Wikidata:Property proposal/Wikipedia suggested article name

Wikipedia suggested article name edit

Originally proposed at Wikidata:Property proposal/Sister projects

Not done

Description	suggested article name for a missing article (red link) in this Wikipedia language edition
Data type	Monolingual text
Allowed values	Dutch
Example 1	Friedrich Ris (Q116510) → Nederlands: Friedrich Ris
Example 2	Wilton Ivie (Q2055054) → Nederlands: Wilton Ivie
Example 3	Danie Visser (Q965762) → Nederlands: Danie Visser
Example 4	Kreischberg (Q875914) → Nederlands: Kreischberg
Example 5	Agapetus (Q10402787) → Nederlands: Agapetus (genus)
Planned use	Add some links for missing articles in the Dutch Wikipedia

Motivation edit

I first brought this up on the Wikidata mailing list to get some feedback.

According to https://www.youtube.com/watch?v=TLuM4E6IE5U : "Semantic annotation is the process of attaching additional information to various concepts (e.g. people, things, places, organizations etc) in a given text or any other content. Unlike classic text annotations for reader's reference, semantic annotations are used by machines to refer to." (more at https://ontotext.com/knowledgehub/fundamentals/semantic-annotation/ )

On Wikipedia a red link is a link to an article that hasn't been created (yet) in that language. Often another language does have an article about the subject or at least we have a Wikidata item about the subject. Take for example https://nl.wikipedia.org/w/index.php?title=Friedrich_Ris . It has over 250 incoming links, but the person doesn't have an article in Dutch. We have a Wikidata item with links to 7 Wikipedia's at Friedrich Ris (Q116510) , but no way to relate https://nl.wikipedia.org/w/index.php?title=Friedrich_Ris with https://www.wikidata.org/wiki/Q116510 .

To keep usage from exploding, some usage rules:

Link only to a non-existent article in a valid Wikipedia language for which the incoming links have been disambiguated
The article has at least one link from other articles in the main namespace and these links are in the article text, instead of only in the (navigation) template

If we have this, we have a connection between a red link on Wikipedia and the Wikidata. We would be able to offer all sorts of nice features to our users like:

Hover of the link to get a hovercard in your favorite backup language
Generate an article placeholder for the user with basic information in the local language
Pre-populate the translate extension so you can translate the article from another language

(probably plenty of other good uses)

Multichill (talk) 18:10, 29 September 2018 (UTC)[reply]

Discussion edit

Oppose I've seen plenty of examples where multiple proposed names - many of them not disambiguated - were (or are) used in red links for a single subject on Wikipedia; and an article eventually created may be at a different name to any used in a red link, even if only one such name occurs. This property will therefore not solve the problem described. The templates described at Template:Interlanguage link (Q13606208) exist to embed links to Wikidata, and to articles in alternative languages, in red links on Wikipedias; even if two Wikipedia articles each use a different proposed name for the needed article. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 03:08, 30 September 2018 (UTC)[reply]
- You are mono-lingual so I assume your only frame of reference is the English Wikipedia. On the Dutch Wikipedia disambiguation often happens for red links, see for example nl:Scott Davis (tennisser) which has almost 300 links and no links to the disambiguation page nl:Scott Davis. The template doesn't exist on the Dutch Wikipedia and I'm pretty sure the Dutch Wikipedia community rejected the concept of using templates in links. So using the template is not an option. Multichill (talk) 17:09, 30 September 2018 (UTC)[reply]
  - I work across multiple projects, and collaborate with editors who are active on many more, so please desist from making baseless ad-hominem criticisms predicated on false assumptions. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:53, 1 October 2018 (UTC)[reply]
Comment I suppose the interesting thing would be to get it to work with https://nl.wikipedia.org/wiki/Speciaal:GevraagdePaginas --- Jura 16:50, 30 September 2018 (UTC)[reply]
- @Jura1: some time ago I made a version of that excluding templates, see nl:Gebruiker:Multichill/GevraagdePaginas. Was thinking about combining that with the new property here (SPARQL mwapi stuff) to get a combined list. For that I would need this template. Multichill (talk) 17:09, 30 September 2018 (UTC)[reply]
Yes, this is what I want to have for at least two years already, a way to connect red to existing WikiData items. Many rivers, people, taxa, etc, are missing as article in many wiki's, especially the smaller ones, but they do exist in red links. Unless we have a better proposal than this one, we should just do it. Edoderoo (talk) 17:25, 30 September 2018 (UTC)[reply]
Support - will not fly anyway on enwiki because current anti-Wikidata climate in general, but this is a door opening to solving the whole issue about Bonnie&Clyde redirects too. It would be nice if this could possibly link to a redirect somehow and not just a redlink. Jane023 (talk) 17:34, 30 September 2018 (UTC)[reply]
Comment This sounds like it is coming with an expectation of WMF developer effort to make this work in some way - have developers been consulted on this already? What prevents doing something like this already using the labels for a given language? ArthurPSmith (talk) 14:44, 1 October 2018 (UTC)[reply]
- @ArthurPSmith: that assumption is wrong. No need for any WMF developer time at all. See nl:Gebruiker:Multichill/GevraagdePaginas met Wikidata for a mock up that uses Commons category (P373) instead of the new property. Labels are not disambiguated so no way to distinguish things. First example I could find is "Agapetus", see Agapetus (Q4691242) and Agapetus (Q10402787). I would disambiguate the links to nl:Agapetus to something like nl:Agapetus (genus), but no way to link it to the item. Multichill (talk) 20:42, 3 October 2018 (UTC)[reply]
  - Support Ok, that's clear enough - you need something like this for disambiguation, and at least on some wiki's there is a disambiguation standard already in place for these redlinks so it would work for them. ArthurPSmith (talk) 20:49, 3 October 2018 (UTC)[reply]
Weak oppose As far as I understand your proposal, your need is mostly covered by Template:Interlanguage link (Q13606208). Visite fortuitement prolongée (talk) 15:57, 1 October 2018 (UTC)[reply]
- @Visite fortuitement prolongée: you understood it wrong. It's not covered by that template. See what I wrote in the introduction and the comment to Andy. If you don't understand it, please explain what part you don't understand so I can explain it better. Multichill (talk) 20:42, 3 October 2018 (UTC)[reply]
  - Among your needs, which are not covered by Template:Interlanguage link (Q13606208)? PS: I do not read nl neither sparql. Visite fortuitement prolongée (talk) 22:29, 3 October 2018 (UTC)[reply]
    - On some Wikipedia's people have tested with local templates (like the interlanguage link one) around the red links. That's not structured data, clutters up the Wikitext, it doesn't scale and the local communities generally don't seem to like the approach. It doesn't even exist in the Dutch language.
    - A template is not structured data so queries like I made a mock of at nl:Gebruiker:Multichill/GevraagdePaginas met Wikidata are not possible. With the new property it's possible to do queries. So instead of the huge unstructured list at w:Special:WantedPages Wikiprojects could set up reports of interesting subjects to write about. Multichill (talk) 10:43, 6 October 2018 (UTC)[reply]
      - Please give evidence that "it doesn't scale". Visite fortuitement prolongée (talk) 11:23, 6 October 2018 (UTC)[reply]
      - Please give evidence that "the local communities generally don't seem to like the approach".
      - Noted. Be happy, I wrote "weak oppose" and not "oppose". Visite fortuitement prolongée (talk) 11:23, 6 October 2018 (UTC)[reply]
        About the scaling, take for example "Agapetus". That's used on about 200 pages. With a template that would be 200 edits. With a property only one edit.
        With your property, you still need 200 edits to change [[Agapetus]] to [[Agapetus (genus)|Agapetus]]. Visite fortuitement prolongée (talk) 10:00, 20 October 2018 (UTC)[reply]
        
        James provided some links at https://lists.wikimedia.org/pipermail/wikidata/2018-September/012468.html about how happy local communities are about Wikidata link templates in article text. Multichill (talk) 11:38, 6 October 2018 (UTC)[reply]
        I see no mention of Template:Interlanguage link (Q13606208) in 012468.html. Visite fortuitement prolongée (talk) 10:00, 20 October 2018 (UTC)[reply]
        Oh it's there; the second of the two RfCs referred to included a proposal to prohibit its use to link to Wikidata; and found no consensus to make any such prohibition. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:05, 20 October 2018 (UTC)[reply]
Oppose with the same question that Arthur has and per Visite fortuitement prolongée. Mahir256 (talk) 16:58, 1 October 2018 (UTC)[reply]
- @Mahir256: see the answers above. Does this address your concerns? Multichill (talk) 20:42, 3 October 2018 (UTC)[reply]
Oppose Why not use badge function instead? --Liuxinyu970226 (talk) 23:23, 1 October 2018 (UTC)[reply]
- @Liuxinyu970226: I don't thing you understand what I'm proposing. This is about red links (articles that don't exist yet). Badges only work on existing articles. So please explain how I could use badges in this situation. Multichill (talk) 20:42, 3 October 2018 (UTC)[reply]
  - @Multichill: What about introducing Template:Internal link helper (Q6462931) to nlwiki instead? This way we can handle "available in other languages" redlinks locally, regardless anything happened on Wikidata. --Liuxinyu970226 (talk) 23:06, 13 October 2018 (UTC)[reply]
    - @Liuxinyu970226: as I explained before: That is not structured data, that doesn't scale and no local community support. No more badges? Multichill (talk) 11:08, 14 October 2018 (UTC)[reply]
    - @C933103: What do you think about that? --125.38.13.141 07:12, 16 October 2018 (UTC)[reply]
      - Why am I getting pinged?C933103 (talk) 07:23, 16 October 2018 (UTC)[reply]
        @C933103:Because this Dutch user think that our Chinese Wikipedia is non-structured (criterial? something must use Wikidata to be structured?) --Liuxinyu970226 (talk) 10:19, 19 October 2018 (UTC)[reply]
Comment given that this based on processing at a given Wikipedia, could we limit its use to languages where this is actually being done? Others some users might think they need to add values for 100 languages on the entity they are currrently editing. Initially maybe the allowed language code should be just "nl". --- Jura 10:50, 6 October 2018 (UTC)[reply]
- You are writing this like the property should be an experiment, for a time. Visite fortuitement prolongée (talk) 11:23, 6 October 2018 (UTC)[reply]
  - Call it a pilot. We could do that and evaluate after a couple of months. I guess in practice without this formal limitation it will at first only be used for one or a few wiki's because it hasn't been picked up yet. Multichill (talk) 11:38, 6 October 2018 (UTC)[reply]
- It requires some support on the use side. If there are no plans for its use on enwiki (e.g. because its users are happy with adding templates to redlinks), there isn't much benefit in generating countless strings in English. --- Jura 11:47, 6 October 2018 (UTC)[reply]
  - So probably first a lot of Dutch for starters. With more data, it's easier to generate nice reports and to develop new tools (gadget) based on it. Once that works for one language, it's easy for another language to adopt it. Multichill (talk) 12:45, 13 October 2018 (UTC)[reply]
Question How do you plan to populate the property? Visite fortuitement prolongée (talk) 11:23, 6 October 2018 (UTC)[reply]
- (Semi-)automatic based on lists like nl:Gebruiker:Multichill/GevraagdePaginas and nl:Gebruiker:Multichill/GevraagdePaginas met Wikidata. Probably also a bit of cherry picking based on whatever writing project is going on (for example female scientists). Multichill (talk) 11:38, 6 October 2018 (UTC)[reply]
Support for Dutch pilot. Suggested alternative template isn't used there anyways. --- Jura 12:55, 13 October 2018 (UTC)[reply]
Comment left a note about this at the Dutch Village pump. Multichill (talk) 14:52, 17 October 2018 (UTC)[reply]
Support Would make it easier to make a good list for Dutch Wikipedia (and possibly later other Wikis) of a list of wanted articles. Mbch331 (talk) 16:07, 17 October 2018 (UTC)[reply]
Support The current listing is not practical. A tryout for an alternative would be appreciated. --VanBuren (talk) 16:38, 17 October 2018 (UTC)[reply]
Support I support to introduce this for the Dutch Wikipedia at least. It is usefull to find which articles have many internal links in Dutch Wikipedia, which is now virtually impossible, because the automatic list is flooded by especially non-relevant species. With this new property, specialised lists can be made to motivate various project teams to write these highly needed articles (like the gendergap team). Ellywa (talk) 17:12, 17 October 2018 (UTC)[reply]
Support. I gave this some thought before expressing support because my initial thought was that this would only be needed when the name of the item is different than what one would expect the name of the article to be, but then if it wasn't there would be no way (without human intelligence) to tell whether that's the case. So yes, we do need this property and I expect it to be useful for Wikipedias other than Dutch as well. Two questions: Do you expect every item that doesn't have a Dutch Wikipedia article associated with it (but has others) to have this property? Relevance criteria might not be the same across all languages. Also, when an article appears, who or what will remove this property which is no longer needed (and I assume, no longer desired)? –Frank Geerlings (talk) 16:52, 18 October 2018 (UTC)[reply]
I don't expect every item that doesn't have an article on the Dutch Wikipedia to have a link, because not every item will have a red link on Wikipedia. See also above "The article has at least one link from other articles in the main namespace and these links are in the article text, instead of only in the (navigation) template".

When an article gets created, this property should be removed and contents moved to the sitelinks. I'll probably setup some reports and a bot to handle newly created articles. Shouldn't be too hard to combine nl:Speciaal:NieuwePaginas with SPARQL. Multichill (talk) 17:44, 18 October 2018 (UTC)[reply]
Thanks for clarifying. –Frank Geerlings (talk) 16:48, 19 October 2018 (UTC)[reply]

A few hours ago, I removed the "Ready" marker from this proposal because the required consensus does not exist. I have been reverted by the proponent with a claim that "The fact that you don't agree doesn't mean we don't have consensus". On checking the discussion here, and confirming that there is no consensus, I see also that at least two supporters arrived after being canvassed on another project. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:55, 19 October 2018 (UTC)[reply]

Arthur was pretty clear in his edit summary. You're not giving any new arguments. Are you trying to just frustrate my proposals? Multichill (talk) 21:08, 19 October 2018 (UTC)[reply]

Observers will note that I don't even comment on the majority of your property proposals (here, for example, are your last three before this one: a, b, c; covering a period of one quarter of a year. I commented on none of them). I've already asked you once on this page to "desist from making baseless ad-hominem criticisms predicated on false assumptions". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:14, 19 October 2018 (UTC)[reply]

Observers will note that you are not answering any questions. So new arguments? Just meta wikilaywering?

Please use the "Show preview" button before pressing "publish changes", it took you 4 edits and that caused an edit conflict. Multichill (talk) 21:28, 19 October 2018 (UTC)[reply]

The only question you (or indeed anyone else) asked me is "Are you trying to just frustrate my proposals?" You can see my answer to that in my preceding comment. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:06, 19 October 2018 (UTC)[reply]

You didn't actually just answer it, you just talked around it.

I have another question, maybe you're able to answer that one: Do you have any new arguments in favour or against the creation of this property? Multichill (talk) 22:15, 19 October 2018 (UTC)[reply]

@Multichill: And can you please please and please ask other Wikipedias that if they kindly wanna this or not? Tones like "it's easy for another language to adopt it.", TBH, are a bit more argumentum ad populum (Q251695) than useful suggestions if you don't have proofs of that. --Liuxinyu970226 (talk) 11:34, 20 October 2018 (UTC)[reply]

But, @Pigsonthewing:, I eventually can't agree that there's only one question that you mentioned, there are also other questions that you may faced-to-faced: <a name="2q">1. Do we need properties that are only intend for only one or two, or not more than 10 languages? 2. How do we ensure that such properties are well-used? I ask this issue because of Commons compatible image available at URL (P4765), that property can also have abuse ways that linked a file that has URAA problems (don't surprise, this problem is very common in ESEAP countries and Canada) that simply public domain in these areas but non-PD in the United States.</a> --Liuxinyu970226 (talk) 11:34, 20 October 2018 (UTC)[reply]

Given that we now have further opposes, I have again removed the 'ready;' marker. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:59, 21 October 2018 (UTC)[reply]

Oppose per Andy Germartin1 (talk) 14:17, 21 October 2018 (UTC)[reply]
Oppose. There are several issues with storing redlink targets on Wikidata, especially if they're not integrated in some clean way with the client wikis, but even if we were to store these here, the correct way would be to integrate it directly into the sitelinks system, rather than using statements for each link. --Yair rand (talk) 16:57, 21 October 2018 (UTC)[reply]
@Yair rand: Your suggestion here would definitely require developer effort to accomplish - in principle it sounds like a nice idea, but given that we've had zero support from the development side for the strong consensus to allow sitelinks to redirects, adding sitelinks to pages that don't even exist yet seems like a real stretch. Are you willing to put in effort to push for this on the development side? There's going to be a need for a detailed specification, this is definitely a nontrivial substantive change. This property proposal is close to what the sitelink option would give but it's something we can do without developer effort, so that'a a huge plus. Maybe once this property exists it would give the developers a hook to allow such sitelinks too (otherwise how would the software know what is an allowed string for linking?) ArthurPSmith (talk) 14:20, 22 October 2018 (UTC)[reply]
And @Germartin1, Liuxinyu970226, Pigsonthewing: - Multichill has repeatedly explained that the template option is a nonstarter, and has discussed a number of additional benefits such a property would bring. What is your proposed alternative for this instead of the template? Or do you commit yourselves to going over to nlwiki and convincing users there to create and use this template to at least partially address the problem here? ArthurPSmith (talk) 14:20, 22 October 2018 (UTC)[reply]
- @ArthurPSmith, Lydia Pintscher (WMDE): And can you please answer my <a href="#2q">two questions</a> above? --Liuxinyu970226 (talk) 15:09, 22 October 2018 (UTC)[reply]
  - @Liuxinyu970226: On properties intended for only 1 or a small number of languages - we certainly have some of those already (and not just for lexemes) - for example classifier (P5978), stroke count (P5205), revised romanization (P2001) etc. We also have properties associated with a single wiki site (generally Commons): Commons category (P373) for example. As to "ensure that such properties are well-used" - that would be a question for any property, and some of them certainly are not well used at all. Having a large active community (as nlwiki does) is the main thing that would help in this regard I think. ArthurPSmith (talk) 17:33, 22 October 2018 (UTC)[reply]
    - P373 is used by several Wikipedias, and Wikispecies. P5978 is used for Vietnamese, Chinese, Cantonese and Japanese lexemes; not for Wikipedias in those languages. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:20, 22 October 2018 (UTC)[reply]
      - @ArthurPSmith: And how do you think about Commons compatible image available at URL (P4765)? --Liuxinyu970226 (talk) 10:57, 24 October 2018 (UTC)[reply]
        @Liuxinyu970226: I don't have any personal experience with P4765. If people are abusing it (linking to images that do not meet the property criteria) then those abuses should be addressed by more experienced editors - deprecate with a "reason for deprecation" pointing to URAA for example, or just remove the link. If there's some disagreement about this among regular editors it should be discussed on the property talk page, and if agreement can't be reached, maybe an RFC. But I don't see what that has to do with this property? ArthurPSmith (talk) 13:33, 24 October 2018 (UTC)[reply]
- "repeatedly explained"? More like "repeatedly asserted". It seems that I am not alone in not finding those assertions convincing. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:20, 22 October 2018 (UTC)[reply]
Oppose I don't think that this site even need to handle redlinks. --125.36.185.213 08:34, 22 May 2019 (UTC)[reply]
Oppose--神樂坂秀吉 (talk) 09:57, 26 May 2019 (UTC)[reply]