Talk:Community Wishlist/Wishes/Suggest media set in Wikidata items for their Wikipedia articles

This page is for discussions related to the Community Wishlist/Wishes/Suggest media set in Wikidata items for their Wikipedia articles page.   Please remember to: Sign posts using the four tildes (~~~~) Remain civil and polite during discussions. Place new text under old text (start a new post).

Options of technical implementation

There are probably more than the 5 options listed here. I think option 4 would be best. The bot could post to the talk pages across all the language versions and if nobody declined the request (then in a subcat of Category:Declined suggestions of Wikidata media additions) e.g. by replying with anything that contains phrase "Request=declined" and the media is still set on the Wikidata item and hasn't been declined in some threshold of language versions (e.g. 5) then the bot could add it to some location of the article after a specified time such as 2 months. An issue here could be the image location and caption of the file(s) – however captions could also be added after the file has been added if no caption with the language version has been set on Wikidata and the Commons description. The most difficult thing would be the correct location of the file within the article, please comment if you have any idea. Maybe it could use some AI to determine the best location via natural language processing the texts and detecting the most appropriate place for the file. If the file is used in any other language version it could also get info on where and find the corresponding section/part of the article to also add it there. Prototyperspective (talk) 18:25, 20 October 2024 (UTC)

Challenge: videos missing lang qualifer

A key challenge here is that many uses of audio and video files in Wikidata items do not have the language of work or name (P407) qualifier set.

This could be solved via this (requested) SPARQL query (by M2Ys4U) that shows files in a given category that are used in video (P10) statements on Wikidata that don't have a language of work or name (P407) qualifier on them.

One can go through the results of queries like the above and set the qualifier and there should probably be some script that automatically sets the language qualifier if the video is in any subcategory of e.g. c:Category:Videos in German. It could probably be implemented already via adjusting your query somehow and QuickStatements. One could also make it a constraint that the qualifier is set where "language-neutral" would be one possible value.

A subissue here is that some videos have subtitles and sometimes the video would be good to add (to suggest to add) despite the audio being in another language if it has good-quality subtitles for the target language. One could think about this later, I think for the initial version only videos where the audio language matches either the language of the Wikipedia or matches a language in the set of languages are widely understood in the WP's regions (like English in German-language regions) would be suggested.

Likewise, audios including spoken Wikipedia audios have the same problem but I think for these it's far less common that the language qualifier is missing. Is there already some kind of script or bot that auto-adds spoken Wikipedia audios to the External links section of Wikipedia articles? Rarely there also are images (mainly maps & datagraphics) that need to have a language qualifier set but don't have it. Another way to solve this challenge to what's proposed here is that contributors declining a media addition suggestion could be shown a reason field or directly the lang qualifier to enter the info that the media languages mismatches (and/or specify the qualifier directly) so the same suggestion is not shown for other Wikipedias if either only x (eg 2) number of contributors declined the media suggestion or a lang qualifier was set by them. Prototyperspective (talk) 14:01, 7 November 2024 (UTC)

(Also @M2Ys4U:) For setting the qualifiers semi-automatically I'm using this method:

1. Copy-pasting directly from this adjusted query (select top left with mouse and scroll to bottom) to a local text file (I don't see any QuickStatements-suited export option or SPARQL-QuickStatements integration there)
2. Open the local text file with Visual Studio Code (or VSCodium) and use the ctrl+f +expand search and replace function to replace all "wd:" with "" (remove these 3 chars)
3. Append P407 Q188 (there's two tabs in between) after every line also with the search and replace function by turning on its regex mode entering ^(.+)$ in the first field and $1" P407 Q188 (again there's two tabs in there) in the second field to replace all
4. Replace File: with P10 " (one tab in there)
5. Save to a txt file and then enter a few of these into QuickStatements via "Import V1 commands" for testing. If it works well, enter all of the lines (including those run for testing). Here I will leave it out 2 files so the two queries will still show some results (ultimately/later it should not return any results).

Notes:

• it can only add qualifiers. I could not find any info on how to make it add-or-replace qualifier values instead of add-or-do-nothing so maybe that's not possible at all. It would be good to have a query show Wikidata items with seemingly contradicting qualifier statements; e.g. languages with different audio languages in the language qualifier.
• As always with tasks like this, such tutorials can help others as well as allowing me to quickly revisit or resume this. I modified the query for this this adjusted SPARQL query (note that replacing wd: in the item column didn't work) but it only made things harder to search & replace to make it QS-ready. If you know any ways to improve on (e.g. make redundant) any of these 5 steps (also if only relevant if this method was to be used for all videos in lang categories), please comment. A related issue is that videos should probably always at least have one caption – a caption in the language of the file even if that is just the file title or the Wikidata infobox (on WMC & WP) should show the file title if no caption is set.

Prototyperspective (talk) 15:35, 7 November 2024 (UTC)

However, videos set on Wikidata items where the language does not match should probably also be suggested in some way, e.g. if an option for that which is disabled by default is enabled. This is useful for two reasons: 1. the videos may have subtitles or are illustrative despite of another language 2. people may use that to find files to redub (likely mainly using a method like this) so that they add a newly dubbed video to the Wikidata item and to the Wikipedia article. Prototyperspective (talk) 17:53, 7 November 2024 (UTC)

Closing wish

Hey @Prototyperspective we plan on marking this wish as "delivered" given that the Growth team plans to cover some of the requests through their Structured Tasks work (talk page). If you think there's a more appropriate status, please let me know! JWheeler-WMF (talk) 15:13, 13 November 2024 (UTC)

That's amazing to hear! I think proposals should only get the status "delivered" set if they have indeed been implemented, i.e. reached publicly available alpha or beta testing stage, and that it would be best if there was a status for 'planned and/or under development' (probably some other issues would also get get that status set instead of setting them directly to 'delivered' before they actually have been and before knowing for certain whether or not they ultimately will be). Prototyperspective (talk) 15:30, 13 November 2024 (UTC)

Image suggestions is currently live on Arabic, Bengali, Czech, Greek, Spanish, Persian, French, Indonesian, Polish, Portuguese, Romanian, Turkish, Vietnamese and Chinese wikipedias. I think it's fair to count that as "delivered", though maybe it also counts as "under development" seeing as work to roll it out to more wikipedias is ongoing? I don't think it's worth quibbling about too much in any case - we're expecting the rollout to all wikipedias to be complete by the end of March 2025 CParle (WMF) (talk) 14:30, 21 November 2024 (UTC)

Oh and also it's available to all users with 50+ edits on the Android and iOS apps for all languages CParle (WMF) (talk) 14:32, 21 November 2024 (UTC)

Thanks for explaining. There are two problems here: 1. they do not suggest videos or files set in e.g. schematic but only files in the image WD property 2. since Including media files set on Wikidata item in Wikipedia articles across languages by default has been archived, the part about automating this more is not in any wish anymore and not implemented here. I think option 4 (if option 5 isn't done) in the "Technical ways to implement this" would be best or needed considering the many small Wikipedias with too few contributors / people going through image suggestions. Note that usually there are just 1–3 media files per item at most, it would also be useful to have these in even short articles. So I could make two new separate wishes about also including other media files, not just those in the image property, and about automating this process more (probably via a bot adding them if the suggestion has not been declined) or this wish simply gets back the Open or Under development status. Prototyperspective (talk) 15:18, 21 November 2024 (UTC)

Number 1 is not really accurate. We suggest images for articles based on

• the wikidata image property
• the wikidata commons category property
• whether an image is used on the page with the same wikidata id on a wiki in another language
• values of the wikidata depicts property in the image's structured data

However you are correct in saying that schematic might be another good source of suggestions - looks like there are ~7k wikidata items with the property. I added a phab ticket, but please note that we already have a great many suggestions (there are 97k articles on ptwiki with suggestions, and 111k on frwiki) CParle (WMF) (talk) 13:07, 25 November 2024 (UTC)

Thanks for clarifying and for creating that issue.

1. However, the video (d:Property:P10) set on the Wikidata items is not suggested. For a good example see d:Q827617#P10. In that case the video matching the WP language would be suggested. For other language WPs the English-language one could be suggested. For items with only one video set without a lang qualifier or with a mismatching one that could still be suggested as these often have subtitles or are useful despite of the mismatching language (e.g. like the video in en:Persepolis before I replaced it with an English redub version).
2. I think there are several further props to suggest media from like audio which is e.g. useful for articles about music. One could postpone that for now.
3. but please note that we already have a great many suggestions (there are 97k articles on ptwiki with suggestions, and 111k on frwiki) Exactly this is the point of the other archived proposal and option 4 and 5 in this proposal and why I think this proposal should be reopened since the other one has been archived: it's entirely infeasible that all of the articles will get their suggestions checked/added across all the many Wikipedias – it needs some kind of more automated process and this is key. A bot suggesting the media on the talk page and adding it to the article if nobody declines the proposal within 2 months would probably be best if no MediaWiki change like option 5 is added.

Prototyperspective (talk) 13:26, 25 November 2024 (UTC)

Ok, so given that we have image suggestions pretty well covered already perhaps what we need is a new wish specifically to expand suggestions to audio and video. That one definitely couldn't be considered "delivered". What would you think of that?

On automatically adding suggestions - I've added a comment in the other ticket CParle (WMF) (talk) 15:30, 27 November 2024 (UTC)

Well I think this proposal already starts off with Many items on Wikidata have a video, audio, or image set while the corresponding Wikipedia articles don't so that other media types are also included is kind of a core part of this proposal. For the other issue and automatic addition, I'll take this to the ViewIt page at some time and then consider creating a separate wish about a bot suggesting the media and auto-adding these if there is no negative response or developments based on the ViewIt tool. Prototyperspective (talk) 18:44, 27 November 2024 (UTC)

Indeed ... I just think it's a bit confusing that suggestions for images are already in production, and so (a significant) part of this wish is already delivered. Rather than debating the ins-and-outs of whether that means this wish is delivered or not, I think it'd be simpler to have a separate wish for video/audio that's definitely NOT deliivered already

(cool on the ViewIt thing) CParle (WMF) (talk) 13:04, 28 November 2024 (UTC)

Lists of Wikidata items missing images

Such queries and lists are relevant to this wish. When people set sufficiently high-quality and representative images on Wikidata items using such lists or in general, that should trickle down into the often many Wikipedia articles for these items which have no or only much less useful media. d:Category:Wikidata items missing images. Prototyperspective (talk) 09:57, 9 March 2025 (UTC)