN (talk) 19:39, 23 October 2020 (UTC)

The community can decide whether to keep Q1700123 as a conflation, or delete it, I have no firm opinion as to whether it is in use anywhere. --RAN (talk) 19:16, 24 October 2020 (UTC)

Having issues with trial parameters

I've been trying to improve Wikidata's of persons that have been criminaly charged, but I'm having a hard time figuring out how to add that they've been acquitted of their crime, or pardoned from their penalty, or taken and Alfrod plea later on after conviction.*Treker (talk) 04:21, 24 October 2020 (UTC) I'm also not too fond of that general concept of "innocence" and the legaly defined plea of "not guilty" are the same item.*Treker (talk) 04:24, 24 October 2020 (UTC)

Merge items

Could someone make sure Q9211385, Q9068341 and Q33124656 get merged? This did not seem to have happened automatically before. --Glennznl (talk) 13:31, 24 October 2020 (UTC)

 Done Michgrig (talk) 15:44, 24 October 2020 (UTC)

Ontology fix for "badges"

All digital badges (including Wikimedia badges) are now appearing as <instance of> "clothing" because up the tree, they are subclasses of badge (Q336533). I'm not sure where (or how) to break this chain - any ideas? Clearly all (notional) badges are not clothing accessories. - PKM (talk) 22:43, 22 October 2020 (UTC)

Welcome back to Wikipedia categories ;) --- Jura 08:21, 23 October 2020 (UTC)

After chewing on this for a couple of days, I think I have it sorted out now. - PKM (talk) 23:07, 24 October 2020 (UTC)

Andreasmperu never changes

https://www.wikidata.org/wiki/User_talk:Andreasmperu/2019/2016-2017#Q154242 https://www.wikidata.org/w/index.php?title=Q154242&diff=1296596047&oldid=1294905427

They don't seem to care at all. 88.147.188.20 13:27, 24 October 2020 (UTC)

• Anonymous personal attacks should not be given much weight. If you want to criticize another user, how about setting up an account so that we know more about who you are? Otherwise, content issues should be discussed as content issues and likely on the talk page in question. ChristianKl ❪✉❫ 22:16, 24 October 2020 (UTC)

Popes

At Talk:Q19546 we have a table of the popes, but there was an import from the Catalan Wikipedia that caused all the early popes to be duplicated, because the Catalan Wikipedia used dates that are slightly off from the canonical dates used by the Vatican. Is there a way to automatically delete them? Or should they be deprecated so that the bot does not add the information again, and the duplication restored? I would prefer deletion and rely on the canonical dates supplied by the Catholic Encyclopedia. --RAN (talk) 19:30, 24 October 2020 (UTC)

• Is it a transcription error of cawiki or are they dates also found elsewhere? In the later case, it's better to mark preferred whatever is preferred. In the first case, maybe it's worth to tell cawiki about it. --- Jura 13:26, 25 October 2020 (UTC)

@Jura1: I see what the problem is, one source is using the day of the election as the start date of being a pope, and the other source is using the date of the inauguration as the start date for being a pope. The modern popes are all using one date, but the early ones are using a mixture of the two dates, so we are getting two entries. I do not know which one is correct. For a US president the authority does not begin until the inauguration, but you only get a new pope when the old one is dead, so the authority may begin at the election, and the inauguration just a formality. I will write a professor at University of Notre Dame and see if I can get a response. --RAN (talk) 19:39, 25 October 2020 (UTC)

Horsedrawn tram or railway?

Q874504 is an example of an horsedrawn railway. However Q832003 is also used for typical Horse-drawn trams (see Commons:Category:Horse-drawn trams). I suggest that a seperate data item is created for 'Horse-drawn trains' typicaly used by very early railways such as Q874504.Smiley.toerist (talk) 11:59, 25 October 2020 (UTC)

Political Graveyard

Can someone peek at Property:P8462, the links do not work. I think the scheme has changed for the database. --RAN (talk) 00:59, 23 October 2020 (UTC)

@Richard Arthur Norton (1958- ): User:GZWDer adjusted the formatter URL to bypass the Wikidata UI problem. It takes a little while for that to come into effect; if you purge a page with one of these id's now it should work though. ArthurPSmith (talk) 16:20, 23 October 2020 (UTC)

OK, I'll wait and check tomorrow, btw, what is the secret to getting the new scheme? When I visit the website via Google, all I get is a page for a surname. --RAN (talk) 19:47, 23 October 2020 (UTC)

@GZWDer: I see what the problem is, the Unicode for # is being interpreted as ASCII code for #. See that "archer.html#811.30.85" is converted to "archer.html%23811.30.85" when it becomes a URL, giving an error. Click on the two examples at Property:P8462 and see if it happens to you. The error must be in the "format as a regular expression" formula. --RAN (talk) 02:17, 25 October 2020 (UTC)

The problem is the UI automatically URL-encodes all special characters in any id. The external-id tool undoes that encoding and forwards you to the right place. ArthurPSmith (talk) 16:50, 26 October 2020 (UTC)

Golden Song by Harry Styles

Golden (Q100701128) and Golden (Q91250107) both relate to the song "Golden". I don't know the correct modeling for compositions and releases, so maybe they should stay separate. Can someone please complete these items? By the way they were both targeted by vandals, I requested semi-protection. --Pyfisch (talk) 09:52, 25 October 2020 (UTC)

I've merged the two items because they are evidently about the same object - a song that is part of an album. Michgrig (talk) 14:47, 25 October 2020 (UTC)

@Michgrig: That is not correct. Per Wikidata:WikiProject Music overview the track should not be merged with the release. Golden (Q100701128) is the single and Golden (Q91250107) is the track that is on the single and album. - Premeditated (talk) 10:17, 26 October 2020 (UTC)

@Premeditated There was no indication that Golden (Q100701128) is a single, otherwise I wouldn't have merged these items. If you take a look at previous versions, you'll see that the former was about a song from an album. If there are articles about both a song and a single, my merge can be safely reverted. Michgrig (talk) 11:30, 26 October 2020 (UTC)

@Michgrig: If you now look at Golden (Q100890626) and Golden (Q91250107) there is a clearly difference between composition and vocal track. Another example of why this modeling is favorable:

|--Stardust (Q91190107): The composition

|--Stardust (Q1543022): The lyrical adaptation of the song

|--Stardust (Q93717770): The vocal track by Willie Nelson

|--Stardust (Q94544075): The vocal track by Rod Stewart

--Premeditated (talk) 14:02, 26 October 2020 (UTC)

Wikidata weekly summary #439

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

• Events
  • Past: Providing sustainable data services through Wikibase and Wikidata, SEMIC Conference 2020 - (replay)
  • Past: Introduction to Wikidata by Frimelle, Software Freedom Kosova 2020 - (replay)
  • Past: What's new in Wikidata? by Lea Lacroix, Lydia Pintscher and Mattia Capozzi, ItWikiCon 2020 - (replay)
  • Ongoing: plenty of Wikidata birthday events, online and onsite! Here's an overview of the upcoming events, you can find the links and details in the main calendar.
    • WikiCite virtual conference on October 26-28, live-streamed on Youtube and Twitter, with sessions across many timezones and languages – English, French, German, Indonesian, and Portuguese.
    • Wikidata Education Week, online, on October 26-30
    • 24-hours online meetup, on October 28-29: an open discussion with plenty of community members facilitating exchange around the topic "what makes you enthusiastic about Wikidata"
    • Gene Wiki: how to synchronize and curate primary sources with and in Wikidata, online, on October 29th
    • Birthday celebration in Rennes, France, on October 29th
    • Birthday Celebration in Accra, Ghana, and online, on October 29th
  • Upcoming: Wikidata & Wikibase office hour, October 27th at 17:00 CET (16:00 GMT)
  • Upcoming: live SPARQL queries on Twitch and in French by Vigneron, October 27 at 18:00 CET

• Press, articles, blog posts, videos
  • Excerpts from Wikidata Birthday events
    • Video: How the Wikimedia Brazil contributed to Wikidata in 2020 (in Portuguese) - YouTube
    • Video: Introduction to Wikidata and data reconciliation (in Portuguese) - YouTube
  • Analyze Swedish politics with Wikidata (Blog post)
  • WikiCite awards 23 grants & eScholarships to improve open citations (including plenty of Wikidata-related projects)
  • Video: Eyoungstrom and Evolution and evolvability explore and explain different Wikidata tools in 15-part series on YouTube - Playlist
  • Video: Wikidata Lab XXV: Interoperability and data access (in Portuguese) - YouTube
  • Video: Wikidata for cultural institutions and GLAM-WIKI partnerships (in Portuguese) - YouTube
  • Wikidata Training Workshop 4, by Canadian Arts Presenting Association - YouTube
    • Wikimedia Commons and Wikidata - (Parts 1, 2, 3, 4 & 5)
    • Wikimedia Commons and Wikidata (in French) - (Parts 1, 2, 3, 4 & 5)

• Tool of the week
  • WikibaseJS-cli allows to edit Wikidata from command line, including creating items from samples ("templates") and do complex transformations of statements

• Other Noteworthy Stuff
  • Server switch: all wikis will be in read-only on Tuesday, October 27th at 14:00 UTC for up to an hour (more information). All services hosted by WMF (e.g. tools hosted on wmcloud.org) are also impacted. Non-essential code deployments will not happen this week.
  • Wikimedia Israel (WMIL) develops instructional material for the Wikidata Query Service (WDQS): wdqs-tutorial.toolforge.org

• Did you know?
  • Newest properties:
    • General datatypes: number of rooms, permits, prohibits, economy of topic, check-in time, check-out time
    • External identifiers: CTAN package ID, Dresden Street ID, Emojipedia ID, FIBA Hall of Fame ID, Stadtwiki Dresden article, GADM ID, Lambiek Special pages ID, Mir@bel publisher ID, ILO Thesaurus ID, AVN performer ID, CCFr library ID, xHamster performer ID, AWMDB performer ID, Normattiva ID, American Heritage place ID, BHL name ID, Hungarian Water Polo Federation player ID, Irish Statute Book ID, Minecraft UUID, Nachlässe in Austria ID, AniList anime ID, National Park Service people ID, AniList manga ID, Archivio Storico Ricordi opera ID, O*NET OnLine code, Opta football competition ID, Opta football player ID, Opta football team ID, The Boardr profile ID, Great Places ID, PIM award ID, Scholars Strategy Network ID, DC Historic Sites place ID, Rheinland-Pfälzische Personendatenbank ID
  • New property proposals to review:
    • General datatypes: annual average daily traffic, isotopically modified form of, Numérisé par, Online catalog, funeral, Inaugural address, Hong Kong film rating, word lookup, hair or facial hair style, Cup size, for color scheme
    • External identifiers: aagm people ID, aagm sites ID, Yandex Zen ID, GeoGuessr ID, Humanitarian OpenStreetMap Tasking Manager project ID, Polski Słownik Judaistyczny ID, SMB-digital ID, identifiant station Vélib' Métropole, Base constructions bibliothèques ID, Dicionário de Historiadores Portugueses ID, Unione Romana Biblioteche Scientifiche ID, Malmö Museer, IMVDb music video ID, NSSDCA ID, Docker Hub repository, Ontology of units of Measure 2.0 unit ID, Persée article ID, Geneanet Family Name ID, Blogger user profile ID, USL League One ID, Stolperstein identifiers, State Catalogue of the Museum Fund of Russia artwork ID, Royal Horticultural Society plant ID, EGAFD ID
  • Query examples:
    • Commemorative plaques and their subjects in Aberdeen, Scotland; with photos of plaque and subject and Wikipedia links. (Source)
    • Cities with female mayor (Source)
    • Most common variants of the name "Susanne" in each country (Source)
    • Birthplaces of art historians (Source)
    • Invasive species in EU and whether they have articles in different languages (Source)
    • Birthplace of Olympique de Marseille players (Source)
    • COVID-19 deaths by month (Source)
  • Wikimedia Commons Query examples:
    • Get information of Europeana item using federated query

• Development
  • Got the Query Builder to create the first simple queries. You can follow along the development from now on on the test system.
  • Preparing regular JSON dumps for Lexemes (phab:T264883)
  • Analyzed a number of queries to the query service together with the WMF search team to better understand what the queries are doing and what further optimizations we can make to the query service and which queries need different systems to answer them.
  • Working on creating and tracking implicit usages of Wikidata descriptions so description changes show up in the watchlist on Wikipedia and co (phab:T265897)
  • Created Extension:WikibaseManifest documentation on MediaWiki.
  • Looking into what would be considered a mismatch when comparing WIkidata's data against another database. This is groundwork for potentially automating this in the future to find potential issues in the data to flag to editors.
  • Working on deploying the new Item quality scoring model for ORES to production so we have improved quality scores
  • Finished investigating how to handle a restricted list of external Wikibase services. (phab:T265118)
  • Removed noratelimit on bots for wikidata.org (phab:T258354)
  • Published Easier Access for Programmers to Wikidata research report

You can see all open tickets related to Wikidata here. If you want to help, you can also have a look at the tasks needing a volunteer.

• Monthly Tasks
  • Add labels, in your own language(s), for the new properties listed above.
  • Comment on property proposals: all open proposals
  • Contribute to a Showcase item.
  • Help translate or proofread the interface and documentation pages, in your own language!
  • Help merge identical items across Wikimedia projects.
  • Help write the next summary!

Read the full report · Unsubscribe · Mohammed Sadat (WMDE) 16:00, 26 October 2020 (UTC)

Wikiversity:_Lua_error_in_Module:Citation/CS1

What needs to happen to help fix a problem accessing Wikidata from Wikiversity?

Several pages in Wikiversity:Category:Freedom and abundance contain {{cite Q|...}} references that display something like the following:

Lua error in Module:Citation/CS1/Identifiers at line 1148: attempt to compare nil with number., Wikidata Q55670016

This appears, for example, in Note 3 in Wikiversity:Media and corruption#Notes.

I raised this question in Project Chat on 2020-08-19; see Wikidata:Project chat/Archive/2020/08#Wikiversity: Lua error in Module:Citation/CS1/Identifiers at line 1014: attempt to compare nil with number., Wikidata Q36472504. I also asked about it on Wikiversity; see Wikiversity:Module talk:Citation/CS1/Identifiers. Thanks, DavidMCEddy (talk) 05:11, 27 October 2020 (UTC)

Cathédrale or cathédrale?

VIGNERON reverted my edition of changing "cathédrale Saint-Lazare d'Autun" for "Cathédrale Saint-Lazare d'Autun" at Q611944. The issue is the capitalization of an noun. The rules at [12] say that "...however proper nouns such as the names of specific people, specific places, specific buildings, specific books, etc., should be capitalized." "Cathédrale Saint-Lazare d'Autun" is the name of an specific building. So, I think the name cathédrale in this case (and similar) should be capitalized. GualdimG (talk) 07:25, 27 October 2020 (UTC)

• If it's about the French label, please use Help:Label/fr. --- Jura 07:33, 27 October 2020 (UTC)

Hi @GualdimG:,

The names of specific buildings are "cathédrale Saint-Lazare d'Autun"@fr and "cattedrale di Autun"@it. Not entirely sure for italian but 100% sure for French. Take a look at the wikipedia articles, the introduction starts with "cathédrale Saint-Lazare d'Autun" and "cattedrale di San Lazzaro".

Cheers, VIGNERON (talk) 07:38, 27 October 2020 (UTC)

Ok, the French rules about nouns are different from the English. Thanks for the answer. Cheers, GualdimG (talk) 08:17, 27 October 2020 (UTC)

Idea for item update

You found a good idea to add product or material produced or service provided (P1056) on Nokia (Q1418)? --2001:B07:6442:8903:7174:8606:ACD2:282E 17:04, 27 October 2020 (UTC)

Ranks messed up in this one item

Hiya! Could some gentle soul perhaps straighten the ranks of the software version identifier (P348) in Q21041255. Now 2.7.4 would seem to get the lime light instead of newer versions. I would do it myself except it can't be done without JavaScript I think... --Palosirkka (talk) 20:57, 27 October 2020 (UTC)

@Palosirkka, it has nothing to do with JavaScript. I changed the rank of version 2.7.4 from preferred to normal, and the rank of the latest version from normal to preferred. For more details, please see Help:Ranking. Michgrig (talk) 06:26, 28 October 2020 (UTC)

@Michgrig: Thanks for changing the pref rank to normal. I wonder if setting the latest to preferred will cause the same problem once a new version appears. Or are they supposed to go change the rank of the other older release. Should they all be normal? Ranks have nothing to do with JS but editing Wikidata has. Try it. :) --Palosirkka (talk) 10:35, 28 October 2020 (UTC)

@Palosirkka: "I wonder if setting the latest to preferred will cause the same problem once a new version appears." - Yes, it will. As soon as a new version appears, the ranks must be changed. This way, Wikipedia articles will display only the latest version. If all versions are normal, then all of them will be displayed in Wikipedia. Michgrig (talk) 13:26, 28 October 2020 (UTC)

Carthage

There seems to be a high degree of conflation between Carthage (Q6343) and Ancient Carthage (Q2429397). Do we have a subject-specialist who could sort them out, and make the distinction between them clearer? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:36, 28 October 2020 (UTC)

Updating SPARQL federation input review process

Tracked in Phabricator
Task T265290

wikidata:SPARQL federation input page has been abandoned for some time and on October 8 @Multichill: marked page as historic to make it visible. (discussion) After that, the phabricator ticket phab:T265290 (Review the federation input process for WDQS) was created.

The phabricator ticket defines as a target that adding new SPARQL-endpoints to Wikidata Query Service requires community input on the process. Also, the Search Platform team should not be in charge of deciding which request makes sense and which does not.

After the community decides, there may be technical reasons why and when new endpoints will be added, but defining the community process is one step going forward with the ticket.

One simple process could be (more or less the same as before)

1. Add proposal of the new endpoint to the wikidata:SPARQL federation input page
2. Community would review the proposal
3. If rejected then mark the rejection to the proposal
4. If approved then the community will create phabricator ticket for a SPARQL endpoint to be enabled where the relevant information is described including a link to the discussion.
5. When phab ticket is ready and the endpoint is working mark the status of the proposal

--Zache (talk) 09:33, 25 October 2020 (UTC)

Thanks for picking this up. I think the criteria for acceptance should be made very clear to avoid disappointments or needless discussions:

1. The data in the SPARQL endpoint should be licensed under a free license. Rule of thumb: If Commons accepts the license, you're good. A link to the license should be included in the proposal so it's easy to verify it
2. The data in the SPARQL endpoint is somehow relevant to combine with Wikidata. Just to prevent people from nominating every free SPARQL endpoint because they can.

Maybe model the flow of requests like Wikidata:Requests for permissions/Bot? Multichill (talk) 10:51, 25 October 2020 (UTC)

I agree with #1, though it is not always so easy to see what is in the endpoint without actually looking into it. (examples: Wikidata:SPARQL_federation_input#European_Environment_Agency_SPARQL_endpoint) About #2, if we are gonna create separate pages for approvals it could be more useful to just to archive the discussion part to the talk page for example, and convert the rest of the content as the living documentation of the endpoint. So editors could add new examples and update links to documentation etc. Currently, the biggest blockers for using federation and SPARQL, in general, is that it is hard to know what is in the endpoint and how to access it. --Zache (talk) 11:27, 25 October 2020 (UTC)

Multichill Is there a bit more reasoning behind the license requirement? I ask because it is distinctly different than Wikidata's license requirement (CC0). If we allow federation to depart from that, would there be any value in going further, such as allowing "non-commercial" licenses as well? That would obviously expand the possibility of such a service, but I feel uneasy about it. Matthias Winkelmann (talk) 14:01, 28 October 2020 (UTC)

The reasoning behind this licensing restriction is that Wikimedia websites should only offer free content. Multichill (talk) 17:35, 28 October 2020 (UTC)

If you wonder what federated input or queries mean then the simple answer would be that those are database queries that can use multiple databases simultaneously. The common use case in Wikimedia is to query both Structured Data on Commons and Wikidata and combine the results. However, it can be other sites too. I wrote an example query to Commons for a query which uses Wikidata, Commons structured data, Europeana (Q234110) and Finto (Q61677804) for reading subjects of photo from Europeana and then converting them to Commons categories. In the Wikipedia context example, it can be used for reading numerical data like population time series which are currently too complex to store in Wikidata from dedicated services like Eurostat (Q217659). --Zache (talk) 08:10, 26 October 2020 (UTC)

Error adding date and time

Tracked in Phabricator
Task T57755

Hello. Could someone familiar with debugging Wikidata errors have a look at Property talk:P585#Adding date and time, please? I get an error when trying to add date and time using point in time (P585). Rehman 06:06, 27 October 2020 (UTC)

Is there a constraint somewhere that I'm overlooking? Rehman 03:49, 28 October 2020 (UTC)

See phab. --Jklamo (talk) 11:22, 28 October 2020 (UTC)

Thanks, Jklamo. I had always thought we could already do this... Rehman 02:04, 29 October 2020 (UTC)

Exists edit interval limit?

I couldn't find policy that specifying edit interval limit per time-unit (have searched 'insource:"edit interval"'). Could someone help me with it? Thanks.--Semi-Brace (talk) 12:33, 27 October 2020 (UTC)

8 edits/min for IPs and Newcomers (i.e. not yet "autoconfirmed"), and 90 edits/min otherwise. The limitation is technically enforced by MediaWiki, and the values are default values for Wikimedia projects. —MisterSynergy (talk) 12:54, 27 October 2020 (UTC)

Thanks for telling me that, I'm glad to hear. --Semi-Brace (talk) 12:58, 27 October 2020 (UTC)

In addition, for uses of the API, there's a maxlag parameter which signals a requirement for the user (normally a bot) to decrease its edit rate when load on the servers is too great - see e.g. https://www.mediawiki.org/wiki/API:Etiquette#Use_maxlag_parameter --Tagishsimon (talk) 22:49, 28 October 2020 (UTC)

Can I edit with this account? (alternate account)

I've made this alternate account with the purpose to be used on my streaming computer. I'd like this account to be limited from the beginning. I will link to my main username User:Datariumrex and I've linked back from it. RustProton (talk) 17:16, 27 October 2020 (UTC)

There is no problem with having multiple accounts, so long as you don't use them to evade a block/ban or try to manipulate a consensus discussion by using the accounts to support each other. The fact that you have made an open statement on the user page of each account about the link is a positive. From Hill To Shore (talk) 17:35, 27 October 2020 (UTC)

I do see this as a valid use for an alternative account.

@From Hill To Shore: Given our current rules making an open statement on the user page is not only positive but required. ChristianKl ❪✉❫ 15:23, 28 October 2020 (UTC)

It is a positive that the user has made an open and frank statement about the link prior to seeking clarification on whether what they are doing is allowed. I have encountered other editors in the past who have not been as honest. From Hill To Shore (talk) 23:08, 28 October 2020 (UTC)

Can't set ISQ dimension property

Tracked in Phabricator
Task T266496

I've just tried to set the ISQ dimension (P4020) of specific impulse by mass (Q100793317) to "\mathsf{L} \mathsf{T}^{-1}" (that is to say, the ISQ dimensions of velocity in math markup). However, when attempting to publish the change, I'm getting the message "Could not save due to an error. Bad value type unknown, expected string". The property ISQ dimension (P4020) has datatype "Mathematical expression", and the value I'm using certainly looks like a valid expression: it's directly cut-and-pasted from the "ISQ dimension" field of velocity (Q11465).

I've done a bit of testing, and seem to get the same error regardless of the string I enter, whether or not it's a valid math formula. I also get similar results if I try to set defining formula (P2534), so this might possibly be a problem with other types of math expression property.

What's going wrong here? Can anybody help me with this, please? -- The Anome (talk) 07:47, 28 October 2020 (UTC)

• @The Anome:} Seems to be a problem for Wikidata:Contact_the_development_team. --- Jura 08:05, 28 October 2020 (UTC)
  • @Jura1: Thanks, I've now posted this message there as well, and hope someone will take a look soon. -- The Anome (talk) 08:07, 28 October 2020 (UTC)

I had filed the same issue yesterday: https://phabricator.wikimedia.org/T266620 Toni 001 (talk) 09:28, 28 October 2020 (UTC)

Thanks. This is definitely a blocker for math-related tasks. -- The Anome (talk) 18:18, 28 October 2020 (UTC)

Type @ in search

If I type @ in search bar on left site then ill getreuslts but If I type it in Search ill get no resuls and also why Wikipdia search is so bad? Eurohunter (talk) 14:39, 28 October 2020 (UTC)

1. They're two distinct systems, the first capable of invoking the second. 2. CirrusSearch is rather good, but like any search technology, it is not great at everything. It will not search for metacharacters without reaching for regex, afaics, and even then it finds it all a bit hard. More info here: https://www.mediawiki.org/wiki/Help:CirrusSearch

By all means come back with issues that are not sweeping generalisations if you want sensible discussion. --Tagishsimon (talk) 22:44, 28 October 2020 (UTC)

Automated finding references: dump and dashboard

Hello all,

As you may know, in May 2020 we released new data for automated references, as well as a game that you can use to associate references with statements. We released this game containing 4200 potential references (see statistics). In the meantime, we parsed many more websites and collected 529K potential new references.

These new references will not be added to the game, because they are too many for their relevance to be checked by hand. As requested by some of you after the previous announcement, we published the list of all references in a dump available here.

Subsets of this dump can be reused by bots and tools, however, we advise you to be careful when using it and to not mass import them to Wikidata without careful review: it is quite raw, some references may be wrong or irrelevant. In order to help you analyze these references and filtering the most useful ones, we are also providing a dashboard containing an overview of the judgements made in the game so you can see which parts are more likely to be of higher or lower quality.

We’re happy to release the dumps and the dashboard just in time for the Wikidata birthday :)

If you have any questions or encounter issues with the dump or the dashboard, please let us know on the talk page. Cheers, Lea Lacroix (WMDE) (talk) 08:56, 29 October 2020 (UTC)

Better Item quality judgments from ORES

Hello all,

Wikidata’s content is growing and our data is used in more and more high-profile places. This means the pressure around data quality is rising. We want to provide people with good data. One important piece in the data quality puzzle is being able to understand where we currently stand quality-wise and how that changes over time. We need to be able to do this at scale and in an automated and repeatable way because no-one of us wants to do this by hand for 90 Million Items for sure.

That’s where ORES, the machine learning system, comes in. One of the things it can do is judge the quality of an Item. Or to be more exact it can judge some aspects of the quality of an Item. It puts each Item into a quality class between A (amazing) and E (ewwww, terrible). It’s been doing this for a while already but the quality judgments it provided were not very good. The reasons for this were that it took only a limited number of signals into account (that’d be something like the number of References on the Item or the number of Labels) and because it was trained on rather old data. Since then Wikidata’s data has changed a lot so ORES could not tell what to do with the new kinds of Items like astronomical objects because it had never seen them before.

We wanted to improve that and make the quality judgments ORES provides better. We did this by:

• adding a number of new signals (e.g. does this Item have an image attached)
• changing existing signals (e.g. missing references on external ID statements no longer punish the Item so much)
• retraining the model on more current data so it better understands scientific papers, astronomical objects, etc.

While we were at it we also wanted to better understand how data quality changes over time on Wikidata. Before we only looked at the global average quality score. But how do Items change over time? How many Items are being improved from D to C or even B class for example? To better understand this we started creating diagrams like the following one. It shows the development from January 2019 to January 2020.

We’re happy to present these improvements for Wikidata birthday, and we hope this will help us get a better and more accurate view of the data quality on Wikidata now.

If you want to see the quality score near the header on each Item you can include the following user script in your Common.js page: importScript("User:EpochFail/ArticleQuality.js")

What’s coming next on the same topic?

• ORES can’t judge all aspects of quality. It for example can not tell if a statement is generally considered true. We will look at ways of judging this aspect of quality as well but it’s considerably harder. If you have ideas how to go about it let us know.
• We will build a small tool that’ll make it possible for you to provide a list of Items and then get the quality of that subset of Wikidata as well as the lowest and highest quality Items. This will hopefully help wiki projects etc to have a good overview of their data.

If you have any questions or feedback, or want to keep discussing about Item quality, feel free to use this talk page. Cheers, Lea Lacroix (WMDE) (talk) 09:03, 29 October 2020 (UTC)

24-hours online meetup for the Wikidata birthday

Hello all,

As a reminder, the 24-hours online meetup to celebrate Wikidata's birthday is starting today at 17:00 UTC/GMT (18:00 in Central Europe), in a bit less than two hours.

This informal discussion is open to everyone who would like to chat about Wikidata and virtually meet other people from the community. The main topic is "what makes you feel enthusiastic about Wikidata" and some people will facilitate certain slots with a specific topic. You can find more information here.

You're welcome to jump in at any point that is convenient for you! To access the call, you will first need to register here (free of charge, valid email address needed). An individual link to the BigBlueButton call will then be sent to you one hour before the beginning of the event. (the service provider complies to GDPR and the email address you provide will be deleted after the event, we will not access or reuse it in any way)

We're looking forward to chat with you and to learn more about your favorite projects, tools or anecdotes on Wikidata!

If you encounter any technical issue, feel free to contact me by email. Cheers, Lea Lacroix (WMDE) (talk) 15:04, 28 October 2020 (UTC)

Can't join the call, but the birthday logo is super cute! Matthias Winkelmann (talk) 20:26, 28 October 2020 (UTC)

I registered about an hour ago. I received an immediate confirmation of my order, but no joining link for the BigBlueButton event. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:44, 29 October 2020 (UTC)

Idea for item update

I wish to add on the item Nokia (Q1418) the section product or material produced or service provided (P1056) and add all the products made by Nokia , for example Nokia 5110, Nokia 3210, Nokia 7110 and the other Nokia models. It is eligible for Wikidata? --2001:B07:6442:8903:808B:BAD6:9C9A:7BF8 17:40, 28 October 2020 (UTC)

Sucky idea. Much better to create items for the company's products. As has already been done. Nokia 5110 (Q2016083) &c --Tagishsimon (talk) 22:16, 28 October 2020 (UTC)

@Tagishsimon: Apple (Q312) also has the section product or material produced or service provided (P1056). Why Apple yes and Nokia no? --2001:B07:6442:8903:6DEE:8CDE:9DC7:6FC9 09:18, 29 October 2020 (UTC)

It's a useful thing do do from a pespective of redundancy. But the cost is bloated items which do not play well with the RDF serialisation process, besides making the items a PITA to navigate. Given that the data is redundant, because it is (or can be) represented in discrete child items, the question would be, why do it? There are, I'm afraid, all sorts of properties for parent items that are more sanely catered for by pointing child items at the parent, given the technical constraints under which wikidata labours. Being able to point to examples of unnecessary bloat is not a good reason for adding more bloat. Clearly, mine is just a view, and if you want to add cruft to an item, knock yourself out, but don't kid yourself that it's unambiguously the case that you're improving wikidata. --Tagishsimon (talk) 13:28, 29 October 2020 (UTC)

Empty ORCID profiles

Is Zhijun Zhu (Q90177186) https://orcid.org/0000-0001-6376-5041 useful at all? There's ZERO info about the ORCID profile.--RZuo (talk) 22:12, 28 October 2020 (UTC)

Yes. https://www.wikidata.org/wiki/Special:WhatLinksHere/Q90177186 --Tagishsimon (talk) 22:14, 28 October 2020 (UTC)

@RZuo: External services (PubMed Central, Crossref) often have ORCID id's for works provided by publishers, which have not been added to the person's ORCID profile. ArthurPSmith (talk) 13:19, 29 October 2020 (UTC)

detect items with P2002 without P580 as a qualifier on SPARQL

How can I find them? I saw the guides, but couldn't find. Thank you. --Semi-Brace (talk) 13:07, 29 October 2020 (UTC)

SELECT ?item ?itemLabel ?P2002
WHERE 
{
  ?item p:P2002 ?stat. 
  ?stat ps:P2002 ?P2002 .
  filter not exists {?stat pq:P580 [] . }        
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}

Try it!

Note also Wikidata:Request a query. --Tagishsimon (talk) 13:15, 29 October 2020 (UTC)

Thanks fast respond, I'll use that page for next time.--Semi-Brace (talk) 13:18, 29 October 2020 (UTC)

Save the date! WikidataCon distributed conference on 29-30-31 October 2021

Hello all,

On this day of Wikidata’s birthday, I’m excited to announce some news about the next edition of the WikidataCon.

The WikidataCon is an event organized by Wikimedia Germany and focused on the Wikidata community in a broad sense: editors, tools builders, but also 3rd party reusers, partner organizations that are using or contributing to the data, the ecosystem of organizations working with Wikibase. The content of the conference will have some parts dedicated to people who want to learn more about Wikidata, some workshops and discussions for the community to share skills and exchange about their practices, and some space left to include side events for specific projects (WikiCite, Wikibase, GLAM, etc.).

During the first two editions, we gathered an international crowd in Berlin for a few days. However, as the global COVID pandemic is still hitting the world, and the forecast for 2021 doesn’t indicate much improvement, the situation doesn’t allow us to plan a traditional onsite international conference. In order to allow the event to take place, and to ensure the safety of all participants, we had to make some clear decisions. In 2021, we will not gather all participants in Berlin, and we will avoid any international travel. Instead, we are experimenting with a hybrid format for the conference: most of the content and interactions will take place online, and small, local gatherings will be possible, if the situation allows it.

In a similar way to the distributed Wikidata birthday events, we will encourage Wikimedia chapters, local groups and communities to organize their part of the event, to support and gather people inside their country, and to contribute to the content, for example with running talks, workshops and discussions. Wikimedia Germany will provide the technical infrastructure and support the coordination of these distributed events. We’re already inviting Wikimedia organizations to include the event in their 2021 plan (especially if you’re requesting an APG), and to reach out to me if you want to be involved in the distributed conference.

On top of this, we would like to partner with a Wikimedia organization outside of Europe/North America, to strengthen the Wikidata community in their country and to allow a fairer distribution of content and speakers. We are currently evaluating several possibilities, focusing on groups who have been very active with Wikidata-related events over the past years. Our criteria include not only the motivation and past activities, but also the ability of the group to support part of the WikidataCon organization workload. The choice of the partner organization will be announced at the end of November. If you would like to know more about the process, feel free to contact me off-list and I’ll be glad to give you some details.

In these quite unpredictable times, organizing an event is a challenge, and the WikidataCon 2021 will definitely be one of a kind: we will have to adapt, to be agile, to make the best out of the situation, and to work closely with the local Wikidata groups all over the world.

Since the first WikidataCon in 2017, I am deeply committed in providing a nice experience for everyone and making sure that this event remains a gathering for and by the Wikidata community. I’m really excited to coordinate this project and run this new experiment with you all!

If you’re interested in joining the effort, if you have questions, suggestions for formats, etc.: feel free to use this talk page or to reach out directly to me.

Cheers, Lea Lacroix (WMDE) (talk) 14:20, 29 October 2020 (UTC)

Old Georgian states

Hello, I don't know very much about Georgian (caucasus) history. I somewhat think there is a mess with, say, Kingdom of Western Georgia (Q56089236) and Kingdom of Imereti (Q1069959) and for instance https://fr.wikipedia.org/wiki/Cat%C3%A9gorie:Ancien_%C3%89tat_g%C3%A9orgien / https://ka.wikipedia.org/wiki/%E1%83%99%E1%83%90%E1%83%A2%E1%83%94%E1%83%92%E1%83%9D%E1%83%A0%E1%83%98%E1%83%90:%E1%83%A1%E1%83%90%E1%83%A5%E1%83%90%E1%83%A0%E1%83%97%E1%83%95%E1%83%94%E1%83%9A%E1%83%9D%E1%83%A1_%E1%83%A1%E1%83%90%E1%83%9B%E1%83%94%E1%83%A4%E1%83%9D-%E1%83%A1%E1%83%90%E1%83%9B%E1%83%97%E1%83%90%E1%83%95%E1%83%A0%E1%83%9D%E1%83%94%E1%83%91%E1%83%98 . Could anyone of better history knowledge help clean and fix their items ? Bouzinac ^{💬●✒️●💛} 21:04, 29 October 2020 (UTC)

Can I borrow your eyes?

Please look at entry number 6 and see if you can find my mistake that is giving the error message: https://www.wikidata.org/wiki/Talk:Q6797896 When you work on a list like this over month's of time, you get brain fatigue, and need a new set of eyes. I was working on the master list at English Wikipedia which led to my being banned mid edit a few years ago. I want to complete the project here. --RAN (talk) 19:59, 30 October 2020 (UTC)

It's fine; was just an effect of the merge on Azariah Dunham (diff) needing to be reflected in the QId used on the P106 on John Bubenheim Bayard. --Tagishsimon (talk) 20:18, 30 October 2020 (UTC)

Thanks! I should have caught that myself. Does a bot eventually replace the QIDs after a merge where they are used? --RAN (talk) 21:42, 30 October 2020 (UTC)

Yes, that's my understanding. Easy to see why tired eyes would not see that cause ;) --Tagishsimon (talk) 21:46, 30 October 2020 (UTC)

Mahmoud Abdelrazek, Egyptian footballer

Shikabala (Q1850464) was heavily vandalized by multiple people. I reverted the recent changes but there are still some clearly wrong claims: height (P2048) and birth name (P1477) contains a nickname. Ideally someone who speaks Arabic looks up the correct values and checks the rest of the item. Pyfisch (talk) 10:37, 31 October 2020 (UTC)

Archived IDs

In case of Google Play Music artist ID (former scheme) (P4198) if should contain two variants of the ID or it should be linked somewhere due to the way Wayback Machine archives pags. Eurohunter (talk) 10:54, 31 October 2020 (UTC)

Exammple:

• https://web.archive.org/web/*/https://play.google.com/store/music/artist?id=Ahatqs4twxuobnxbyhoghtjb64e
• https://web.archive.org/web/*/https://play.google.com/store/music/artist/Basshunter?id=Ahatqs4twxuobnxbyhoghtjb64e

Eurohunter (talk) 10:54, 31 October 2020 (UTC)

Languages available "Traditional Chinese" and "Chinese"?

At Taiwan Transportation Safety Board (Q4828674) the options for the name include "Chinese" and "Traditional Chinese". Isn't there a way to change "Chinese" to "Simplified Chinese"?

Thanks WhisperToMe (talk) 20:36, 24 October 2020 (UTC)

WhisperToMe, I'm assuming that you are talking about the few labels of the item visible before expanding the full list, and that you want to have simplified and traditional Chinese labels there. Did you try changing your language fallback chain, for example by adding some Babel boxes? whym (talk) 11:31, 29 October 2020 (UTC)

Wow, I see that item is endowed with labels in Chinese, Simplified Chinese, Traditional Chinese, Chinese (China), Chinese (Hong Kong), Chinese (Macau), Chinese (Malaysia), Chinese (Singapore), and Chinese (Taiwan). Though as far as I can tell there are only two distinct primary labels across those nine entries.

Babel says it uses this table which has zh, zh-classical, zh-min, zh-yue, and wuu. But if I type “chinese” into the ULS language selector I get 14 hits, including: 中文,

中文（中国大陆）, 中文（简体）, 中文（繁體）[generic, mainland, simplified, traditional (AFAICT)].

For me in Australia, ULS suggests I might read English, 中文, 中文（繁體）, or italiano.

I can switch to another language via ULS, but that disappears from my fallback chain if I switch back.

To add a label in Simplified Chinese orthography I could enable the Label Lister gadget and hand-type zh-hans as the language code. But to see it displayed I need to tap on either Labels list or All entered languages each time.

If WhisperToMe wants to have Chinese (Traditional) or English as their first preference, and also see labels in Chinese (Simplified) without expanding, I can't see a way to do that?

— Pelagic (talk) 21:02, 30 October 2020 (UTC)

Argh! To answer my own question, I did a test. Despite the documentation, zh-hans-0 and zh-hant-0 in Babel do seem to work, but zh-hant-TW-0 or zh-MO-0 don’t. When I say “it works”: I can get Simplified Chinese to show in the labels-and-descriptions box, but it doesn’t get listed at MyLanguageFallBackChain. The API returns normalised codes [13] but in alphabetical order not preference order. – Pelagic (talk) 21:53, 30 October 2020 (UTC)

I suspect there might be a mismatch between Babel "zh-Hant-TW" versus Wikidata "zh-tw"? Or am I barking up the wrong tree? —Pelagic (talk) 22:31, 30 October 2020 (UTC)

@Pelagic: Per our label policy, zh-classical, zh-min-nan and zh-yue are semi-automatically (i.e. by both bots and humans) changed by lzh, nan and yue so these label codes are expected to be different from the URL code, as for zh-* variants' codes, there seems still unclear if there are consensus about them or not. PS: after a Phabricator task's request (ignored which that is), zh-CN, zh-TW, ... etc are already replaced by zh-Hans-CN, zh-Hant-TW, ... etc in the Special:Preferences. --Liuxinyu970226 (talk) 23:51, 30 October 2020 (UTC)

Thanks, Liuxinyu970226. I'm surprised to hear that labels for both zh-yue and yue (etc.) exist in parallel and require ongoing maintenance. I guess maybe they couldn’t be eliminated because some tools and bots still use the old codes?

I see that the Babel language codes have been cleaned up (phab:T101086 “Standardize invalid language codes for Babel extension”) and match those in User Preferences. Seems like the documentation at mw:Extensions:Babel (which points to the list having zh-classical etc.) might need updating. And if anyone reading this knows Babel's internals, could they add to Lea's list?

Oddly, if I select say zh-Hans-SG in user preferences, then I do get Singaporean Chinese in the termsbox; but if I add zh-Hans-SG only to my babel list, then it's not shown.

— Pelagic (talk) 07:29, 31 October 2020 (UTC)

I bit the bullet and created phab:T266940. Pelagic (talk) 18:15, 31 October 2020 (UTC)

Sacral architecture vs. religious building

There seems to be a mix-up of two concepts in sacred architecture (Q47848), which is supposed to be about the architectural practises. However, it is a subclass of religious building (Q24398318) which must be wrong. I already moved the three site-links which were definitely wrong, but as I am not familiar how to model architecture and it gets late here - can someone else take over and clean those two items up? Ahoerstemeier (talk) 21:40, 29 October 2020 (UTC)

I changed sacred architecture (Q47848) to be a subclass of architecture (Q12271). I also thought maybe subclass of religion (Q9174), but since it's not a religion, that's not right. Ghouston (talk) 21:42, 30 October 2020 (UTC)

religious art (Q2864737) is better as a superclass. Ghouston (talk) 00:40, 31 October 2020 (UTC)

@Ghouston: Could you also please change instance of sacred architecture (Q47848) to instance of: religious building (Q24398318)? There seems to be a few dozen of these: https://w.wiki/jhr Vojtěch Dostál (talk) 15:59, 1 November 2020 (UTC)

If it's only a few dozen, I can do it manually. Most seem to be temples, like Sri Lankaramaya Buddhist Temple (Q17054133), and there's surely a more specific class available. Ghouston (talk) 23:33, 1 November 2020 (UTC)