Latest comment: 3 years ago3 comments2 people in discussion
I'm looking to add functionality to import (English) Wikipedia references via Wwwyzzerdd (Q108267084) but I'm not familiar with any tools to convert Wikipedia references (either as Wikitext or HTML) to Wikidata references (or even vice-versa?). It seems like something someone would've made but I can't seem to find such a thing. The only thing similar is Cite_Q but that imports references from an entire item (and does it in the wrong direction). I would also like the tool to work for multiple language Wikipedias at once but that's probably too much to ask. Anyone have any pointers? Having structured references really feels like something that should exist. Thanks. BrokenSegue (talk) 22:23, 29 April 2023 (UTC)
Latest comment: 3 years ago1 comment1 person in discussion
Here's your quick overview of what has been happening around Wikidata over the last week.
Discussions
New requests for permissions/Bot:
ForgesBot (Task: Add licensing information to software forges entries in accordance to what is found in the corresponding Wikipedia page. It is used as a helper in the context of the Forges project)
Next Linked Data for Libraries LD4 Wikidata Affinity Group call May 2, 2023: Fudie Zhao will be presenting on Wikidata in Digital Humanities projects. Agenda
WeChangEd is an ERC-funded research project and online database by Ghent University, about women editors in Europe. The database shows stories about women editors and their publications, and is powered by Wikidata and Sciencestories.io.
Userbox migration seems to be a policy about deemphazing "all controversial and divisive userboxes". If EnWiki wants to do that with certain userboxes not having those user boxes linked from Wikidata is in line with the intention of the policy. Our notability policy explicitely says that sitelinks into the User or Draft namespace don't produce notability.
That does mean that userboxes which are templates that don't have two non-User/Draft namespace sitelinks lose notability through Userbox migration policy and it makes sense to delete those. ChristianKl ❪✉❫ 15:22, 1 May 2023 (UTC)
@ChristianKl Thank you for your explanation. Thus I can safetly remove from items all redirect-sitelinks to userboxes here which were moved to user namespace. And the items like Q43658193 can be deleted. right? Carenanny55 (talk) 16:08, 1 May 2023 (UTC)
Without speaking the language is hard for me to see whether Q43658193 links into the user namespace. If it does, you are right that the link should be removed and it should be deleted. I would expect that some Wikidata admin would delete template items without sitelinks sooner or later without them needing special nomination to be deleted. ChristianKl ❪✉❫ 18:08, 1 May 2023 (UTC)
I think your edit is reasonable, but I don't have a strong conviction about whether or not that link to Jimmy should be there. ChristianKl ❪✉❫ 16:28, 2 May 2023 (UTC)
Data population for Master's Thesis
Latest comment: 3 years ago4 comments4 people in discussion
Hello,
I am a Master's student at Instituto Superior Técnico in Lisbon and, for my thesis, I want to use the Wikidata Knowledge Graph to create a suggestions functionality for text writing. However, I need~the permission to populate the Wikidata Knowledge Graph with IST-related information in order for the suggestions tab to work, in the context of the project.
@Bernardoquinteiro10 Might I gently suggest that your project would go a lot better if you knew more about Wikidata? I note that your edit here was your first and only edit to any WMF project. Maybe you should spend a little time learning about the project before trying to make big changes. Bovlb (talk) 18:35, 2 May 2023 (UTC)
Subclass or instance_of
Latest comment: 3 years ago9 comments5 people in discussion
I think this question is closely linked to whether literature is a class at all. If we can say “Every Elizabethan literature is a literature”, then it is a subclass, but “Every Elizabethan literature” does not make sense, does it? I would still say that subclass fits better than instance, since it works when using “belongs to”: Everything (work etc.) that belongs to Elizabethan literature also belongs to literature in general. Using “instance of” here would imply something could belong to one but not the other (“subclass of” is transitive, “instance of” isn’t). --2A02:8108:50BF:C694:4881:39D7:411A:F02112:34, 2 May 2023 (UTC)
Yes, but there is a rationale given in the constraint violation message (“This statement leads to false inferences that conflate artworks with the practice or study of art”). I don’t know whether it is relevant here, but it seems the constraint is there to solve an actual problem. --2A02:8108:50BF:C694:4881:39D7:411A:F02112:38, 2 May 2023 (UTC)
It 'solves a problem' when 'literature' is applied to a single work/edition. It stands in the corner with a dunces cap on its head when it borks the correct use when describing a literary genre. It's not uncommon for constraint designers to attack the corner of WD they grok, whilst failing to consider side-effects. --Tagishsimon (talk) 12:44, 2 May 2023 (UTC)
Should I create something called "literary era"? That should solve the problem of instance_of= ? --RAN (talk) 12:40, 2 May 2023 (UTC)
To quote the en-Wikipedia article "List of literary movements": "Literary movements are a way to divide literature into categories of similar philosophical, topical, or aesthetic features, as opposed to divisions by genre or period". But the borders can certainly get blurry; the same article lists "Renaissance literature" as a movement. Pfadintegral (talk) 05:27, 3 May 2023 (UTC)
Double wikidata on en-wiki
Latest comment: 3 years ago5 comments2 people in discussion
I have come across some (or even quite a few) where under the hidden category that it’s short description matches wikidata and then the next line it says it does not match it. If neither match or it’s missing on the top line I usually correct it to be one or the other. Then I get the previous matches statement. If it’s not fixed then it only shows one line of not matching. What would call this or do I need to go line by line and see if something was pulled into the article from wikidata and it got corrupted?
@ChristianKl When you look at the article I linked and look at the hidden categories. One says the short description matches wikidata and then I think two cats after it says the short description does not match. ThatFungi (talk) 12:00, 3 May 2023 (UTC)
@ChristianKl I understand. I just wanted to make sure it’s not an issue on pulling the info across. I’m wondering if an image that was brought over doesn’t match what’s on the corresponding wikidata property. I’ll dive into it some more. Thanks for your assistance. It is much appreciated. ThatFungi (talk) 13:38, 3 May 2023 (UTC)
[Breaking Change Announcement] Changes to the wblistentityusage API module
Latest comment: 3 years ago1 comment1 person in discussion
Hello!
This is a breaking change announcement regarding the WikibaseClient wblistentityusage API module. This API module is used for listing all pages that make use of data from a given entity via Lua, parser functions, or Sitelinks.
Currently, this API module has two issues:
When using this module as a value for the list parameter in an API query action, the results are added to the query.pages key in the response object. However, this key is reserved for use with the following parameters: generator, titles, pageids, and revids. This could lead to a collision if used alongside one of these parameters, potentially causing a runtime exception.
The API parameter prefix wbeu is shared between list=wblistentityusage and prop=wbentityusage (a different module that instead lists all entities used on a given page). However, the module prefix must be globally unique.
We are addressing both issues at the time. Starting today, you can use the wblistentityusage API with the parameter prefix wbleu instead of wbeu, and the results will be added under query.entityusage instead of query.pages. However, the old prefix (wbeu) is still supported for some time, and will continue to produce the same output as before (with an additional deprecation warning).
On or shortly after 2023-05-17, we will remove support for the old prefix and output format from the API. Thus, you now have two weeks to migrate to the new prefix and output format. For a comparison between the old and new methods of using the API, please refer to the wbeu version and wbleu version examples.
If you have any questions or concerns about this change, please don’t hesitate to reach out to us via tickets T196962 or T300460.
Latest comment: 3 years ago44 comments13 people in discussion
What is people's opinion on such edits? It seems to me that if an item has an English name and alias, that can be denoted in the English name and alias. Why should foreign-language labels be polluted with exactly the same text as the English? If there is no substantial translation and the name/alias used by speakers of, say, Traditional Chinese is exactly the same as the English name, the English labels suffice, do they not? What more is gained by spamming all languages like this? It seems to be a chilling effect to discourage foreign-language speakers from populating these fields, if indeed there is a variant in that language. Elizium23 (talk) 06:47, 25 April 2023 (UTC)
@Elizium23: There's actually no reason to revert such edits and claim they're non-constructive or spam: first of all, Traditional Chinese was not (!) even labelled, secondly a person's name does not change from English to any (Latin script) language (let's say, Spanish? French? German?), last but least I added labels and aliases for languages that merely use Latin script (and Traditional Chinese is not among them). --France3c0 (talk) 07:01, 25 April 2023 (UTC)
What is the benefit/advantage to filling these in?
Wiki software is perfectly capable of handling the fact that a language does not have an alias and defaulting to English.
You will need to justify the edits by stating your purpose and benefit these additions have to the project.
I see them as a net negative. They are unnecessary bloat and pointless duplication, unless you can adequately explain otherwise.
They are also a burden on future editors. Say that Eilish changes her name, via marriage, transgenderism, or we decide via consensus. What is an editor to do, make 300 edits to change them all at once? They are the same value! Elizium23 (talk) 07:19, 25 April 2023 (UTC)
Furthermore, you have no proof. You have no sources. Have you researched a body of literature in each of 300 languages and determined that each refers to Eilish by this name? Have you walked among the Cherokee to hear them refer to her? It is common to require sources to back up assertions, especially extraordinary ones, and names/labels have the unfortunate feature that they cannot be cited, but I am afraid that I must demand your proof that you have identified her name as such in each of these languages. Elizium23 (talk) 07:23, 25 April 2023 (UTC)
@France3c0, please do not edit-war while your edits are in an unresolved dispute. They may become difficult to undo. Consensus is required for such extraordinary changes. Elizium23 (talk) 10:57, 25 April 2023 (UTC)
There are currently no rules against it. It's however likely that it will be all deleted once we adopt 'mul'. ChristianKl ❪✉❫ 13:50, 25 April 2023 (UTC)
I don't think "no rule against it" is a compelling reason to be doing it. Typically, adding properties, qualifiers etc. serves a definite purpose that improves the project. Nobody's been able to articulate what improvement is brought by spamming these names. Elizium23 (talk) 17:55, 25 April 2023 (UTC)
@Elizium23: It seems to me you're taking this nonsense too far. Why would adding labels in different languages be spam? Perhaps, you gotta reason that first. France3c0 (talk) 18:06, 25 April 2023 (UTC)
I'm worried that you are continuing to do mass changes while this discussion is taking place, even for items with titles like Never Gonna Not Dance Again (Q115108001) where I doubt the song is known by that English phrase in 200 languages. Is your ideal for every WD item to have 200 language labels in the end? Vicarage (talk) 06:21, 26 April 2023 (UTC)
(EC) The addition of the aliases means that a user who has a particular language set in their WD UI, or in their WDQS report, will see the alias value. That's actually quite a big win. You can argue the toss about whether it's appropriate for a name in a particular style to be an alias for a particular language, but I question whether it is worth losing sleep over; aliases, in particular, allow users to find items, and so should IMO obay Postel's law - which would support their addition.--Tagishsimon (talk) 18:08, 25 April 2023 (UTC)
I am not sure I quite understand. In my WD UI I see all aliases in every language. Are you saying that there are users who are not able to see the values of English labels, like if they only edit in Cherokee? Does WD/WDQS hide values that do not match the user's declared languages? Elizium23 (talk) 18:28, 25 April 2023 (UTC)
Surely I see the thirst for more and better data across the board. Certainly filling out blank fields is a plus and inflates our numbers here well. I am afraid that this user does not speak, understand, or really care about these minority/obscure languages, has not done any actual research into usage among the speakers or writers of such languages, and doesn't actually care. Many of these languages are so small they don't even have Wikis on Wikimedia. It almost seems like Cultural Imperialism to impose aliases, derived from English, into their space. Meanwhile this user is less interested in engaging in discussion or building consensus, and would prefer to edit-war to keep their preferred changes in, despite reasonable objections by people like me and Lucas. Elizium23 (talk) 18:32, 25 April 2023 (UTC)
I agree that there’s no benefit to these additions – aliases exist to improve search, and search already uses aliases in all languages, so copying identical aliases into additional languages serves no purpose whatsoever. Lucas Werkmeister (talk) 18:09, 25 April 2023 (UTC)
"no purpose whatsoever" is clearly incorrect, Lucas. If I ask for Welsh language aliases of an item in WDQS I will not see English language aliases. There is a major-language hegemony on WD, which we would do well to break. --Tagishsimon (talk) 18:37, 25 April 2023 (UTC)
Surely the query system allows you to specify a language hieracy for labels. Ensuring it does allow for Welsh, default English in all cases seems much better than having the Welsh editors copy English text. Vicarage (talk) 06:09, 26 April 2023 (UTC)
@Elizium23 the fact that you see no compelling reasons for doing an edit does not give you a right to undo the edit. You need to have a reason why it would be good for the edit not to exist to undo it.
Wikidata is a place where different people with different goals come together and just because you don't care for the reasons why someone else made an edit does not give you the right to undo it. ChristianKl ❪✉❫ 21:00, 25 April 2023 (UTC)
Speaking of 'mul'; What will happen to transliterations/transcriptions of proper nouns once 'mul' rolls out? Will those get removed too, or kept? Or will just transliterations/transcriptions into different writing systems be kept? Infrastruktur (talk) 18:34, 25 April 2023 (UTC)
I don't think we have made final decisions on this but one proposal is "You are not allowed to add values that are duplicates of the value that's in mul into other languages" (not allowed meaning the software won't let you).
I think we had a few discussion on the project chat and there are software changes made that are life on the test wiki, but I don't think we have a centralized place. ChristianKl ❪✉❫ 12:15, 26 April 2023 (UTC)
Please watch your tone. You can ask what the purpose of copying labels to hundreds of languages is without repeatedly calling the edits spam and speculating that the editor is trying to discourage speakers of other languages from editing. - Nikki (talk) 18:40, 25 April 2023 (UTC)
The editor is literally using a tool that's designed to spam as many languages as possible. There is zero chance that the editor speaks most, if any of these languages in question. The editor has no sources, no research, no demonstrated interest in any of these languages. The editor is wholly unable to articulate why they're doing it in the first place. It's spam by an edit-warrior. I stand by everything I've said. Elizium23 (talk) 20:15, 25 April 2023 (UTC)
@Emu, it is not for me to make that judgement. It is my opinion that nameGuzzler is a tool that has been created to spam for no discernible advantage or benefit to the Wikidata project. Whether it is "prima facie spam" can be decided by consensus. I think it makes little difference whether the tool user speaks the languages or not -- how would we even tell? Elizium23 (talk) 12:54, 26 April 2023 (UTC)
@Elizium23 I would seriously advise you to retract your statement that nameGuzzler “is a tool that has been created to spam for no discernible advantage or benefit to the Wikidata project”. This “opinion” of yours goes against WD:AGF and, quite frankly, strikes me as a little absurd, too. --Emu (talk) 15:04, 26 April 2023 (UTC)
@Emu, @Nikki, currently @France3c0 has been warned about edit-warring and continues to make nameGuzzler runs in contravention of emerging consensus here. I see at least 3 editors opposing these changes and I don't see a lot of rousing support on the other side.
Edits should be correct, not some subjective justification of "useful". If you have specific reasons those edits are incorrect or should otherwise not be done, then please raise that. But there's no value to asking how an edit is "useful".
You did raise an interesting issue above where you brought up the burden on future editors if Eilish changes her name. This would require editing every alias. That would indeed be an annoyance. ElDubs (talk) 21:03, 25 April 2023 (UTC)
Edits do need to be useful in a system constrained by computing limits and where information can change. Stating things only once, such that the query process does the duplication, is a good idea. Names, which generally only differ by script, are best done dobe by mul, if that is likely to be Latin script.
Has there been any statement from WMF that we'll need to be careful of the amount of content on wikidata for fear of limited computing resource? I very much doubt it. I do not believe this is a concern. ElDubs (talk) 09:03, 26 April 2023 (UTC)
Recently, there was a post about steps that the developers take to reduce computing usage. One of them is to add the 'mul' type with the hope of preventing data duplication like that. SPARQL is very useful but one of the effects of having all data queriable via SPARQL is that you can't just scale it to whatever size you want the way you could a standard NoSQL database. ChristianKl ❪✉❫ 12:28, 26 April 2023 (UTC)
Thanks for this, hopefully it's an issue that can be resolved as regardless of edits that are considered "useless", edits everyone agrees are useful and legitimate will push us to that ceiling. ElDubs (talk) 21:26, 26 April 2023 (UTC)
I am not sure how a useless edit could be "correct". It would seem that every edit must be a net improvement to the project in some way. It adds a useful fact, it clarifies something, it expands knowledge of the item, it reverses vandalism. "Useful" "beneficial" edits are recognizable, objective, and quantifiable.
Now if the person who is making an edit can't even say why they are doing it, and can't describe how it improves the Wikidata project, I think we can agree that that is an undesirable edit, or at least needs additional scrutiny and analysis before mass changes proceed, yes? Elizium23 (talk) 13:04, 26 April 2023 (UTC)
Just to be sure: Is there any compelling argument that the edit in question is against any guideline or consensus? --Emu (talk) 15:06, 26 April 2023 (UTC)
Of course there is. Several editors in this thread have objected. Therefore there is no consensus in favor of the edits, a possible consensus against the edits, and there is no basis for this editor to continue edit-warring. Further, consensus should be based in policy, and no user who is supporting or neutral in this case has provided any facts about how the edits are a net benefit or improvement to our project. Elizium23 (talk) 15:14, 26 April 2023 (UTC)
Latest comment: 3 years ago3 comments2 people in discussion
The labels in many languages debated has shown that the implementation of a mul(tiple) language code is complicated, needing discussion on its purpose, replacement of en, possible mass deletions of entries, education on its performance benefits, SPARQL queries etc. A quick go on test.wikidata.org suggests the implementation there still has problems. I think a new WikiProject be a good focal point? Vicarage (talk) 07:20, 27 April 2023 (UTC)
Because we have quite lots of pages that have been created, but have not been connected to Wikidata. Therefore, it will be easier to use Bot. Regards. Medelam (talk) 22:33, 2 May 2023 (UTC)
Create a new file interwiki.py and duplicate the whole script in that one file:
:#!/usr/bin/python3::importpywikibot::site=pywikibot.Site("jv","wikipedia"):repo=site.data_repository():page=pywikibot.Page(site,u"TITLE ON JVWIKI"):item=pywikibot.ItemPage(repo,u"WIKIDATAID")::item.setSitelink(page,summary=u'Added link to jvwiki: TITLE ON JVWIKI')::# Copy and paste again from #!/usr/bin/python3 to item.setSitelink...:
Run a coreutils software for splitting the file: split --lines="11" --additional-suffix=".py" "interwiki.py"
Move the generated .py files to pywikibot/scripts/userscripts
I think I have to use the Property:P2013 but when I add it I get an error on the regex, because there is no /pages/, /profiles/ or others. Is it possible that Facebook has changed the format of its links and that the regex is no longer current? Or is it the wrong property?
I'm looking into it, but i dont think so. This example does not work because this item is already imported. But I have an excel containing about 800 organizations on women's rights, in it I have: "Ville de Namur. Cellule Egalité des Chances" which contains the Facebook link you saw.
The error on OpenRefine is: Facebook ID (P2013) statements with invalid format.
this sounds like a bug in OpenRefine. I'm not familiar with that software. The regex works on Wikidata. I'm guessing the issue is that there are multiple regex's for the Facebook property. BrokenSegue (talk) 04:37, 4 May 2023 (UTC)
reverting past merger
Latest comment: 3 years ago2 comments2 people in discussion
I am struggling to resolve Wikimedia disambiguation page (Q4167410) (DAB pages) of Japanese wikipedia connecting to non-DAB wikidata items. Some of such connection derived from past merger between DAB and non-DAB items on wikidata. For most cases, reverting the merger works fine (seemingly). However, Willy (Q416744) was problematic. Willy (Q416744) was originally DAB item, but Willy (Q1459646) representing male given name had been merged in early 2016. I have just reverted the merger, and arranged DAB and non-DAB pages of wikipedia projects connecting to DAB and non-DAB items on wikidata, respectively. The problem is, that Willy (Q416744) is now DAB item, but many (>2000) items refer to Willy (Q416744) since it has long been treated as a non-DAB item. My question is which is better, to request bot edit to fix reference Willy (Q416744) > Willy (Q1459646), or to revert my reverting, create a new DAB item for DAB pages, and keep Willy (Q416744) as non-DAB item for non-DAB pages? Mzaki (talk) 03:27, 4 May 2023 (UTC)
Reverting old incorrect merges is often problematic, for example in cases where there are many links from other items (correcting hundreds of incorrect descriptions is also difficult). Sometimes you might be better off with the new DAB item (and fix "few" sitelinks rather than hundreds of links from other items). Jklamo (talk) 09:06, 4 May 2023 (UTC)
Need-blind colleges
Latest comment: 3 years ago8 comments4 people in discussion
Does anyone have an idea of how we'd represent that a college or university practices need-blind admission (Q936700) (e.g. does not take into account applicants' financial circumstances when deciding admission)? {{u|Sdkb}}talk16:33, 3 May 2023 (UTC)
It might also be worth looking whether there's some organization that has standardized categories for the types of types of admission. ChristianKl ❪✉❫ 00:38, 4 May 2023 (UTC)
Possiby qualify the 'has quality' with 'has quality' (or 'applies to part') with a value coined to indicate coverage is not universal - 'subset of admissions' or somesuch. --Tagishsimon (talk) 00:00, 5 May 2023 (UTC)
Yeah, many schools are need-blind only for domestic applicants, so I considered something like "applies to part" -> "domestic applicant", but there's no item for "domestic applicant" and creating one (as well as ones for "international applicant," "transfer applicant," "out-of-state applicant," etc.) seemed a bit much. {{u|Sdkb}}talk02:20, 5 May 2023 (UTC)
Well, these seem like notable subdivisions of the class of applicants, which WD will need if it wants to represent this quality at a more granular level. I'd not hesitate to create them. --Tagishsimon (talk) 02:27, 5 May 2023 (UTC)
Claims about UTC offsets added to hundreds of items
Latest comment: 3 years ago6 comments5 people in discussion
[3] - are such edits desired? If something is located in a specific city or sometimes country, the offset can be inferred them item. Shall it be added to each street, building, entrance, memorial, tree, tower, grave etc. individually? GeoGQL (talk) 02:54, 27 April 2023 (UTC)
They seem pointless, but the user has been inactive since May last year. In an ideal world
@GeoGQL: If you think about whether or not the property should be added, it makes sense to focus on the constraints. If you think the constraints should disallow such edits, propose how to change the constraints. ChristianKl ❪✉❫ 15:05, 28 April 2023 (UTC)
If you look at the linked property proposal, it seems this property's intended granularity was at the country/state level. The proposal also serves as a consensus. Infrastruktur (talk) 21:39, 28 April 2023 (UTC)
@Vicarage, ChristianKl, Infrastruktur, Multichill: the problem can be reduced by linking to IANA time zone items via a dedicated property and make sure that any place item that has a UTC claim after 1970 has a this new property. Of course it still needs to be solved on which level to put something, but with IANA data better available it should be easier. And when the EU abandons daylight saving time, the change can be added to Wikidata by editing only ~30 IANA zones instead of thousands or millions of individual items.
Latest comment: 3 years ago2 comments2 people in discussion
I believe there is some confusion at Hammering Man (Q685987). It seems to be a mix of at least two different sculptures by Jonathan Borofsky. I'm pretty sure the image there doesn't match the Commons category (the location certainly doesn't because the Commons category is about multiple copies of a sculpture, not one single location) and I don't think it matches the various Wikipedia articles, either, at least not all of them. - Jmabel (talk) 23:52, 4 May 2023 (UTC)
Latest comment: 3 years ago10 comments2 people in discussion
I am actively editing articles about architecture, design, and the arts (and adjacent subjects) on Wikipedia, and have become increasingly interested in (and active on) Wikidata. I would appreciate guidance concerning the definition of what constitutes a valid "Value" under the statement "Employer". A specific example is the industrial designer Richard Sapper (Q64699): currently, the only P108 value listed is the Stuttgart State Academy of Art and Design. My question is whether companies with whom the designer maintained a long-term professional relationship such as IBM, Alessi, and Brionvega should be included here too (i.e., is a client/professional relationship considered "Employment" in this case, regardless of the nature of the contracts between the parties)? Cheers, Cl3phact0 (talk) 07:20, 27 April 2023 (UTC)
Thank you, M2k~dewiki, you suggestions above and the related discussion are very helpful. Both have given me some interesting possibilities to explore. -- Cl3phact0 (talk) 16:01, 27 April 2023 (UTC)
@Cl3phact0: every property also has a talk page, where aspects / usage of that property could be discussed and other releated properties in that area might be found (e.g. by a navigation bar).
Again, thank you! Very helpful. I will (attempt to) put this information to good use too. (Wikidata is fascinating. I find it oddly meditative to peruse. A fantastic tool I've been using is: Reasonator, which, no doubt, anyone reading this thread will know – but it was truly a Eureka! moment for me when I discovered this gem.) -- Cl3phact0 (talk) 16:24, 27 April 2023 (UTC)
Latest comment: 3 years ago4 comments3 people in discussion
Trying to find out the occupation of Charles Barton (Q76262192), I noticed that the item was apparently created based on thisThe Peerage (Q21401824) page. Said page cites “Wikipedia, online http;//www.wikipedia.org. Hereinafter cited as Wikipedia” (sic; i.e. no specific Wikipedia article) and some “e-mail message to Darryl Lundy” as its sources. I have no idea how reliable The Peerage is in gerenal and would usually not care about its sources, but isn’t it unfortunate that the entry is (heavily, considering its size) based on information from an unspecified somewhere in Wikipedia? For example, someone may feel inclined to write a Wikipedia article based on the Wikidata item, potentially creating a circular referencing situation. Or am I overly concerned? --2A02:8108:50BF:C694:4881:39D7:411A:F02110:17, 2 May 2023 (UTC)
Latest comment: 3 years ago57 comments9 people in discussion
Can we please change brightons identity from town to city please? Sources show that although the official name is Brighton and Hove, all sources also consider Brighton in the same equal manner when describing the city. Greenfrog23 (talk) 08:12, 5 May 2023 (UTC)
You can change it, you don't need other people for that. If you want to change it, find the references to the sources that say it's a city and change it. ChristianKl ❪✉❫ 11:53, 5 May 2023 (UTC)
Brighton never refers itself as town its always city. Brighton (City of Brighton and Hove) are my suggestion because town status does not exist any more. 78.86.35.23213:34, 5 May 2023 (UTC)
That would be good, but if not then the description saying town in England removed if that's ok please and changed to something like I said previously or even Brighton officially known as the City of Brighton and Hove or even Brighton part of the City of Brighton and Hove would be better. Greenfrog23 (talk) 13:55, 5 May 2023 (UTC)
No. Please get with the programme, Greenfrog23. Brighton is not 'Brighton and Hove'. The Brighton item should not be renamed or redescribed to suggest that it is. Please understand that these two things are distinct, different, not the same. Just no.
As simply as I can put it, the town of Brighton is combined with the town of Hove to make an area which has city status as City of Brighton and Hove. It's like 1 + 1 = 2; and 2 not equalling 1. --Tagishsimon (talk) 14:15, 5 May 2023 (UTC)
What I’m saying is Brighton isn’t a town anymore that privilege had gone once City status was given to Brighton and Hove. keep Brighton on here but change the description to either seaside resort or part of the City of Brighton and Hove, this needs addressing as it’s incorrect. This is the letter approving City status. “To all to whom these Presents shall come Greeting. Whereas We for divers good causes and considerations Us thereunto moving are graciously pleased to confer on the Towns of Brighton and Hove the status of a city Now Therefore Know Ye that We of Our especial grace and favour and mere motion do by these Presents ordain declare and direct that the TOWNS OF BRIGHTON AND HOVE shall (henceforth)have the status of a CITY (and shall have all such rank liberties privileges and immunities as are incident to a City”). Greenfrog23 (talk) 14:30, 5 May 2023 (UTC)
Yes. The TOWNS OF BRIGHTON AND HOVE shall (henceforth)have the status of a CITY. The TOWN OF BRIGHTON ON ITS OWN IS NOT A CITY. WD has an entry for the City. And one entry each for its constituent towns. The towns on their own are not the same as the city. 1+1, 1 != 2, Greenfrog23. It's not hard. --Tagishsimon (talk) 14:50, 5 May 2023 (UTC)
You don’t understand what I’m saying,I know that the City is officially Brighton and Hove. What I am saying is that Brighton is no longer a town, it’s part of a City it’s not a town in its own City that doesn’t happen or make sense there’s no town centre,there’s a City centre! Brighton is part of the City yes but to describe Brighton as a town is wrong.The description stating Brighton is a town is wrong. Greenfrog23 (talk) 15:05, 5 May 2023 (UTC)
The TOWNS OF BRIGHTON AND HOVE shall (henceforth)have the status of a CITY. So if this said the same for Wolverhampton when it received city status (the town of Wolverhampton shall henceforth have the status of a city),that means it is no long a town doesn’t it. Brighton and Hove aren’t separate towns anymore they are 2 parts of a City. So Brighton ( City of Brighton and Hove). Greenfrog23 (talk) 15:13, 5 May 2023 (UTC)
I think this discussion on the English Wikipedia was initiated by the same user, which didn't reach consensus. Also the various IP's have been blocked there due to abusing multiple accounts. Sjoerd de Bruin(talk)14:29, 5 May 2023 (UTC)
You really cannot just make shit up and expect people to buy into it. Why does Brighton cease being a town merely because it is within an larger administrative territorial entity which has been given city status? From which dark passage have you extracted the "it's probably something like a borough/district". Does being a borough/district disqualify a place from being a town? If so, that'll be a surprise for very many places within London. This whole thing is a farrago of nonsense based on your inability to see that there are a multiplicity of ways in which Brighton may be described, some of which you may disagree with. --Tagishsimon (talk) 17:10, 5 May 2023 (UTC)
None of those are a reliable source for the contention that Brighton is not a town. We know about B&H and its city status. We know that Brighton is referred to as a city. We know that in fact it is B&H which is the city. From w:en:Town: "The distinction between a town and a city similarly depends on the approach: a city may strictly be an administrative entity which has been granted that designation by law, but in informal usage, the term is also used to denote an urban locality of a particular size or importance." You seem to have a problem with a city having towns and, presumably, villages within it. Too bad. --Tagishsimon (talk) 17:41, 5 May 2023 (UTC)
Ok so all the sources I provided from both Brighton and Sussex universities, a national newspaper and the official visit Brighton guide are all incorrect aswel are they ? Greenfrog23 (talk) 22:11, 5 May 2023 (UTC)
www.woodingdeaninbusiness.co.uk/blog/is-brighton-a-city///myhotels.com/ www.visitbritain.com/gb/en/business/england/brighton Greenfrog23 (talk) 22:15, 5 May 2023 (UTC)
I can carry on sending you site after site, I’m not here to argue,I just want to put it right is everyone one of these sites wrong or is it this one ? Brighton never refers itself as town it's always city Greenfrog23 (talk) 22:55, 5 May 2023 (UTC)
Latest comment: 3 years ago4 comments3 people in discussion
Hi, I'm having trouble figuring out which property to use for an organization that's changed its name a few times over the years. I tried "alternative name", but that threw up a warning that it wasn't being used properly. Basically, Q4466010, the Pacific Asia Travel Association, was formed in 1952 as Pacific Interim Travel Association, became Pacific Area Travel Association in 1953 and then Pacific Asia Travel Association in 1986. I added the older names as aliases with a date range afterwards, but that feels wrong. If each name should have a separate entry (there is already Q85724306, Pacific Area Travel Association), what's the best way to link them? I used "replaced by" on the Pacific Area Travel Association to link to Pacific Asia Travel Association (and added "replaces" to the latter entry), but the qualifier "start time" is raising an error. —Tcr25 (talk) 15:21, 5 May 2023 (UTC)
With company items, it's always important to look a bit deeper into the circumstances of such changes. Sometimes they indeed simply just change names, but often enough such changes are due to mergers, holding companies taking the names of acquired subsidiaries, and other circumstances where the company with the old name or legal form isn't necessarily the exact same entity as the newer one. Which unfortunately is rarely adequately reflected in external databases or superficial articles (including Wikipedia), leading to Wikidata items mixing more than one legal entity --2A02:810B:580:11D4:9563:313B:DAB5:4D4317:59, 6 May 2023 (UTC)
Crown possessions, and what is not that...
Latest comment: 3 years ago1 comment1 person in discussion
Because of the Coronation happening nowadays, I got curious and looked up two palaces in my native Wikipedia, and lo and behold: Buckingham Palace is OWNED by the House of Windsor!
Frankly, while I would understand "House of Windsor" as a shorthand for the private properties, what's the point of NOT showing Buckingham Palace (and whichever other properties that might be mislabeled) as being the property of the Crown ?
(Brief sidenote: The estates of Balmoral and Sandringham are known to be the private properties of the Monarch; the question regards the properties of the Crown.)
The Peerage property is so bad because one or more people have been doing automatically matched batches based mostly on name matches. And then someone else uses the bad matches to add mother/father/parent relationships. In my opinion any item with a The Peerage property is probably messed up in some regard. And because Wikidata users aren't looking at items about royals/nobles besides the most famous, these items get almost no scrutiny or checking. I certainly am not inclined to go do verification on items I have no interest in. --William Graham (talk) 16:32, 6 May 2023 (UTC)
Latest comment: 3 years ago5 comments3 people in discussion
Hello,
I'm currently working on automatically updating the population of cities in France, but it seems that another user @Markussep is using another method to merge same values.
I'm not sure if I'm clear enough, so here's an example:
For the town of Peyzieux-sur-Saône ( https://www.wikidata.org/wiki/Q693575#P1082 ), the population for 2020 is 676, but the population for 2016 is also 676, so Markussep merges the two and adds "point in time" and the references for 2020 into the existing 2016 claim.
I didn't write my robot this way, and to be honest, I think merging historical data like this can be easily confusing at first.
So what is the best method, all in the same claim or separate claims?
Hi, I started updating the populations using QuickStatements, which worked fine for the larger subdivisions (departments, regions, arrondissements), but indeed for communes the problem you observed may occur. Also, QuickStatements can’t change the status to preferred or normal. For English Wikipedia that’s fine, I’m not sure whether it causes problems in other wikis. Anyway, the bot you used last year is superior, so if you have time please go ahead with the rest. I won’t have much time to edit the coming weeks. Markussep (talk) 01:00, 8 May 2023 (UTC)
Thank you @Markussep, I was wondering if it was a new way to add claims on Wikidata and I though you did it voluntary.
I will let the bot update the remaining entries and I will make a second pass to fix by removing the duplicate qualifiers/references. Myst (talk) 03:07, 8 May 2023 (UTC)
You're directing us to an item with 7 labels, 3 descriptions, 2 aliases, 2 statements, 4 sitelinks, none of which look hugely problematical. --Tagishsimon (talk) 22:31, 8 May 2023 (UTC)
I've tidied things up a little bit, but nothing was seriously wrong. There is a history of vandalism, but nothing survived to this day. Is there something specific you have a problem with? — Huntster (t@c)23:32, 8 May 2023 (UTC)
Latest comment: 3 years ago5 comments3 people in discussion
From computers to microprocessors, many electronic device models support the Bluetooth (Q39531) wireless protocol. However, I'm unsure how this should be represented in Wikidata. I've provided examples of the properties that are being used for this purpose:
Thanks Jklamo and Christian for the advice: I'll take a look at the Wikiproject. For now, uses (P2283) seems like the most sensible option, so I'll use that. It should be easy to automatically migrate the statements if a more specific property (e.g. "supports protocol") is created. BEANS X2 (talk) 15:00, 10 May 2023 (UTC)
You could provide reasons for why you think the edits should be made. If you want to convince someone to do something, providing arguments is generally a good idea. ChristianKl ❪✉❫ 15:37, 9 May 2023 (UTC)
I thought the reasons are obvious. It seems for "3" this was true enough since it has been solved without an explicit reason-statement. For 1, 2, 4 I added "Reason: " and some text. When do you expect an effect to be visible? GeoGQL (talk) 03:06, 10 May 2023 (UTC)
This seems messy and also not a good candidate for my first merge, but I don't think the right approach is to have English only linkable to one of the two, and for the redirection to preclude the second one from being linked. Gidklio (talk) 17:35, 9 May 2023 (UTC)
@Gidklio, can you explain why you want to merge these two things? They are completely separate entities: one is the bird nest itself, the other is a product which uses the bird nest as an ingredient. — Huntster (t@c)17:52, 9 May 2023 (UTC)
Latest comment: 3 years ago24 comments8 people in discussion
I had some free time today so I decided to look into how big a problem anonymous users are on Wikidata. Below is a chart of the revert rate of edits made by users and by IPs. The chart only considers edits made using the mobile or Web UI (i.e. not QS). Deleted edits are also not included. So the anonymous revert rate probably undercounts the magnitude of the negative effect.
Can you do a plot of the distribution of revert rate and revert count by distinct IP address, to see if its certain bad apples are causing the problem? Vicarage (talk) 06:55, 7 May 2023 (UTC)
@Vicarage: Here's the plot of that. . There are some problem users but 1 in 4 reverted edits are caused by users who have just one reverted edit (and 10% are caused by users with 2 reverted edits). Not sure what I can conclude from this. BrokenSegue (talk) 18:50, 7 May 2023 (UTC)
That looks that most bad edits are either innocent, or by sophisticated bad hats that use IP address switching. I can see that happening for general mediawikis, but not here. And we've only got one example of systemic abuse. So not much point monitoring users after their first few mistakes. Vicarage (talk) 18:58, 7 May 2023 (UTC)
Guess I'll be devil's advocate then and point out that 92% of all IP-edits are still presumably good. Estopedist1 points out a lack of manpower to patrol those edits. There was 27000 patrols last month, 18000 of those were by Msynbot and Deltabot, meaning 9000 manual patrols. The same period there were 139000 edits by IP-users, and 20,47 million edits overall. But that could just as easily mean more people are needed to patrol and block vandals, it doesn't necessarily mean the best course of action is blocking all IP-edits. That said, since most people on Wikidata can patrol if they want to, I attribute the low patrolling to the willingness to put in this kind of work. I think this is safe to infer. Infrastruktur (talk) 08:28, 7 May 2023 (UTC)
+1. There are some issues with the patrolling process: many are not aware of how this works and why this is necessary; tools are scarce and not well known; patrolling activity does not increase your reputation since it is pretty much invisible. Nothing of this implies that we should exclude unregistered editing, though. —MisterSynergy (talk) 10:23, 7 May 2023 (UTC)
true but the chart also excludes automated edits. Most edits aren't made manually so the true fraction of reverted edits is probably much lower for logged in users. BrokenSegue (talk) 18:55, 7 May 2023 (UTC)
When talking about how well we manage to patrol current anonymous edits, it would be worth to look at the numbers for the amount of unpartrolled edits that are out there. If you think that there's a problem with those, please bring the numbers. ChristianKl ❪✉❫ 11:46, 7 May 2023 (UTC)
I already had some scripts for monthly statistics so I was able to tweak them a bit to get the numbers you wanted. For the last 30 days there was 27742 patrol actions of which 11443 and 6650 was Msynbot and Deltabot patrols leaving 9649 manual patrols. So 34.78% of the patrols are by a human.
The second script looked at unpatrolled changes, and after todays tweak, counts various totals, but this only counts edits and page (item) creations. Here we have: Unpatrolled set: 188 728, manually patrolled set: 24 461 and autopatrolled set: 20 881 955. This gives us an apparent patrolling coverage of 11.47%. But since we know that only 34.78% the "manual" patrols is actually patrolled by a human, that leaves us with an actual patrolling coverage of 4.31%. Infrastruktur (talk) 14:58, 7 May 2023 (UTC)
@Infrastruktur To me 139000 edits by anonymous users per month and a total set of unpatrolled edits of 188728 sounds together that the total amount of unpatrolled edits is 1 1/2 months of edits. That's not what you would expect to see if we only manage to patrol a small amount of out edits. In that case we would expect to see a growth in the unpatrolled actions over time to a much larger backlog. ChristianKl ❪✉❫ 18:38, 8 May 2023 (UTC)
Most of my numbers were for the last 30 days, except in my first post where I used the edit counts for April from the official statistics website. I'm not sure, but to me it seems like the patrolled status is only remembered for 30 days and forgotten after that. Otherwise they would indeed have accumulated very fast. The patrolling coverage numbers indicate that four percent of the edits needing checking gets checked quickly and the eleven percent is what eventually gets fixed later that month. 89% is either fine or get fixed later if we're lucky... but have escaped the patrolling. Infrastruktur (talk) 20:06, 8 May 2023 (UTC)
I made an incorrect assumption about the patrolling coverage, my apologies for that. The bots mark edits that have been fixed, so there's no reason to mark edits that have been fixed as being patrolled. The real patrolling coverage is at least eleven percent. Infrastruktur (talk) 16:47, 10 May 2023 (UTC)
The raw values for patrolling of IP-users: Unpatrolled: 125 107, manual patrol (including bots): 16 161, autopatrol: 4 586. Infrastruktur (talk) 15:43, 7 May 2023 (UTC)
I can sort of understand people thinking the modest amount of IP-contributions isn't worth the effort it takes to patrol it, but that raises two questions: 1) Will the resources freed up really be used to raise quality elsewhere, or will people just do other things instead and 2) this could have an adverse effect on recruitment of new users, as there will no longer be as low a barrier to entry. Infrastruktur (talk) 13:58, 8 May 2023 (UTC)
Interesting, my hypothesis would be that disabling IP-edits will likely have a positive effect on the recruitment of new users. One can only really talk to users once they have created an account. And interacting with them in positive conversation should make it more likely for them to become permanent new members. And creating an account seems like an extremely low barrier to me. Most new users will expect it. Some new people even got irritated, when I told them that you don't need to create an account to edit on Wikipedia or Wikidata. This is quite unusual on the Internet, after all. --Manuel (WMDE) (talk) 16:55, 8 May 2023 (UTC)
The overall amount of individuals who contribute to Wikidata would likely decrease, since a requirement to use an account is nothing but an extra barrier compared to the status quo. What is interesting in this context is the distribution of IP work (~18.800 different IPs have edited Wikidata in the past 30 days, with ~142.000 edits):
There are some heavy users without an account (e.g. currently 175 different IPs have made 100+ edits in the past 30 days; these 175 IPs have contributed 61.300/43.3% of all IP edits in the past 30 days). Some of them would register an account and continue since they have a clear mission, but this is a one-time effect: many of these are long-term users. It would of course be better if they were using an account, but I am not aware why they do not do so. The edit quality of these users is usually not worse than for registered users.
However, most IPs contribute less than 5 edits per 30 days (currently more than 15.000 different IP addresses). These are sprodic or one-time editors. It is unlikely that many of these will register an account just to make a couple of edits.
So I think we would lose a significant amount of individuals who edit Wikidata (up to ~25% of the current user base), but not so much in terms of absolute edit numbers. —MisterSynergy (talk) 18:12, 8 May 2023 (UTC)
@Manuel Merz (WMDE) One of the big differences between Wikimedia and many other spaces is that you need to set a password within Wikimedia while many other websites allow for one-click sign-on with Google or Apple.
Software-wise one solution would be to require having an account for edits made with the mobile app but make it really easy to create an account with the mobile app so that there's little barrier. ChristianKl ❪✉❫ 18:45, 8 May 2023 (UTC)
I accidentally create multiple duplicate
Latest comment: 3 years ago12 comments4 people in discussion
Hello,
I made a mistake. I made an import with OpenRefine, but it got stuck for a few hours on a certain percentage (About 12h). So I shut down and restarted and this 2-3 times. I didn't manage to go to the end of my import, but I just realized that it created duplicates. Is there any way to delete a contribution that i made ?
@Ideesculture: I've merged all duplicates which were P31=organisation with a reference pointing to catalogue.amazone.be/index.php/Detail/entities/ ... about 670 merges. Not sure if there were other patterns of item added. Creating dupes sometimes happens; many of us have been there. No worries. --Tagishsimon (talk) 12:01, 9 May 2023 (UTC)
Oh great! Thanks, I was supposed to have 803 items modified/created. I can see that I don't, but thanks to this I can identify which ones and re-import only on the missing ones to create them.
It seems poorly associated. Some of the links from agricultural revolution (Q6499967) lead to generalized articles about the concept, but the one in Dutch mostly describes the British one, and indeed links to the generalized article as well. So I think that first what needs to be done is clear up the inter-language links so that everything in 6499967 links to the general article and not the British article, (also some of the descriptions need modification too). There is certainly space for both items, because they are distinct concepts; the British one is a specific instance of the general one. Elizium23 (talk) 06:12, 14 May 2023 (UTC)
We have frequently discussion about sourcing on various part of Wikidata like the project chat admin noticeboard or other places. Those show the defacto calls our community makes about sourcing requirements. If you want to know what it is, read the archives.
Moving forward, there's the possibility of make proposals for a written policy that reflects the currents usage. Probably, at best we could have an RfC were we decide on an explicit policy. ChristianKl ❪✉❫ 22:16, 14 May 2023 (UTC)
This is your personal opinion, and I disagree. While Help:Sources clearly has potential for improvement, it does in many regards reflect the reality quite well. It is not appropriate to swiftly tag it as "outdated" with barely any explanation. This banner confuses editors, but it does not move anything forward. —MisterSynergy (talk) 22:31, 14 May 2023 (UTC)
The page says: 'This guideline explains when it is necessary to add a source to a statement in Wikidata. [...] Wikidata is a collection of sourced data, which means that most statements should indicate where the data comes from.'
This implies that we forbid adding statements without sources. We frequently had in the past people reading that page coming away with the mistaken assumption that users are not allowed to add statements without sources, delete those claims followed by a discussion on this page where we come to the conclusion that the statements shouldn't be deleted.
We do have a bot policy that actually requires sources and help:sources says nothing about it. We have a living people policy that makes sourcing related claims and the page says nothing about that either. We have a notability policy that interfacing with a need for sources.
If someone reads help:sources they learn nothing about those policies we have that create needs for source and come at the same time with mistaken assuptions about sources being necessary in cases where they aren't. ChristianKl ❪✉❫ 10:37, 15 May 2023 (UTC)
There was already an RfC. If you want to start another one, feel free, but in the interim "read the archives" isn't a good solution, especially for linking to newbies as per Thibaut. Nikkimaria (talk) 01:28, 15 May 2023 (UTC)
There was no RfC that lead to an adoption to a policy for sources. Help:Sources is not a page that went through an RfC that found consensus for it's adoption. ChristianKl ❪✉❫ 10:28, 15 May 2023 (UTC)
EntitySchemas: We've continued work on the new datatype to link to EntitySchemas in statements. (phab:T214884)
Converted a number of Properties from string to external ID datatype (phab:T334450)
mul language code: We've worked on preparing the special pages related to labels, descriptions and aliases for the new language code. (phab:T330193, phab:T329626)
Wikibase REST API: We continued working on the ability to modify labels of an Item using the HTTP PATCH method. (phab:T332739).
Is there presedence for such kinds of statements? Other P31 statements on Wikidata properties have as object items that have a name starting with "Wikidata". Only few items seem to use "numeric ID" (e.g. 2023-01-09 OpenStreetMap numeric user ID P279 ... [7]).
Because some existing properties link to Q93868746 via P31, for example, X (Twitter) numeric user ID (P6552) and Genius artist numeric ID (P6351) (Q93868746 links to these two examples at that time), and I think it is a good idea to link every other numeric ID properties to Q93868746 as well. This bot is used as a generic QuickStatements replacement because it's quicker to hit run than login on QuickStatements. Midleading (talk) 14:43, 18 April 2023 (UTC)
@Eurohunter: seems to have added those existing one's back in 2020.
I think that removing all of them from properties would make more sense given that they follow a different form than the usual P31 we have for properties. ChristianKl ❪✉❫ 15:03, 18 April 2023 (UTC)
I still think linking numeric IDs to Q93868746 is useful, and P31 is the appropriate property. You can create a new item specific to Wikidata with label beginning with Wikidata for this purpose and make it subclass of Q93868746. Midleading (talk) 15:48, 18 April 2023 (UTC)
"has quality": the object should be a quality, right? Currently Q93868746 is a subclass of ID rather than a characteristic of ID. Midleading (talk) 09:10, 20 April 2023 (UTC)
I'm going to reiterate my suggestion from edit group discussion: if a statement about numeric value is really necessary then a clear solution would be to create "Wikidata property for numeric identifier" or "Wikidata property with numeric value" class or something like that, and use this as P31 value instead. numeric identifier (Q93868746) as P31 in property page makes little sense as property that is used for some (numeric) value is different from its values (a true instance of Q93868746 would be some value like "185407". Neither does the suggestion to use P1552 (has quality) make much more sense as numeric ID is not really a quality of any sort. 2001:7D0:81FD:BC80:BC5E:FBC6:E24B:756707:53, 5 May 2023 (UTC)
Agreed. But then how is "numeric identifier" or "numeric value" defined? By a regex? But the regex is already stated. Maybe improve the regex statements first, e.g. standardize on a form that means the same in any dialect, 0-9 instead of \d ot make sure it is universially understood as ASCII numbers etc. GeoGQL (talk) 21:22, 5 May 2023 (UTC)
These phrases, respectively referring to an identifer or value consisting of numeric characters, seem pretty much self-explanatory to me. As what else could these be defined? I don't think there's any need to mention "regex" in defintion as that is merely one technical application that can make use of numeric characters. 2001:7D0:81FD:BC80:C05B:F975:D55C:DF016:44, 6 May 2023 (UTC)
See below where 1) a user wants to enforce /[0-9]+/ on an item named "numeric ID" and 2) a quote from a Wikipedia article attached to "nominal number" is given which implies no restriction to numeric characters. Re "merely one technical application that can make use of numeric characters" - not sure that is true, because "ex" stands for expression, but more importantly a regex can be used to define. GeoGQL (talk) 14:46, 9 May 2023 (UTC)
nominal number (Q2004972)
There is already 'nominal number (Q2004972)' - those identifiers that use [1-9][0-9]* are probably of type 'nominal number'. The description of numeric identifier (Q93868746) says 'identifier in the format of a number, regex [0-9]+' which maybe isn't restrictive enough, or can numbers start with leading zeros and at the same time is too restrictive or do numbers have to use only [0-9]? GeoGQL (talk) 15:38, 19 April 2023 (UTC)
Q93868746 is [0-9]+, which means they can have leading zeros, like Lobatse (Q165768)postal code (P281)"0000". Q2004972 is superclass of Q93868746 and also doesn't specify whether leading zeros are allowed. Maybe you want to create a new item for [1-9][0-9]*. Why are you removing descriptions for Q93868746 in all languages? Midleading (talk) 09:17, 20 April 2023 (UTC)
Before you removed descriptions from all languages and changed P279 of Q93868746, it is [0-9]+, so why are you asserting it isn't? Previously Q93868746 has nothing to do with Q2004972, and Q2004972 isn't restricted to [1-9][0-9]*. Midleading (talk) 05:46, 23 April 2023 (UTC)
I can't believe you removed all descriptions again [9] while saying you didn't remove them! Please look at [10] before making edits! Q93868746 is "identifier in the format of a number, regex [0-9]+". Q2004972 is "numeric or alphanumeric code (without independent meaning) used for labelling or identification". "Q1", "2nd", "III" and "3a" are instances of Q2004972 but not Q93868746. "0000" is an instance of Q93868746. Midleading (talk) 13:18, 28 April 2023 (UTC)
Re "!" and "!" and "!" - Cool down and be more careful with your claims. Q2004972 is "nominal number" and has an English Wikipedia article attached. Q93868746 is "numeric ID" and has no Wikipedia article or any other source attached. GeoGQL (talk) 21:18, 5 May 2023 (UTC)
The English Wikipedia article you mentioned made it clear that letters are allowed in "nominal number", not only digits, and that's why Q2004972 doesn't have a regex:
For the purposes of naming, the term "number" is often used loosely to refer to any string (sequence of symbols), which may not consist entirely of digits—it is often alphanumeric. For instance, UK National Insurance numbers, some driver's license numbers, and some serial numbers contain letters.
In contrast Q93868746 only allows digits in the form [0-9]+. Please don't remove all descriptions and deny the definition. There is no item for [1-9][0-9]*. Midleading (talk) 14:18, 6 May 2023 (UTC)
Your edits just removed all original descriptions and statements. You can't remove the evidence then pretend it doesn't exist. Q93868746 and Q2004972 are two different items with different IDs. Wikipedia article linked to Q2004972 is not the definition of Q93868746. Midleading (talk) 04:09, 16 May 2023 (UTC)
News : new version for the classification gadget
Latest comment: 3 years ago1 comment1 person in discussion
Detect loops in the superclass tree, shows them. Classification gadget screenshot — color blindness loop
Detect if a class is a subclass of several classes listed as disjoint with a disjoint union of (P2738) statement. Classification gadget screenshot page "sigmoiditis" disjointness conflict graph visualisation
be deactivated if it bothers you.
It has also been refactored and shows its informations in a colored box, so if you want to discuss the aesthetics or would like to improve the UI, please talk. author TomT0m / talk page11:54, 16 May 2023 (UTC)
How do you add a constraint?
Latest comment: 3 years ago3 comments2 people in discussion
According to the query service, there are at least 420 instances of the former properties in use on road pages (not including other page types where this may be inappropriate). I am fairly new to Wikidata, so I am not certain what to do.
I recently raised a similar question here (it was related to which coordinates should we add to the street, and whether there are properties which would indicate the ends of the street so that coordinates of two ends could be added); the answer was that there are no such properties. Ymblanter (talk) 19:00, 17 May 2023 (UTC)
I think this is a sound assessment, and it covers the topic well. However, is there a property suitable for the direction of an ascending sequence of address numbers? Also, in cases of roads terminating in an intersection of multiple roads, how should this be handled (repeat for each intersecting road or...)? Ilawa-Kataka (talk) 02:16, 18 May 2023 (UTC)
Small bot job
Latest comment: 3 years ago3 comments3 people in discussion
A user has created some 100+ Croatian settlement stubs on enwiki but has not added WD site links. I'd like to fix that by a bot, but I'm not sure if I need to go through formal bot approval or if this small job can be considered a test run ("50 to 250 edits"). Enwiki to Qid(hrwiki) mapping might (Special:Diff/1897794173) or might not (Special:Diff/1897798065) be 1:1 — I'll review every case manually. Ponor (talk) 12:47, 17 May 2023 (UTC)
There is also an item for the company founder, Paolo Venini Q7132258, who was a noteworthy figure in the history of Murano glass in the 20th century (perhaps I should have mentioned this in my original note – apologies for the oversight). As such, I maintain that there should be a distinction between the two. In my view, including the abbreviation (either S.p.A. or & Co.) for the item which references the company would help make this clear to a user without particular knowledge of the subject. [NB: I haver also tried to open a discussion on WP Talk regarding this distinction.] -- Cl3phact0 (talk) 16:36, 18 May 2023 (UTC)
Violation of Wikidata:Requests for comment/Sort identifiers
Latest comment: 3 years ago8 comments4 people in discussion
Wikidata:Requests for comment/Sort identifiers specified ordering of properties having datatype external-id should - with some exceptions - be done in alphabetical order by English label.
Whatever has happened or not happened, is not a "violation" of anything. At worst it is delay b/c prettymuch everything done here is done by volunteers. Getting stampy-footed with volunteers is quite sucky, if you think about it. Maybe try other ways to get progress? --Tagishsimon (talk) 18:00, 18 May 2023 (UTC)
The community decided. The rule exists. The sorting is not done according to the rule.
If something is not according to the rules it is commonly called a violation of the rule, even in Wikidata as can been seen by several thousands of constraint violations listed in hundreds of pages.
This project chat page isn't critical either. Individual items aren't. Still, the idea is that all parts combined serve to further the goals of the Wikimedia Foundation.
The page is problematic yes. Do you know where the technical implementation has been discussed? Or was it a decision by some programmers without RFC or talk here?
RE "I am not surprised that nobody is willing to put effort into edit requests related to this functionality." - then, maybe unprotect the page? Are those that put the edit requests on the talk page included in "all" and are "here voluntarily" too?
I would like to improve the page:
make it more discussable - use "#" instead "*" for lists, so in the discussion one can refer to the numbers in the list, the change does not interfere in any way with the item and property pages
move properties converted from string to external-id into the external-id section
move properties that are qualifier-only to the section qualifier-only
collect properties of datatype URL that should have been created as external-id, and later work on a process to convert them to external-id
After that basic clean-up, there should be more work on logical placement, currently there are groups by type of item that carries a property, there are others by type (not datatype) of property, and the rest is by datatype. Maybe that could be improved.
With a better picture of the situation one could work on a better technical solution: probably trying to get rid of that page or fill it based on settings on the properties.
Priority one is to implement changes that are not controversial.
Why? Because the page is critical enough. There are some human editors that work on the item pages manually. Random placement of properties consumes their time and makes them less likely to enjoy the editing. GeoGQL (talk) 23:35, 18 May 2023 (UTC)
The solution evolved in 2016 from older (2013) user developed Javascript solutions, and the desire to list statements sorted by property on item pages. We had way fewer properties back then. The RfC was held much later, in order to avoid conflicts regarding the "correct" order. —MisterSynergy (talk) 07:55, 19 May 2023 (UTC)
The old javascript gadget used DOM-manipulation which is a relatively expensive operation, especially on items with lots of data. In other words it makes the pages sluggish. It still works though. Infrastruktur (talk) 15:13, 19 May 2023 (UTC)
Latest comment: 3 years ago1 comment1 person in discussion
I had proposed property Gujarati Vishwakosh entry (P9863). The property is widely used by Gujarati Wikipedia. After its use for months and adding many values, I realised that the property needs to change some constraints. As it is a technical thing, I request it here:
distinct-values constraint: It should not have distinct/unique value. A same value can be added to multiple items. It was found that Gujarati Vishwakoash (GVK, Gujarati encyclopedia) has entries which club multiple topics under single title. e.g. 'Tooth and Dentistry' entry in GVK cover Tooth as well as Dentistry.
format constraint: It is found that sometimes GVK URL indetifiers contain English alphabets and English numbers too so please add them format too. Currently only Gujarati alphabets are included in the format.
Labels: Kanton Bern vs Bern, Republic of Ireland vs Ireland and so on
Latest comment: 3 years ago11 comments7 people in discussion
Hello. Wikidata has items about administrative-territorial division, whose labels in some languages look like "type name" + "name proper". For example: Kanton Bern Q11911, Republic of Ireland Q27, Distrito de Aveiro Q210527 or even Autonome Gemeinschaft Madrid Q5756. I think the "type name" is part of the item description, not part of the label. The same information is written on the Help:Label: A label is like a page title, but is the smallest unit of information that names an item (e.g. "Paris", not "Paris, France"). Labels are allowed to be ambiguous - they are disambiguated using descriptions.
Member @Ludo29 thinks that "type name" is an integral part and should be in the label (you can read more on my discussion page). For example, when you talk about the canton of Bern, you must add the word "Canton". This member has undone many of my edits. For example: [12][13][14].
Definition in English Wikipedia: Ireland (Irish: Éire [ˈeːɾʲə] (listen)), also known as the Republic of Ireland (Poblacht na hÉireann),[a] is a country in north-western Europe consisting of 26 of the 32 counties of the island of Ireland.
Definition in German Wikipedia: Bern (Kürzel BE; berndeutsch Bärn [b̥æːrn], französisch Berne, italienisch Berna, rätoromanisch Berna?/i) ist ein Kanton im Westen der Schweiz.
In my opinion, this proves that names can be used without "type name" and so the "type name" should not be a part of the label.
In my case I don't really care if the label is "Kanton Bern" or just "Bern", but in the cases you change the former for the latter you should add the former as an alias, that way it will make it easier for people finding it when trying to add it to a claim. Agabi10 (talk) 10:11, 4 May 2023 (UTC)
Hello. Just to make it clear: There is a canton of Canton of Berne (Q11911) and a city of Bern (Q70). The same principle applies as well to, e.g. St. Gallen, Lucerne, Geneva, etc. Thus, I care, as it is different and adding a correct label helps and avoids confusion. AnBuKu (talk) 14:36, 20 May 2023 (UTC)
@Ludo29: The edits you reverted are almost certainly not wrong. Your reversions may well be wrong - or at least, unnecessary. The Republic of Ireland is commonly known as Ireland and yet this. WD is *very* clear that Descriptions are the place where disambigation happens: per Help:Description: "The description on a Wikidata entry is a short phrase designed to disambiguate items with the same or similar labels". Labels are not. And yet you seem to be arguing for disambigation to take place in the label per your Jura example. If so, that is an incorrect use of the label and a misapprehension about the basis for the label. --Tagishsimon (talk) 14:23, 4 May 2023 (UTC)
I have a problem with someone who adds disambiguation in ship labels, as in Jean Bart (Q1638571). They argue that as British warships have a HMS prefix, French ones, which traditionally don't have a prefix, need something, so "French battleship" is required. I think this argument is nonsens, and is a right pain if a query returns ships from both navies. Vicarage (talk) 06:09, 5 May 2023 (UTC)
Kanton Bern (canton) or Stadt Bern (city) are two different things. Sometimes people here in Switzerland talks just about Bern, sometimes they mean the canton and sometimes the city. Thus, "correct" labels make life of users easier. Just my 2 cents AnBuKu (talk) 22:10, 14 May 2023 (UTC)
Catholic church building in Switzerland IS NOT Kostyol
Latest comment: 3 years ago19 comments6 people in discussion
Please, do not replace "Catholic church building" in statement "instance of" by "Kostyol", an expression which is not known in Switzerland and not part of Swiss wording. Thank you for your understanding. AnBuKu (talk) 13:57, 15 May 2023 (UTC)
So it appears that AnBuKu (talk • contribs • logs) has changed some of these, as well as Cruzate1492 (talk • contribs • logs) who has changed several recently. Ыфь77 (talk • contribs • logs) may also be involved. I don't know how many, because both are prolific contributors and I couldn't find a recent run of matching contribs. (Cruzate1492 says they do not understand or speak English, so we may need translation to Spanish for this discussion. Ыфь77 speaks Russian and not English.)
Now regarding Catholic church building (Q1088552), I think there's an ontological problem here. We need a speaker of such a language to explain to us: is there a substantive difference between a "kostyol" and a "church building" in some way? It seems that "kostyol" is a topic for various Russian languages and it is linked thus, but why is there are completely separate ontology, if kostyol is merely a synonym for "church building"? What makes it different from the term in English that it would merit a different item?
Oh, as I have discovered, somebody was running a script, which replaced ALL (must be a couple of thousands) "Catholic church building" in statement "instance of" by "Kostyol" without any announcement or information ... I'm speechless
14:42, 15 May 2023 (UTC)14:42, 15 May 2023 (UTC)~~
If you give an example, we can see what sort of script, and it might be possible to bulk revert the changes trivially.
The first one is about a church building, the second one is about a term (which usually refers to Polish church buildings). Ymblanter (talk) 18:25, 15 May 2023 (UTC)
Exactly what the English name says: church building, catholic church building, and kostyol which is an article about a term mainly referring to Polich catholic church buildings. Ymblanter (talk) 20:37, 15 May 2023 (UTC)
Не только католические храмы Польши, но и Литвы, а также других восточноевропейских стран (но не России). Translation: Not only Catholic churches in Poland, but also Lithuania, as well as other Eastern European countries (but not Russia).--Ыфь77 (talk) 21:05, 15 May 2023 (UTC)
Только географическому положению: в Польше, Литве и некоторых других восточноевропейских странах любой католический храм называют костёлом. Швейцария в этот список стран точно не входит. Я не специалист в религиоведении, чтобы раскрыть полный перечень стран, к тому же на Украине на западе католический храм может называться костёлом, а на востоке -- нет. Translation: Only by geographical location: in Poland, Lithuania and some other Eastern European countries, any Catholic church is called a kostyol. Switzerland is definitely not included in this list of countries. I am not an expert in religious studies to reveal the full list of countries, besides, in Ukraine in the west, a Catholic church can be called a kostyol, but not in the east. Ыфь77 (talk) 22:00, 17 May 2023 (UTC)
There is a reverse problem for кирха 'kirkha' / kirch / Kirche (luth.) / zbór (kirch (Q1989839)). The English and German descriptions say it is a {term for Lutheran organisations/congregations}, Polish has {local community of faithful Protestant...} (not "term for"), but the Russian description says {church as a building (German or Lutheran)}. It seems кирха and zbór may need to be split? MalachiiGreen (talk) 04:40, 20 May 2023 (UTC)
Польскую страницу однозначно отсоединить нужно. Замечание: изначально элемент был о типе религиозного здания. Translation: The Polish page definitely needs to be disconnected. Note: Initially, the element was about the type of religious building. Ыфь77 (talk) 06:37, 20 May 2023 (UTC)
The unsolved backend scaling issue
Latest comment: 3 years ago4 comments2 people in discussion
I just talked at the Hackathon with @Fuzheado about the issues with the backend.
WMF just finished the review of the current possibly viable SPARQL-compatible backends. None of them can solve the scaling issues in any way feasible. We are in a bit of bind. The board has not allocated resources for solving the issue and it really needs to be adressed.
Tinkerpop is based on the Gremlin traversal language so converting RDF to that is not a trivial task and few have done it. I found rdf2gremlin.
We could add a feature like the current MWapi-keyworkd to Blazegraph to make it possible to do this kind of "pre-search" for queries that are currently timing out/requires more items than Blazegraph can ingest.
It's an interesting topic, but if you're looking for a good and fun discussion you should get into specifics.
For starters what makes queries with gremlin different from SPARQL? I'm entirely unfamiliar with it. SPARQL allows some expensive operations such as string manipulation, sure. This can increase the amount of compute the server has to do.
As for horizontal scaling, I will say this for starters. Blazegraph already does sharding. And if you intend to spread data across computers in a cluster, you will then get a lot more network traffic. The solution that MapReduce uses is it stores a small chunk of all the data, so that a part of the computation can be performed locally on that machine, which means it won't saturate the network as much. This model works fine with document databases, but it is not immediately obvious if you can even apply this to data stored as triples.
The second point I will make is that triplestore was chosen for Wikidata to begin with because this is very nice fit for interconnected and non-uniform data. The data stored on Wikidata is not a natural fit for a relational database (ignoring that you can can still technically use it as a backend).
I believe the WMDE is also looking into a modernized version of the Label service that does not suffer from the caching-hostile access-patterns that SPARQL enforces. Very cool I might add. Infrastruktur (talk) 17:48, 19 May 2023 (UTC)
JanusGraph looks interesting since it uses NoSQL backends. I couldn't find any description of how it uses those backends, this is an important detail. Are there any description of how it works any place on the website or elsewhere? Benchmarks? Infrastruktur (talk) 18:15, 20 May 2023 (UTC)
Found this on the website: "When the graph is small or accommodated by a few storage instances, it is best to use random partitioning for its simplicity. As a rule of thumb, one should strongly consider enabling explicit graph partitioning and configure a suitable partitioning heuristic when the graph grows into the 10s of billions of edges." This makes me wonder how well they have managed to fit the graph model on top of a bigdata model. Infrastruktur (talk) 18:28, 20 May 2023 (UTC)
Blocking abuse filter for impossible self-referencing
Latest comment: 3 years ago4 comments3 people in discussion
Hi y'all,
We already have an abuse filter for self-referencing (Special:AbuseFilter/49, @Matěj Suchánek:) but it's non-blocking. It shouldn't be as there is some strange edge case where self-referencing is possible - fictional timetravelers for instance - and even sometimes wanted - for country (P17)... Also, obviously, blocking the entering of data is not something that should be done lightly.
But for some specific and a limited number of properties, it's impossible to be self-reference. In particular said to be the same as (P460) and different from (P1889), I regularly remove these kind of self-references for almost 10 years and I never noticed a false-positive (unlike other properties). Wouldn't it be a good idea to have a blocking filter in this case ?
first line (P1922) seems to me an example where self-referencing makes sense. A novel is a good source as a reference for it's own first line. Recording information of how the author name is stated on the book would also need self-references. ChristianKl ❪✉❫ 17:52, 15 May 2023 (UTC)
@ChristianKl: yes absolutely, first line (P1922) falls into the first category (which is the vast majority of properties) where the current filter is more than enough. My suggestion is about the other case and for now only the two properties I explicitely mention (maybe we could add a few others later, but it will always be a small numbers). Also first line (P1922) rarely trigers the filter anyway, while said to be the same as (P460) and different from (P1889) are quite often wrongly used ; hence the need for a stronger filter.
Is there any objection or obvious things I might have missed? (otherwise, in a few days/weeks, I'll ask for this filter creation on Wikidata talk:Abuse filter).
The problem is that many of such pages on various Wikipedias are both an article about what the title means and the history of the title, and then a list of people who held that title. In a way, that makes them halfway houses between articles and lists, which is quite annoying for categorisation and linking purposes. Nederlandse Leeuw (talk) 13:11, 21 May 2023 (UTC)
Q17375735
Latest comment: 3 years ago4 comments2 people in discussion
Q17251820 and Q17375735 were always Mesembryanthemum inachabense; Q96171854 is Mesembryanthemum longipapillosum. The title "Mesembryanthemum longipapillosum" is a redirect to "Mesembryanthemum inachabense" in some sites, which were using a source that said they were synonyms - some of the identifiers at Q96171854 say the same thing and the Catalogue of Life ID (P10585) is a redirect. What possibly happened is that when af:Mesembryanthemum longipapillosum was linked to a redirect on another site, a link was made to the target of the redirect instead. Peter James (talk) 12:37, 21 May 2023 (UTC)
I corrected the label and added the links to Mesembryanthemum longipapillosum (Q96171854); I don't know if there are any statements that could be added to say that it is sometimes considered to be a synonym. The redirects linked to in Q96171854 and articles in Q17251820 were created by a bot, and there are probably many more similar. Peter James (talk) 15:38, 21 May 2023 (UTC)
Do wikidata allow reusing redirected items?
Latest comment: 3 years ago165 comments14 people in discussion
Absolutely not. @Seanetienne: please do not do that ever again. If you want a new item, create a new item. If you change the meaning of existing items - redirects or not - then you break wikidata in the most fundamental way. I have reverted your edit. --Tagishsimon (talk) 23:48, 12 May 2023 (UTC)
My view is that a user who is warned against reusing merged items in January 2023, and reuses a merged item in May 2023, is asking to be blocked. --Tagishsimon (talk) 00:12, 13 May 2023 (UTC)
Ouch! Such disheartening comments... First of all, I do recognize the importance of integrity of Wikidata items. Yes. If the items are replaced, existing internal and external links are broken — the rationale of prohibiting such practice. On the other hand, apart from this rationale, I also believe in being economical. Given how simple it is to create new item on Wikidata than an article on Wikipedia (hundreds can be produced within seconds by bot) news items could be created at whim. The random items, which can be expected not going to be linked anywhere, are either deleted or become redirects, resulting in item numbers being wasted. The amount is expected to be huge (number of active items on Main Page vs newest item number).
Instead of creating new ones and inflating the item number in vain, why not salvage these and turn them into something more beneficial and with expansion prospect (figuratively giving them new life)? In the case of Q117476381, it is a randomly created item without significant history (They are not going to be linked anywhere, are they?). Dedicating it to a heavily reported incident is a more ideal treatment in my view (I even considered its creation date to let it seem reasonable!). As for the editing history 132.234.229.249 cited to suggest I am a chronic disruptor, the "removed redirect" edits shown are done for various reasons and are different in nature and NEVER have I built items on items with long history of reference value. If had been a vandal I would have been banned already! Seanetienne (talk) 03:59, 13 May 2023 (UTC)
@Seanetienne: There is no such thing as being "economical" with Wikidata items. You do not save server space or increase efficiency by reusing redirects. It costs nothing more to create a new item, and it saves everyone the trouble, both potential and real, of reusing the redirects. In the future, create a new item. — Huntster (t@c)06:21, 13 May 2023 (UTC)
@Seanetienne: It's unclear to me whether "OK... Thank you for your feedback." is agreement that you will not reuse redirected items in the future, or a brushoff. Please confirm which it is. There's also the matter of the redirects you have reused in the past, all of which need to be reverted back to being redirects. Will you be doing this? --Tagishsimon (talk) 12:24, 13 May 2023 (UTC)
Not anymore henceforth. No, I will not act retrospectively; the redirects reused are not going to be reverted as they are already well established with internal and, for sure, external links. Seanetienne (talk) 14:40, 13 May 2023 (UTC)
They are going to be reverted. The only question was whether you were going to join in cleaning up your mess, or leave it to others to do. --Tagishsimon (talk) 15:26, 13 May 2023 (UTC)
And the disadvantage of you not doing the reverts, is that your work gets thrown away. No-one owes you the courtesy of saving that work whilst you insist that your (good faith) vandalism should persist. --Tagishsimon (talk) 16:35, 13 May 2023 (UTC)
While you might be correct. What use is it tearing the poor man a new hole? I'm sure he's received the message. Anyways, I've done my share of mistakes, but I've always cleaned up after myself. Infrastruktur (talk) 17:05, 13 May 2023 (UTC)
It's just the way it is. The items need reverting. Retaining the new data is additional work which I'm not up for doing. More salutary lesson than hole. --Tagishsimon (talk) 17:09, 13 May 2023 (UTC)
Reverting won’t be enough in many cases. When some time have passed, we are better off deleting the repurposed items lest we create even more chaos. I have started to do this for two items (see list below). --Emu (talk) 23:24, 13 May 2023 (UTC)
Some of the repurposed items are linked from elsewhere, such as Q95339544. Should these be left alone? And should anyone helping bother to create new items, or just revert? Infrastruktur (talk) 10:00, 14 May 2023 (UTC)
I think it’s a matter of timing. If the repurposing took place a very short time ago, reverting is the preferred option. If it happened months or years ago, deletion because of conflation makes more sense. Linking from VIAF is not a problem (VIAF clusters are transitory by nature), other projects should probably be approached via email if deemed important. --Emu (talk) 12:49, 14 May 2023 (UTC)
If if happened months or years ago it is probably better to leave it, particularly if it was originally an internal item. An alternative would be to use statements to suggest alternative items. Deletion just makes it more difficult, and when the items are not linked in the deletion log there's no way of knowing the replacement item. Peter James (talk) 15:13, 14 May 2023 (UTC)
I agree that the items should be mentioned in the deletion rationale. I don’t agree that it’s a good idea to risk the integrity of our Qid system just because some time has passed. --Emu (talk) 16:14, 14 May 2023 (UTC)
I deleted some items from 2021 and earlier, the ones that wasn't already reverted. Anyways, it looks like most if not all the items are unconflated now. Infrastruktur (talk) 15:43, 14 May 2023 (UTC)
I found what one was (Q94732662) and the replacement item has no statements and a different label (an alternative name). It was created today from an empty item with a different label (unrelated to the current use) that once had a Commons sitelink, so it looks like the repurposing has not stopped, the difference is that now empty items are being used. Peter James (talk) 16:34, 14 May 2023 (UTC)
I've reverted a bunch of these, and then sent the reverted item for deletion; only going back to ~13 May. Unsure if Seanetienne was doing this in earlier periods. --Tagishsimon (talk) 18:08, 15 May 2023 (UTC)
I created a brand new item like Chiara Taigi (Q118370443) (duplicate of the repurposed form of the item)
I fixed sitelinks, MnM links, etc.
I then deleted the repurposed item with a note explaining what happened (see for example Q94733006)
Since there seems to be some controversy about what’s the right way to do it, I will now pause and wait for comments if this course of action is acceptable. Thank you! --Emu (talk) 19:37, 14 May 2023 (UTC)
It is a lot of work to manually type in the information from deleted items. If I had known in advance I could have just copied the JSON blob from an appropriate revision and created the items using WB-CLI. Infrastruktur (talk) 19:46, 14 May 2023 (UTC)
Revert or delete are the only options that work for me. I don't think WD should tolerate the meaning of items being changed b/c that undermines the whole endeavour. Whatever we do, the damage is already done; there's no good outcome. --Tagishsimon (talk) 19:48, 14 May 2023 (UTC)
Tagishsimon, Seanetienne, Infrastruktur: I think our work is mostly done. I did not fix most items where the meaning changed from one category to another. Those changes in meaning shouldn’t happen of course but the damage off-project is probably zero and my time is limited. --Emu (talk) 20:22, 19 May 2023 (UTC)
Ὠ just a comment on cleaning up computers are full of files and archives but all drill to drive on electricity. It is the copper thieves or thefts that break the connection but worst of all is the collection of archives that seem to want their collection in the collection and treated with respect and no shouting. Sadly there is also the spiritual collection on religeon with a full history so if you know how to count foward and haven´t done any cleaning then it is still there amongst the collection.How did the plague set or get in well why do name the flowers and fauna yesterday today and tomorrow for what grew , still growing and still to grow on. Everything is a souvenier including soaps but you don wash with what is dirty or recycled water so with fresh water to remain clean and smell clean you leave it clean.just like dirty souveniers you throw away so to dirty water and wasted files of no interest on a program that is not changing of a doctrine for silence to the silent to keep silent but then information is for publishing and dirt is for criticizing it. 195.235.124.13014:25, 22 May 2023 (UTC)
We don't generally make items for Wikispecies categories in that series; but should we link them to our items where they already exist?
In other words, should current practice change, where such items already exist, given that we do not create items for categories that exist on Wikispecies alone? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits14:42, 22 May 2023 (UTC)
Wikidata weekly summary #573
Latest comment: 3 years ago1 comment1 person in discussion
Here's your quick overview of what has been happening around Wikidata over the last week.
Discussions
CEUR-WS (Task: CEUR-WS Semantification bot to be used on publishing Workshop proceedings via http://ceur-ws.org)
Wikimedia Deutschland is launching a laptop donation program to support Wikidata contributors with limited access to technology. If you've been actively contributing to Wikidata for at least a year, you can request a free laptop to support your work. WMDE Hardware Donation Program#Applying
Data Donation City of Zurich - How has data donation be done?
Latest comment: 3 years ago11 comments6 people in discussion
We (the statistical office of the City of Zurich) would like to donate statistical data to Wikidata. Examples: population data for the entire city (Q72; https://s.zazuko.com/2LGEQmH), or smaller areas such as 'Kreise' (e.g. Kreis 9, Q456159, https://s.zazuko.com/2b2JNhb) or 'Quartiere (e.g. Albisrieden, Q80797, https://s.zazuko.com/2zeSYPL).
How should we donate the data (including annual or more frequent updates)? asked Klemensrosin - Friendly speaking, I have no idea, how this has to be done in Wikidata. Is there a techie or an administrator, who can tell us how to do or can help? Thank you a lot for your support AnBuKu (talk) 14:25, 20 May 2023 (UTC)
Hello, for example in d:Q72 the population has been added on a yearly basis for the last years. If a wikidata object gets "to big" eventually (e.g. maybe by adding numbers quarterly ore more often numbers for a lot of years), then it might take longer to open an wikidata object in the browser (and scroll down to IDs, for example), for example with chess players and ELO numbers, like
What might be a possible use case for quarterly numbers (instead of yearly numbers)? As far as I know, most articles show/update the population numbers on a yearly base. M2k~dewiki (talk) 15:46, 20 May 2023 (UTC)
Latest comment: 3 years ago2 comments2 people in discussion
What happened to the item labeled "set" with description "group of items held singly"? It's a common parent class, but can't find it now. Daask (talk) 20:18, 22 May 2023 (UTC)
Any reason as to why 'Encyclopedia.com' is not included in WikiData?
Latest comment: 3 years ago6 comments3 people in discussion
When musing around Wikidata's encyclopedia entries I discovered the entry "Electronic library encyclopedia.com.ua (P11645)", created 03:07, 13 April 2023.
To my surprise I found no entry for the well-known and useful "Encyclopedia.com".
I then searched within Wikidata for "encyclopedia", ".com" and other plausible phrases. Also ran through the Wikidata list "Wikidata property related to encyclopedias (Q55452870)". Results in all cases: Electronic library encyclopedia.com.ua YEA, Encyclopedia.com NAY.
Then looked up the site's relevant Terms of Service "https://www.encyclopedia.com/TermsOfService" and compared these to the Encyclopædia Britannica's Terms of Use "https://corporate.britannica.com/termsofuse.html", which I found to be overall rather similar, albeit my lacking proficiency in USA legal terminology.
Is there any obstruction, legal or otherwise, that I am not aware of that has unto this day prevented adding "Encyclopedia.com" as a new Wikidata item? If, in your opinion, no such impediments exist, I would introduce this as a new item to the best of my abilities.
I have added a number of Wikidata items before, mainly concerning camera lenses, and am quite knowledgeable about Wikidata statements, regular expressions and software in general, but this is another theatre, possibly riddled with legal caveats. Especially my place of residence, Germany is "renown" (in a very negative sense) for its world-wide singular legislation that allows even unintended breaches of licence by simple commoners to be punished with 4- or 5-digit fines in Euros. Just to be sure, I would appreciate feedback from one or more Wikdata admins before setting off with such as task. Franz van Duns (talk) 14:58, 23 May 2023 (UTC)
PROBLEM SOLVED! Many thanx for the highly useful link to "encyclopedia.com (Q48968006)". I just wonder why I didn't find it on my own.
I will now create the missing "encyclopedia.com ID" in a basic version for use as a Wikidata identifier and will add further statements in due course. Franz van Duns (talk) 16:20, 23 May 2023 (UTC)
SORRY! I noticed your hint just too late and had in the meantime mistakenly created a Wikidata item encyclopedia.com ID, which is, as I now know, completely incorrect.
With mainly Wiki Commons / Wikipedia experience, finding out how to do more than basic edits correctly within Wikidata is a bit tricky.
After having read your notice I inserted a short text indicating that the item "encyclopedia.com ID" should be deleted by an admin, as my initial entry doesn't offer an undo button.
Now I will focus on placing a Property proposal, as prescribed, and will follow the instructions there for data type "External identifier".
The general objection is that it's a transitive property inherited through e.g. P131 and/or P17. Obvs, there are antipatterns. Not infrequently discussed, iirc, such as at P421 talk and on here a month or so back (?). --Tagishsimon (talk) 16:17, 23 May 2023 (UTC)
Latest comment: 3 years ago1 comment1 person in discussion
Property:P5097 (Logainm_ID) gives the outline of townlands in Ireland, is there any way to automate "shares border with"? Something similar was done for USA human settlements. RAN (talk) 18:23, 23 May 2023 (UTC)
Fictional members of groups as well as real ones
Latest comment: 3 years ago4 comments4 people in discussion
I was trying to produce a list of (United Kingdom) members of parliament. To my surprise, the list I got include several fictional characters (e.g. James Hacker). To be complete, my program is clearly not right, as it only gave a small number of outputs, some fictional, others real. But I am wondering if there is a general solution to finding people who hold some role in real life, but omitting those who hold the same role in fiction.
@Ja malcolm - as a little bit of background, this one is a bit of an unusual case because for various reasons we model people as member of the 58th Parliament of the United Kingdom (Q77685926) (the current one), 57th Parliament, etc. A side-effect of that is that the ones who are just generically "UK MP" tend to be fictional, because they don't line up neatly to a real parliamentary term.
If you're wanting to do some stuff with the MPs data, then hopefully the building-queries page is helpful, but I'd be really happy to help with any additional questions - please feel free to ask here or drop me an email. Really keen to see it get better use :-)
More generally, a good way to do queries like this is to look up one or two examples of things you expect to find and see how they're modelled - this usually flags up why the approach you're using isn't working. It might be that the data model is slightly indirect (as is the case here), or that it uses a different property from the obvious one, or (if unlucky) that the data just isn't consistently modelled. But either way, working "backwards" can be really helpful. Andrew Gray (talk) 12:05, 24 May 2023 (UTC)
Raul Sarinana
Latest comment: 3 years ago2 comments2 people in discussion
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Bovlb (talk) 19:13, 30 May 2023 (UTC)
Viewing angle of subject
Latest comment: 3 years ago5 comments4 people in discussion
Thanks, @Tagishsimon and @Pelagic. Suitable values are harder to find than I would have thought. "Direction relative to object - rear" has the right semantics, but with current labels "direction relative to location - backwards" (backwards (Q16938807)) is unlikely to be used. Is it meaningful to have separate direction items for "to the back" and "from the back"?
Part of is an inappropriate choice here. There is, too, a copyright question attached to the import of instances of a proprietary ranking schema, which may well be why there is not a property for this information. --Tagishsimon (talk) 13:02, 23 May 2023 (UTC)
Latest comment: 3 years ago6 comments3 people in discussion
Hello, I need an advice. The English article about the person (the corresponding item is in the title) has been created recently, but with the naked eye it's seen that it's extremely biased and is written in a propaganda point of view. I must admit that it's not totally libelous but it's written in a deliberately repulsive manner which also has some contradiction with the data in the Wikidata item which is more-less neutral. How correct would it be if I'm going to keep on hindering/undoing the connection between the article and the item here? Is it possible to add the sitelink (reluctantly) in such a way that the enWP information won't be imported here and messed up the local data? Also, I just realized that the badges like 'not proofread' or 'problematic' can be assigned for a specific Wikipedia link, do they fit here? What are they really intended for? Thanks in advance for any answer, and sorry if my problem seems too weird to somebody. Wolverène (talk) 04:56, 24 May 2023 (UTC)
We are only concerned about the facts recorded on WD, which in this case are minimal. You could change the neonazi part of the label to something else that honestly distinguishes him from other people with the same name, and keep an eye on other statements being added without references (wikipedias are not considered good references). You should not try to break links to wikipedia entries. We are not interested in politics here, edit wars are strongly discouraged. Vicarage (talk) 08:32, 24 May 2023 (UTC)
I see, the links shouldn't be unconnected in any case. I have renamed the English article ('militant' seems neutral) and merged the items, gonna see what will happen next. --Wolverène (talk) 09:24, 24 May 2023 (UTC)
On the last point, "not proofread"/"problematic" are statuses assigned to pages by Wikisource - they shouldn't be used on projects that don't use those as a page "badge". s:Help:Page status. Andrew Gray (talk) 21:38, 24 May 2023 (UTC)
Having said that... this report lists the number of Wikisource page status badges assigned to Wikipedia sitelinks, and this report lists them all. Seems like there are a couple of hundred that might need looking at. Andrew Gray (talk) 22:12, 24 May 2023 (UTC)
Latest comment: 3 years ago14 comments3 people in discussion
The City of Zurich question inspired me to work on the 2021 Croatian census data for WD. I have a manually curated mapping for Croatian Bureau of Statistics xlsx data to hrwiki article to Qid for about 6700 Croatian settlements at PAWS. I produced QuickStatements of the form
Before I make 5×6700 batch edits I'd like to know if everything is right with the ones I made in Q1722, i. e. if there are any other statements that should be added. Do I need to change the rank of the most recent entry (and how)? Can someone please take a look. Thanks! Ponor (talk) 13:37, 18 May 2023 (UTC)
@Ponor Those look good to me I think - my only suggestion would be that if there is a landing page for the dataset it might be better to give that as the source rather than the Excel file itself?
Unfortunately you can't change ranks in QuickStatements. You can do this using a tool like wikibase-cli, however, if you feel up to running that instead - it is a little more complicated though. Alternatively, I'd be happy to run a second batch of WB-CLI edits when you're done to upgrade all the 2021 ones to "preferred". Andrew Gray (talk) 16:40, 19 May 2023 (UTC)
"a landing page for the dataset" translates as, make an item for the overall data (e.g. 'Croatian Bureau of Statistics xlsx data 2021') and then use it as a value for a stated in (P248) reference. The 'Croatian Bureau of Statistics xlsx data 2021' item can then have a pointer back to the URI of the XLS file. --Tagishsimon (talk) 16:50, 19 May 2023 (UTC)
Thank you both, @Andrew Gray & @Tagishsimon. I am not sure if you're talking about the same landing page so I'll say this: I've asked, and they said the only guaranteed permalink is to the xlsx file itself. I'd rather have a regular html page as the source as well, but it seems that https://podaci.dzs.hr/en/statistics-in-line/ is the best we can get; as they keep adding more data the link is gonna get more and more buried. Also, they're notorious for changing their web site every too often.
As for the ranks: since I'm rarely here, I don't think I'd have enough patience to learn wikibase-cli; I do have a PAWS bot (python), but even that would take some trial-and-error and I'm not sure if it's worth it (for me), given there are so many other things I need to fix. So yes, maybe a separate run by you, Andrew, would be best. Thanks! Ponor (talk) 18:24, 19 May 2023 (UTC)
Well thanks, @Andrew Gray! I didn't want to cause you too much trouble, but if you insist... :) Is there any way for me to help you? I did everything in a number of batches. I can (at least) provide you with Qid's, if needed. Ponor (talk) 19:21, 24 May 2023 (UTC)
@Ponor You can tell me if this makes sense :-). I make it 6716x items with a population value in Croatia, and where there is a 2021 census statement (report). Of those, 48x already have one statement marked as preferred, no items have 2+ preferred. We need to:
where there is 1 normal statement, leave it be [5589 items]
Does that all sound correct? So about 1100 items will need editing. Group #4 might need a quick check by hand but should be easy enough to do. Andrew Gray (talk) 20:46, 24 May 2023 (UTC)
@Ponor Okay! I extracted one from group #2 and did some test edits - this shows what the script does. It unfortunately has to make two edits (one to make it preferred, one to add reason) but it should be straightforward to run both scripts. I will do them shortly. We can then sort out batch #4 tomorrow, and do a final check to make sure that #3 all have the reason listed. Andrew Gray (talk) 21:28, 24 May 2023 (UTC)
Everything in batch #2 done now, so we only have the complicated cases to sort out. I'll have a look at them tomorrow night if they're still there, but if you would prefer to do them by hand, please feel free :-) Andrew Gray (talk) 22:22, 24 May 2023 (UTC)
I don't think they're still there and our job is done ;) I did not set the reason P7452 for those I fixed, but I think we can live with that. Thank you very much, @Andrew! Ponor (talk) 14:57, 25 May 2023 (UTC)
@Ponor Hurrah! And thankyou for the impetus to sort this out - I'd been meaning to figure out how to work ranks like this for a while, so it was a good learning experience. Andrew Gray (talk) 21:53, 25 May 2023 (UTC)
performers in a music festival
Latest comment: 3 years ago3 comments3 people in discussion
Hello - I was working on adding performers who partipated in a music festival last summer - I believe I asked about this and 'participant' was suggested. I see now that there are exclamation marks wherever I've added the musicians under 'participant', so looks like I've got that wrong. What is the best way to add musicians who were part of a festival if not using the participant property? For a performance it is cast member, and there's performer (actor, musician, band or other performer associated with this role or musical work) but a festival isn't a musical work, so I'm not sure? JollyJamboree (talk) 00:33, 25 May 2023 (UTC)
@JollyJamboree A "participant" could be anyone doing anything at such an event, I think something more specific would be better. At least in the Rock & Pop world (and sports), the term used for this is line-up (Q55064180). We don't seem to have a property for that. El Grafo (talk) 08:29, 25 May 2023 (UTC)
Latest comment: 3 years ago5 comments4 people in discussion
Hello All! I needed a help with getting a list of articles from a specific Wikipedia which do not have links to any Wikidata entry. For example; i am working mainly on Marathi Wikipedia these days; which is one of the many Indian languages. It has nearly 92k articles. But it would be helpful if i could get a list of all articles within this wikipedia which are completely unlinked with any Wikidata entry. §§Dharmadhyaksha§§ {T/C} 05:24, 26 May 2023 (UTC)
@Dharmadhyaksha:: Just passing and noticed your question as it was the bottom of the page and as it happens I have done and am doing similar work on enWikiquote. Does querying unconnected pages in mainspace help? That list will probably or mostly be in descending order of creation but that is not guaranteed. I have to thank someone, I think it could have been Ferien, for pointing out Special:UnconnectedPages to me, but it needs to be on mainspace only (namespace=0) to be useful for these purposes. Also be aware that there can be a delay of usually say 10 minutes before a sitelinked wikidata item comes off of that list. Thankyou. -- DeirgeDel tac09:37, 26 May 2023 (UTC)
@MisterSynergy:: Apologies I missed you you mention Special:UnconnectedPages just before my post. I personally find namespace=0 is useful to restrict to mainspaace articles only, certainly where I use it on Wikiquote. Looking at Tagishsimon reply I must look at the duplicity tool if only I have time.-- DeirgeDel tac09:45, 26 May 2023 (UTC)
Selection of the U4C Building Committee
The next stage in the Universal Code of Conduct process is establishing a Building Committee to create the charter for the Universal Code of Conduct Coordinating Committee (U4C). The Building Committee has been selected. Read about the members and the work ahead on Meta-wiki.
Database replication is currently very unwell - https://replag.toolforge.org/ - and according to phab T337446 may not be fixed until the middle of next week. IMO tools which take data from replicas (Petscan?, but probably quite a few more) need to be considered unreliable & out-of-date until the issue is cleared.
Latest comment: 3 years ago5 comments3 people in discussion
Hi all,
I'm trying to import in street names, including the country (Australia) and LGA (New South Wales, City of Liverpool/Camden Council), geographic coordinates and location (suburb). However, I'm getting errors with importing the coordinates.
Here is the sample CSV (using the sandbox):
qid,P31,Len,Den,P17,P131,P131,P625,P276
Q4115189,Q34442,"Brauch Street Test","street in Leppington, New South Wales",Q408,Q3224,Q979106,"-33.988761398912224, 150.79433763238862",Q6527803
Oh… apologies for that. How many decimal places should they be rounded to? Any guidance would be brilliant! I actually hopped onto IRC and was told how to do it, so I did for Leppington but wasn’t aware the decimal places were a problem. - Chris.sherlock3 (talk) 17:45, 28 May 2023 (UTC)
en:Decimal degrees#Precision has a table showing how the number of decimal places corresponds with the precision of the coordinates. 4 decimal places (~10m resolution) would be more appropriate for locating streets. M2Ys4U (talk) 18:38, 28 May 2023 (UTC)
Thank you, I will start rounding to 4 decimal places. I really appreciate the advice! I’m really enjoying contributing to Wikidata, and it’s great how helpful the community has been :-) - Chris.sherlock3 (talk) 19:05, 28 May 2023 (UTC)
Patrolling with scripts
Latest comment: 3 years ago10 comments5 people in discussion
Doing a bit of patrolling using scripts. It's similar to the badwords filter, except I also check for Q-values you would expect to find in bad-faith edits. I also check some combinations of action (create, change, remove etc.), property and value in an attempt to find edits that don't make sense. Currently this covers changes to P31, P279 and occupation (P106) statements.
Do you have some ideas for things to check for? Things that commonly pop up in vandalism and are high signal-to-noise. I'm particularly interested in combinations of action, property and value. You can also tip about suspicious Q-values, but I already have a decent list of those.
I'm actually working on a machine learning powered patrol system and have been thinking along these lines. A lot of vandalism occurs in the free-text fields (title/name/label/description) where just a simple spam-like filter would do well I think. One thing I wish we had was a corpus of "sneaky bad-faith" edits and "sketchy-looking good faith edits". If you have any of those to share I'd appreciate it. BrokenSegue (talk) 18:00, 24 May 2023 (UTC)
That's interesting. How are you training it? I suppose one way to try and detect bad edits that is indistinguishable from good edits would be to analyze user's edit patterns. If a new user is particularly eager to change statements, that's a bit of a red flag. There's already a filter for that, but it doesn't aggregate. I might look into that and see if I get any good hits. Infrastruktur (talk) 18:48, 24 May 2023 (UTC)
This kinda relates to the general problem of patrolling at Wikidata: it is inherently revision-based, which is not ideal for atomic edits (many small incremental edits, rather than many changes in one edit). For Wikidata it is, in many aspects, better to assess at large all edits by a given editor, or all edits made by a given editor on a given page. So, a user-based or page-based revision system would probably be a better fit for us.
Even without changes to the patrol functionality itself, there is IMO some potential to patrol with a user- or page-based approach. However, it requires aggregation of edits by user or by user+page edited, and some useful features need to be defined that indicate bad-faith behavior (or incompetence: good-faith, but poor quality edits). I don't think there is anything really useful available right now. —MisterSynergy (talk) 19:16, 24 May 2023 (UTC)
I'm training it based on revert logs. The data isn't great or close to fool proof (hence why I'm trying to manually curate a dataset) but initial results are promising anyways. I'm also merging consecutive edits together into one larger reverted diff to improve the signal. BrokenSegue (talk) 19:35, 24 May 2023 (UTC)
I see a lot of "immature" edits in "In more languages" descriptions. Is there something on Wikipedia that drives those people to WD? Just noticed Cirkvena (Q753994): Serbian description is "Strange populated place in Croatia". User Andromator191 has made numerous low quality, borderline vandalism edits from his Android app, inserting words such as "čudan/čudn-" (strange) "naj-" (most) "ružniji/lepši/agresivniji/kreativniji" (ugly/beautiful/aggressive/creative), "pre-" divan/lep-" (most beautiful). 300 edits in ~2 days, with little experience on other projects. It would take a human patroller quite some time to clean up after this guy, and the patroller would need to know the language. These descriptions are so prominent on mobile wiki pages and make the whole project laughable. I'm really hoping AI and edit filters (throttle?) would do more about it. Number changes in referenced data (usually added by bots?) are also very suspicious. Ponor (talk) 12:28, 26 May 2023 (UTC)
So I tried my model against @Andromator191:'s edits. [18] was found to be "bad faith" but [19] was thought to be good faith with probability 60%. [20] and [21] were found to be bad faith. But [22] was not found to be bad faith. [23] was correctly identified as good faith. So it's a mixed bag at the moment (I don't speak serbian so this is just my guess on which are bad faith based on gtranslate). BrokenSegue (talk) 22:39, 27 May 2023 (UTC)
Added a new functionality today, showing unpatrolled changes by people who have been blocked at some point in the last year. This gave a whopping 2880 hits, pun not intended. Infrastruktur (talk) 17:32, 27 May 2023 (UTC)
Do you have some ideas for things to check for?
For statement changes, use property constraints. As [a] lot of vandalism occurs in the free-text fields, you can use a heuristics like "if the old value would match the regex and the new value does not, it is a suspicious change." Analogously for enumerations, quantities with bounds (in theory, you could even memorize the distribution of a quantity), etc. (This is what phab:T190529 is asking for.)
Edits that change well-sourced statements are suspicious, too. (phab:T213630)
Removal of deprecated statement from P18 image property.
Latest comment: 3 years ago13 comments6 people in discussion
The image good faith removal in question is here, that being the only image for Johann Gottfried Walther (Q61716)/image (P18). I want to check if that removal is compliant and consistent with Help:Deprecation. The background is that image, long standing for over a decade, was not actually of the subject but a different person. I've asked the actioner @Jmabel: about this and they have questioned me as to why it would be retained and have also expressed concern that if left as deprecated it could still be picked up from infoboxes on other projects. I would like to ask experts here for their answers to those questions. Be aware I have some caused some controversy on commons by nominating the image for removal & also here where Jmabel was really helpful. That said I'd like to see what people here think. Thankyou. -- DeirgeDel tac21:30, 25 May 2023 (UTC)
Now that the image has been renamed and redescribed on commons - commons:File:Thomas Christian Walter 02.jpg - there's no good reason to keep it associated with Johann Gottfried Walther (Q61716). Removing it from the item - as has been done - seems more appropriate than retaining plus deprecating it. Nothing good seems to be achieved by retention plus deprecation not least since, as I understand it, the source of the image no longer makes an erroneous claim about the image. Difficult to see how the image would in the future ever become reconnected with Johann Gottfried Walther (Q61716) and so there is little point in retain+deprecate merely to warn against an improbable eventuality. --Tagishsimon (talk) 22:42, 25 May 2023 (UTC)
I don't think infoboxes will pick up deprecated statements, not unless they are programmed incorrectly, in which case they should be fixed. Infrastruktur (talk) 23:15, 25 May 2023 (UTC)
In general, yes. But in this case the image has no real ties to Walther anymore, so there should be no reason for a bot to try and re-import it. — Huntster (t@c)13:44, 26 May 2023 (UTC)
An interesting set of replies so far. No-one has referred to Help:Deprecation which specifically says: "Statements in Wikidata should be ranked as deprecated (and not removed) if they are: ... now known to be wrong, but were once thought correct." which applied ot the case here. That is consistent with 2A02:810B:580:11D4:F8BB:49DE:73D7:11E0's viewpoint. It has been confirmed that should an infobox pick the depreciated image up then it is the infobox that needs to be changed. And the precautionary approach would seem to return the depreciated statement. -- DeirgeDel tac22:39, 28 May 2023 (UTC)
Well, that's a point of view. I don't know in how many ways we can say to you that there is no plausible connection, now, between the photo and the item, and no point at all in adding the photo as a deprecated statement. --Tagishsimon (talk) 22:49, 28 May 2023 (UTC)
The fact its been wrong on various Wikipedia articles for a considerable period of time is a actual connection; and quite frankly we can't be 100% sure where that has mde its way into to or which offline book or journal or article that has ended up in. And opinions which avoid referencing Help:Deprecation carry less weight surely? Or should Help:Deprecation be scrapped/modified or am I interpreting what it actually says incorrectly? Thankyou. -- DeirgeDel tac23:05, 28 May 2023 (UTC)
WD is riddled with incorrect statements, which when found are routinely deleted and (sometimes) replaced by correct statements. I think, for instance, of the thousands of incorrect P131 and P625 statements on Scottish geolocatable items which have been dealt with over the past few years. Is it your position that every mistake made on WD should instead be deprecated? Deprecation is generally used when an external source references the incorrect statement - where there is a clear, obvious and real risk of the reintroduction of the erroneous value. Here, there is no clear, obvious nor real risk. Nor would adding commons:File:Thomas Christian Walter 02.jpg as a deprecated statement preclude the readdition by a bot of commons:File:Walther-Johann-Gottfried-01.jpg should the same image, with that filename, ever be readded to Commons. Nor is it likely that the same image would be readded to Commons, since the Commons' upload facility spots duplicate images at upload time.
So. You have asked your question, and a bunch of long-time, very experienced wikidata & commons users, all of whom are well aware of Help:Deprecation, have advised against. Your horse is dead, but you continue to flog it. --Tagishsimon (talk) 23:35, 28 May 2023 (UTC)
@Tagishsimon:: Thankyou for your response which now explicitly refers to Help:Deprecation. While seemingly making you very angry, and you have somewhat upset my reference to my dead horse, albeit it was only a pony and it wasn't mine. It would have been better if others have referred to it. While I have pressed this somewhat hard it helps me make community consistent decisions in future based on information from this discussion. I myself have deleted statements at time, not many, but certainly some. And I hope this conversation will enable me to make better decisions in further. Thankyou all for your time and apologises if I have annoyed people. -- DeirgeDel tac23:57, 28 May 2023 (UTC)
I must say that I was also misled by Help:Deprecation, and so if this accepted system-level guideline/policy/whatever is giving bad advice to editors, perhaps we should consider revising the source file, and not castigating users for attempting to follow it. Elizium23 (talk) 00:42, 29 May 2023 (UTC)
Help:Deprecation is not 'giving bad advice'. It seeks to explain what deprecation is good for (and what it should not be used for). It does not exclude the use of deletion as an alternative way of dealing with bad data. The question to be weighed is whether deprecation is proportional to the risk of reintroduction of erroneous values. In this case, for the lengthy reasons set out above, probably not. --Tagishsimon (talk) 01:03, 29 May 2023 (UTC)
Well we shall wait wait. Should the image re-appear then I shall be somewhat amused. Overall in this saga though the good thing is the image misused has been corrected across multiple wiki's. And I won't be laughing if when next week I am sitelinking a Wikiquote article without an image from a wikidata item that only has a deprecated image if I accidently choose to "improve" the article at the same time by adding the image to the Wikiquote article. Hopefully a bot would catch it but I'd be eating the trout regardless. Thankyou. -- DeirgeDel tac08:04, 29 May 2023 (UTC)
Wikidata weekly summary #578
Latest comment: 3 years ago1 comment1 person in discussion
Here's your quick overview of what has been happening around Wikidata over the last week.
Next Linked Data for Libraries LD4 Wikidata Affinity Group call May 30, 2023: We will discuss results from our community survey on advocacy for Wikidata in libraries and hear from community members Mary Aycock, Steve Baskauf and Eric Willey about their experiences with advocacy and gaining support for Wikidata work within their respective libraries. Agenda
TalkPageHeader.js shows the talk page header for languages other than English.
Other Noteworthy Stuff
The libraries of the London School of Economics and the University of York have collaborated to produce a toolkit for uploading research theses metadata to Wikidata. Full announcement and access to the toolkit
The contents of most Wikimedia Cloud Services Wiki Replicas are out-of-date owing to large & increasing replication lag; see https://replag.toolforge.org/ & Phab T337446. Exercise caution using data from any tools that have a dependency on these databases.
ISCED category orientation (International Standard Classification of Education category (based on focus/orientation) applicable to an educational program)
Learning outcome (specific knowledge, skills, and abilities that students are expected to acquire as a result of participating in a particular education program)
Latest comment: 3 years ago5 comments3 people in discussion
I added urban population (P6343) as a qualifier to population (P1082) in Solin (Q642828), but the system complains (qualifier not allowed). Not sure how that's different from male population (P1540) and female population (P1539), which are allowed. In the case of Croatian municipalities and towns, they usually share the name with the most populous/urbanized settlement, but include a few satellite settlements that are more suburban or rural. Those settlements have their own population data but are also included in the municipality's total population. That leaves the "core" ("proper") settlement without its own population data; I thought that should be added as urban population (P6343) under population (P1082) (makes it easier for me because the data source/reference is the same). Is that a problem? Should we allow P6343 in P1082's constraints? Is there some other way to do this? Ponor (talk) 12:54, 26 May 2023 (UTC)
As I'm doing this mostly with Wikipedia(s) in mind, having two Qids would only complicate things. Not many wikis have separate articles for municipalities and their "core" settlements, for many good reasons (...). In the case where there's only one article for the two, it'd be harder to have templates use wikidata because only one Qid is default in function calls (→en:T:Wd) per article. Ponor (talk) 13:56, 30 May 2023 (UTC)
If the suggestion above does not work for you, why don't you use P6343 as a main statement on your municipality items, instead of hijacking P1082? Vojtěch Dostál (talk) 13:37, 29 May 2023 (UTC)
Mostly because of these reasons:
urban population is a subset of total population, just like male/female populations are
the two population data come from the same source (there's no need to repeat references)
after the 2031 census it shouldn't happen that one is still "valid" and the other is not
I was hoping for some widely used solution, but was unable to find one. TBH, I'm not sure that "urban population" is the right description in all cases; some "satellite" settlements are not that non-urban after all: outer parts of the capital city (Zagreb (Q1435)) are probably more urban the "urban" cores of most municipalities/towns. Ponor (talk) 14:19, 30 May 2023 (UTC)
Merge request
Latest comment: 3 years ago2 comments2 people in discussion
Apart from the fact that merging is not possible because of the distinct DE articles, the distinction being drawn on DE is between fibre (a strand of material) and wool (a mass of fibres). Meanwhile, ceramic fiber blanket appears to be a distinct concept, just as (sheep's) woolen blanket is distinct from sheep's wool. --Tagishsimon (talk) 13:47, 29 May 2023 (UTC)
Latest comment: 3 years ago1 comment1 person in discussion
Hello. South Korean television channels have special program blocks that are named like "CHANNELNAME Fri-Sat Drama". This is often also featured in the logo of the show. Moreover, in all Korean wikipedia articles for shows running in those blocks, it is mentioned in the first sentence. And the Korean wikipedia has articles and categories for every block. However, I am not even certain if block is the most appropriate way to say it in English. Anyways, when wanting to add it to the wikidata item, what is most appropriate to use? part of (P361) or part of the series (P179)? The latter one always causes a limitation warning.
And then, probably inside this block, you would use follows (P155) and followed by (P156) for the series that was broadcasted the previous season and the next season.
It happens that some Korean shows have a sequel when it comes to the content, but they also have the chronological predecessor and successor when it comes to the broadcasting slot. I am just thinking how the best way is to implement it on wikidata.
Here are just some items for the program blocks and you can easily find the Korean wikipedia article and category associated with it. Those blocks every few years change the time of the day they are starting. E.g. 22:00 instead of 21:00. And sometimes, they also go on hiatus. E.g., maybe five year long, a channel shows a Thu-Fri drama and a Sat-Sun drama, but then, those go on hiatus and they install a Fri-Sa drama and a Sun-Mon drama instead.
Just thinking how to implement them the best as currently, they are not really considered on wikidata and their items are almost empty.
Latest comment: 3 years ago2 comments2 people in discussion
Hi. Does someone have a bot that fixes links in references? In particular I'm interested in having links to search engines expanded so that they point to their search results. E.g.:
This is important because such links can be abused for tracking purposes or hiding malware. IIRC Google also have exploitable redirect services. Infrastruktur (talk) 16:29, 31 May 2023 (UTC)
We definitely shouldn't have such links, you're right. Honestly a bit puzzled how someone would go about adding them - copying straight from a search result without following the link? Andrew Gray (talk) 19:58, 31 May 2023 (UTC)
Done -- Reise Reise (talk) 15:03, 1 May 2023 (UTC)
Comment: my vision is that anonymous edits should be banned because we don't have power to patrol these edits. If we will have power, we can allow anonymous edits. If anonymous creations are banned, we can focus more on new users (and definitely new users should have limited options in Wikidata, see Wikidata:Administrators'_noticeboard/Archive/2023/02#New_items_creation_to_be_restricted_for_anonyms_and_for_not_autoconfirmed_users?). Some Wikipedias has option that anonymous edits are allowed, but not published until patrolled (not sure about the mechanism, though). Pinging BrokenSegue (talk • contribs • logs)--Estopedist1 (talk) 06:36, 7 May 2023 (UTC)
statement. 
Notified participants of WikiProject Switzerland for information, also @AnBuKu, Fundriver, Eihel:
