Wikimore

Welcome to the general talk page of Structured Data on Wikimedia Commons. See also specialized talk pages about data modeling and about Lua.

	SpBot archives all sections tagged with {{Section resolved\|1=~~~~}} after 7 days.

Talk pages of subpages and archives

Wikicrowd

I don't know much about Wikicrowd, and don't particularly care to, but is it actively encouraging edits like , or did this user just happen to do that? I suppose the edit is not actively harmful, but to take a picture of two people, at least one of whom has a Wikidata item, and say that what it depicts is a microphone seems really useless. It might as well say it depicts a wall and water bottle. - Jmabel ! talk 18:50, 18 December 2025 (UTC)

(Courtesy ping @Ranjithsiji, who made the edit, and @Addshore, who made the tool.)

The edit summary would seem to suggest that the statement was effectively copied from the file’s membership in Category:Microphones (or more specifically Category:People with microphones). I’m not convinced that depicts statement is significantly worse than the category, which has been on that file since 2021. Lucas Werkmeister (talk) 19:05, 18 December 2025 (UTC)

Depicts says entity visually depicted in an image . In this image there is a Microphone and it is clearly visible. There are humans, May be people who has wikidata item. We can add more than one depicts into an image. So I don't see a problem here. THe microphone is not make prominent here. Ranjithsiji (talk) 19:16, 18 December 2025 (UTC)

I also believe the edit is fine, though of course I'm happy to tweak the tool to guide based on what the community in general wants and finds acceptable. My understanding though mirrors the above comment, which is the microphone is clearly depicted, and thus should likely be included.

Really, the unfortunate thing is that JessAnn Smith and pianist and music historian Eric Hung who are photographed don't have entries on Wikidata, and thus can not also be added as depicted. For the benefit of this conversation, however, I have gone ahead and said that they also depict "person" ·addshore· ^{talk to me!} 12:25, 21 December 2025 (UTC)

And this? - Jmabel ! talk 20:48, 31 December 2025 (UTC)

Again, there probably isn't anything wrong with that image saying depicts trousers, likely it should also have other depicts statements also (which I have gone ahead and added https://commons.wikimedia.org/w/index.php?title=File:No_Kings_NYC_(October_2025)_41.jpg&diff=1140570455&oldid=1139705827)

It all really depends on what people want to be able to look for within the dataset of commons images. ·addshore· ^{talk to me!} 14:25, 2 January 2026 (UTC)

Caption expand button is not usable, as it is barely visible

As shown in the image caption expand button is barely visible while using on Firefox 148.0 browser (image shown) or Google chrome Version 143.0.7499.169 (Official Build) (64-bit) on Ubuntu 22.04 Arjunaraoc (talk) 07:26, 3 March 2026 (UTC)

There are multiple gadgets that intentionally hide or minimize the SDC interface, because many people dislike the whole concept of SDC. Do you happen to use one of them? If yes, https://commons.wikimedia.org/wiki/File:Mallavaram_Dam.jpg?safemode=1 should appear fine for you, as ?safemode=1 disables all gadgets, user scripts and similar. If that URL indeed appears fine, you’ll need to go through your gadgets and find the problematic one. —Tacsipacsi (talk) 19:56, 3 March 2026 (UTC)

Search for capture date

How can I use SDC search to find all images in a category tree, captured at a specific date (e.g. March 16th, independent of year)? I failed with various values for haswbstatement:P571=. (Intersection of commonscat and SDC property)? best --Herzi Pinki (talk) 13:49, 16 March 2026 (UTC)

@Herzi Pinki: I would guess you would do better with deepcat:"6 March, by year" than with anything off of SDC. Neither will be comprehensive, of course, but I don't see a way to tease out day-of-year from date in a search based on structured data. - Jmabel ! talk 18:05, 16 March 2026 (UTC)

Thanks, but to work this needs full categorization by date of all images. My focus was on finding images that where taken on a specific date constellation and do not have the by date categories at the moment of search, for example to set those categories or to find images without such categories. hoped that the automatic propagation of capture dates to SDC would help. --Herzi Pinki (talk) 23:24, 16 March 2026 (UTC)

@Herzi Pinki: It might be doable with a SPARQL query as long as the commonscat isn’t too large… which category are you interested in? Lucas Werkmeister (talk) 17:57, 20 March 2026 (UTC)

@Lucas Werkmeister: I see some eager people hiding images in those by-date-categories. If these images (by date) can be retrieved by a smart search, there would be no further need to hide the images deep down the category tree of these by-date categories. e.g. category:Großglockner.

sparql for regex date & depicts (P180), alas no Großglockner image has a suitable depicts (P180)

is there a way to mix sparql with incategory: search?

btw: dates BC with day precision - unbelievable! best --Herzi Pinki (talk) 19:19, 20 March 2026 (UTC)

@Herzi Pinki: Yes, it’s possible using the MWAPI service. (Note that, while that page has a find category members example, to get deep category members you need to use the search instead.) This query returns one result for 17 March. Lucas Werkmeister (talk) 20:58, 20 March 2026 (UTC)

AFAIK haswbstatement has never supported time values: phab:T238498. --Matěj Suchánek (talk) 09:31, 20 March 2026 (UTC)

deepcategory:"San Marino" insource:/[0-9]{4}-03-16/ https://commons.wikimedia.org/w/index.php?search=deepcategory%3A%22San+Marino%22+insource%3A%2F%5B0-9%5D%7B4%7D-03-16%2F

@Herzi Pinki this is my approach without using sdc, with which i implemented Template:Category search by/month Template:Category search by/hour about finding files according to date/time. it works on the assumption that files contain yyyy-mm-dd hh:mm:ss in their wikitext.

so, this method works for most files, but not for files that dont have such info in wikitext but only in sdc, e.g. files in Category:Uploads using Flickypedia.

sorry if this is offtopic since my method is non-sdc. RoyZuo (talk) 16:23, 15 May 2026 (UTC)

@RoyZuo: , thanks for the hint. The background of my question was: If we can find images via sdc dates and if there are these irresistible bots adding the date/time info into sdc, there would be no more need to manually clutter all the image descriptions with these annoying date and corresponding cross-cutting categories. We could free sparse resources for genuine categorization work (look at the image and recognize what is shown there). Of course your proposal will also help to not add these date categories. best --Herzi Pinki (talk) 17:45, 15 May 2026 (UTC)

@Herzi Pinki yep, part of my motivation was also to make such searches easier "for dummies", so all these Category:Dresden photographs taken on 2024-05-04 xx photos taken on yyyy-mm-dd (Commons:Requests for comment/Categories of photographs by country by date Commons:Categories for discussion/2016/10/Category:September 2007 Finland photographs Commons:Categories for discussion/2017/08/Category:March 2011 France photographs) can all be gone, since some users' argument for using these cats is to "find photos about a certain topic taken on a specific date/period" in an otherwise big category if it's not divided into these subcats.

in reality, that utility can be done using this simple search.

i have plans to make gadgets using graphical user interface, to make doing these searches more intuitive, but only when i find the time... RoyZuo (talk) 19:43, 15 May 2026 (UTC)

date errors / BC / upload / ?bots

Hi, found some date errors and a consecutive misinterpretation by automated processes: See e.g. File:Our Lady of the Sacred Heart Church christmas 001.jpg with a syntactically wrong date, which is misinterpreted by the automated upload process regarding the SDC and is set to a BC date ().

see SDC query for more cases

Can somebody please fix the wrong dates? Experience with also fixing SDC is also needed.
Can someone fix the upload process to check for invalid / questionable dates?

EXIF ↔ SDC mapping

Do we habe any listing of SDC properties equivalent of EXIF (and analog) atributes? I guess I'm talking about every attribute you can extract with API:Imageinfo. —Ismael Olea (talk) 19:29, 23 March 2026 (UTC)

It's here: Commons:Structured data/Modeling/Meta --Schlurcher (talk) 10:35, 24 March 2026 (UTC)

@Schlurcher but there is not equivalency to EXIF/XMP/... :-m Is this information elsewhere? Maybe you have it hardcoded in one of your scripts? —Ismael Olea (talk) 16:35, 24 March 2026 (UTC)

I thought the table included on that page under the Heading "Modeling various types of metadata and exif(-like) data related to Wikimedia Commons files" was exactly what you are looking for. My bot extracts the imageinfo and matches it according ro that table. Otherwise, I am misunderstanding your question. --Schlurcher (talk) 06:44, 25 March 2026 (UTC)

(Side note: the “Data in RDF” table on that page also points out that WikibaseMediaInfo already emits much of this metadata automatically, making your bot edits redundant and, in my opinion, an enormous waste of resources. Queries should generally prefer e.g. schema:height over wdt:P2048, because the former is automatically kept up to date by WikibaseMediaInfo if e.g. a new version of the file is uploaded.) Lucas Werkmeister (talk) 21:21, 26 March 2026 (UTC)

Just a +1 on the side note that I think we should bring up again somehow. What would be a possible way forward to stop this waste or resources? Ainali (talk) 07:44, 27 March 2026 (UTC)

This would be my way to go: Commons:Requests for comment/Technical needs survey/UploadWizardSDC --Schlurcher (talk) 13:46, 29 March 2026 (UTC)

Does the Structured Commons have a future?

When the Structured Commons started, the expectations were that it would replace the system of categories. Now, if I get it right, we are not there, and the WMF stopped developing the Structured Commons. Is the project useful in its current stage? Is it going to survive (I mean not to be closed as having no future)? I am specifically asking because I add to every image I upload two properties, d:P31 and d:P1071 (as well as d:P571, but this is mostly done by bots, sometimes incorrectly). It would be useful for me to know whether I am just wasting my time, but I would be also interested in a broader perspective, for example, an answer to the original question - it is a possible, perspective replacement of categories? Ymblanter (talk) 16:58, 26 March 2026 (UTC)

That there is no development is a political decision of the WMF. This can change any time. We can not do more than demanding things and hope to be heard. We do not know what happens in the near or far future. GPSLeo (talk) 17:35, 26 March 2026 (UTC)

To me, structured data has a value in itself now, and for me, replacing categories could happen or not. Structured data is still great in search, SPARQL queries and working with APIs. So I don't think you are wasting time, you're making Commons better. Thanks! Ainali (talk) 19:34, 26 March 2026 (UTC)

I think it's pretty useful for what it is. I also think it is not on track to replace categorization, probably ever. - Jmabel ! talk 04:34, 27 March 2026 (UTC)

instance of (P31) location of creation (P1071) inception (P571). RoyZuo (talk) 16:03, 15 May 2026 (UTC)

Video

2 years after Commons_talk:Structured_data/Archive_2025#c-RoyZuo-20241119142700-Modeling_video_/_audio_/_pdf_/_djvu_files i wanna ask about videos again. there is still absolutely no info about anything related to videos. last time i asked about languages and time when a video is made.

this time i was prompted by the wish to classify a video according to its type: animation / feature film / podcast / tv episode / tv news package... e.g. File:What is Magna Carta-.webm is animated film (Q202866).

dont know which property to use or if there is actually one for this purpose.--RoyZuo (talk) 15:59, 15 May 2026 (UTC)

i found filming location (P915) and recording location (P8546) that probably should be used.

i'm gonna use filming location (P915) for all my videos if i want to register the location wd item where they were recorded. RoyZuo (talk) 18:55, 25 May 2026 (UTC)

State of neglect

it seems to me sdc is in a state of neglect. there are multiple discussion pages including but not limited to Commons talk:Structured data/Modeling Commons talk:Structured data/Properties table, but all seem to attract very little attention from users and whoever that's supposed to be in charge of maintenance and development of sdc.

supposedly Commons_talk:Structured_data#c-RoyZuo-20260515155900-Video should be raised at Commons talk:Structured data/Modeling but i just post them here since those pages are hardly monitored.

i wonder:

how many people are actively involved in maintenance and development of sdc? paid or volunteer? as of right now, 2026?
in addition to #1, how many people are actively involved in the development and curation of data models?

i ask because if the number is >100, then i can reasonably expect someone should notice questions on these discussion pages soon. but if the number is <10, then i understand why no one seems to be answering. also whether these people are paid or just volunteering makes a big difference, coz it's within my expectation that volunteers might only get to work on something much longer after the issues are raised.--RoyZuo (talk) 15:59, 15 May 2026 (UTC)

i skimmed thru Commons talk:Structured data/Modeling just now. there're in depth discussions that remain unanswered for years.

i think that answers my questions. probably <10 people or no one is in charge of sdc development.

everyone is just following their own intuition to decide which properties to use and how they assign values. RoyZuo (talk) 16:18, 15 June 2026 (UTC)

ISCC

Today, for the first time, I came across an ISCC code (ISCC code (P13150)) that a bot generates. The code sounds interesting, but I’m wondering how it should be managed. Where can I get the code? I don’t expect to have to generate a code somewhere for every photo and set it manually; instead, I’d like to be able to retrieve it from the metadata via an API, for example. But so far, I haven’t been able to find it there. XRay 💬 07:40, 17 May 2026 (UTC)

I've since realized that this isn't the first time I've noticed the code. The question remains: how can I easily access the code so I can set it myself? -- XRay 💬 16:00, 17 May 2026 (UTC)

Hi @XRay. The bot I run is not generating the code, it is just publishing the code that was generated previously. It's also not stored in files, you actually have to generate them with an algorithm. The specs are at https://iscc.codes/ and if you just want to try to generate a few codes, you can do it at https://demo.iscc.io/ The code Wikimedia Sverige is using for this now is available at https://github.com/Wikimedia-Sverige/commonsdb-commons-supplier but that is doing a few more things, namely get more metadata and declare the images in the CommonsDB registry. If you're not in a rush, you can just lean back and we'll come around to your images on Commons eventually (but it might take some time). Ainali (talk) 17:54, 17 May 2026 (UTC)

Thanks for the info. In short, the ISCC codes are stored in a database and updated as needed, and your bot then transfers this data from the database into the structured data and ensures that it remains up to date. -- XRay 💬 18:24, 17 May 2026 (UTC)

That might be a future workflow, yes. For now, the ISCC codes are stored in a database on toolforge, and my bot copies them (using dumps of that database) to the structured data. But we haven't generated the codes for all files yet. When we have, then we can start taking new versions of files into account. Ainali (talk) 21:29, 17 May 2026 (UTC)

Generic wd item labels and descriptions

i raised d:Wikidata:Project_chat#c-RoyZuo-20260525191400-Large_number_of_homonyms. RoyZuo (talk) 19:17, 25 May 2026 (UTC)

Depicts no value

https://commons.wikimedia.org/w/index.php?diff=670348721 how is this possible?? RoyZuo (talk) 20:26, 26 May 2026 (UTC)

Using the three-dot menu behind the input field (if you mean “how can this be achieved” by “how is it possible”). —Tacsipacsi (talk) 00:23, 27 May 2026 (UTC)

thx. that's opposite of wd interface. i was looking to the left of the input field. RoyZuo (talk) 05:58, 27 May 2026 (UTC)

Also: no idea what someone was trying to convey there. @RoyZuo: does it make any more sense to you? - Jmabel ! talk 14:53, 27 May 2026 (UTC)

It doesn’t make any sense to me, I’d say it’s YouTube spam (I didn’t click on the links to avoid increasing the view count, but I can tell it’s quite obvious spam even without that). —Tacsipacsi (talk) 23:45, 27 May 2026 (UTC)

Book (pdf / djvu)

Commons:Structured data/Properties table#Document, article and book files very little instruction.

should title (P1476) be used for title?

two things i notice now:

it is limited to 1500 characters. not sure where that limit is set.
for compilations, how would titles be handled?
some publications are 1 book containing multiple separate works (sometimes even by different authors). how to enter this info in sdc? that this pdf contains book A by author A, book B by author B, and so on.
multiple values of title (P1476) and author, etc.? title (P1476)=author A; author B...?
then how to establish the relation that book A is by author A, book B is by author B...?

--RoyZuo (talk) 11:45, 5 June 2026 (UTC)

@RoyZuo I’d say that does not matter. The publication of a collection is the work. WikiCommons describes the work as uploaded. That’s the title (say: collected works on chairs). There is no real author in that case, there is a publisher and curator who made the collection. The individual books are ‘components’ included into the work. You’d either NOT describe them in commonsdata, or you’d use wikidata’s ‘has part’ to add references to sub parts. —TheDJ (talk • contribs) 08:49, 13 June 2026 (UTC)

thx for reasonable idea.

for reference, File:NDL993064 歌文枢要外三種.pdf is an example.

歌文樞要 is the title of one of the works (usually the main work), but "外三種" ("in addition, 3 works") are included together.

some explanation for this "外*種" practice, commonly seen in east asia: “外二种”是什么意思？ - 扫书萌主的回答 - 知乎 https://www.zhihu.com/question/55935841/answer/146945161 . i'm not sure how common this is in europe or america. RoyZuo (talk) 15:58, 14 June 2026 (UTC)

Museum label

File:(20260517 144022175 HDR) Kulturhistorisches Museum Magdeburg.jpg = museum label (Q6940727) of File:(20260517 144013358) Kulturhistorisches Museum Magdeburg.jpg.

how to efficiently record this relation on a large scale?

initially, i tried to stick to Commons:Overwriting existing files#Secondary images Commons:Overwriting existing files#Secondary images 2 but quickly abandoned it coz it cannot be scaled. when someone has hundreds of exhibits and their labels to upload, this method wastes a lot of time clicking thru the pages and waiting for the pages to load and refresh. RoyZuo (talk) 08:26, 10 June 2026 (UTC)

Have you considered to just add the information on that "Museumsschild" to your description of the file of the actual object? You took your time to write "Kulturhistorisches Museum Magdeburg" which is an actual Non-description" and in my eyes valueless. With the same effort you could have just added the information from this Museumsschild to your description. With just a little more effort you could use the art-photo template as I I have just done here Wuselig (talk) 09:24, 10 June 2026 (UTC)

as i said these complicated manual edits dont scale.
that description is indeed pretty useless, but it was a must due to stupid uploadwizard design. i copied "Kulturhistorisches Museum Magdeburg" from the category name by 1 click coz i have a script that does that. then i pasted that in uploadwizard and copied it to every file i uploaded in that batch. because uploadwizard makes com:file caption mandatory (which i strongly disagree with), i have to fill something in. i always fill it with the common location of the whole batch of photos. then uploadwizard also ticks "same as caption" which copies that to description.
not too long ago it was possible to use MediaWiki:Gadget-GoogleImagesTineye.js for quick and easy ocr. still that was quite a bit of work. but now this gadget is broken indefinitely, it adds more work to the originally relatively simple task of copying the text from the image. i'm unwilling to waste my time on unnecessary manual work that is forced by bad software.
still, trying to ensure that knowledge is not lost and to make editing easy for other people -- in addition to making use of commons' filename by using a consistent format with timestamp so that the related files stay close together in a category -- if there is a way to quickly identify file A is of a certain relation (in this case, being museum label (Q6940727)) to file B, i'm happy to do that.

RoyZuo (talk) 12:12, 10 June 2026 (UTC)

There is no need to upload these images of "Museumsschilds". If it is just short information (data) you should put that text into the information. If it is more text like here, it is actually a copyright violation and in danger of speedy deletion.

I guess your intention was to add the file with description as "another version". But that is not a good alternative to missing description.

Even the Filename you chose is utterly useless and doesn't tell us anything about what your image shows.

Who do you expect to reuse your images? Who is supposed to sift through this spam and find something usefull within our projects, or outside?

Look at the gallery below. I changed the file name and the description of your first file. Do you see how it stands out from the others?

Eckmaskenkapitell (Kulturhistorisches Museum Magdeburg) (20260517 144013358)
(20260517 144048450) Kulturhistorisches Museum Magdeburg
(20260517 144221609) Kulturhistorisches Museum Magdeburg
(20260517 144243434) Kulturhistorisches Museum Magdeburg
(20260517 144303960) Kulturhistorisches Museum Magdeburg
(20260517 144343608) Kulturhistorisches Museum Magdeburg
(20260517 144405000) Kulturhistorisches Museum Magdeburg
(20260517 161235991 HDR) Kulturhistorisches Museum Magdeburg
(20260517 160055524 HDR) Kulturhistorisches Museum Magdeburg
(20260517 154054830 HDR) Kulturhistorisches Museum Magdeburg
(20260517 153710787) Kulturhistorisches Museum Magdeburg
(20260517 153618616) Kulturhistorisches Museum Magdeburg
(20260517 153549610) Kulturhistorisches Museum Magdeburg
(20260517 153358307) Kulturhistorisches Museum Magdeburg
(20260517 153309133) Kulturhistorisches Museum Magdeburg

Look at this gallery. How does your bell stand out from the others?

(20260517 144541341) Kulturhistorisches Museum Magdeburg
Gröninger Glocke
Ratsschulglocke von 1578 01
Ratsschulglocke von 1578 02
Ratsschulglocke von 1725 01
Ratsschulglocke von 1725 02
Ratsschulglocke von 1725 03

I want to laud you for the categorization effort you have put in for File of your bell. It shows that you are not careless and want to put in an effort. So please reconsider what you can do regarding file-names and descriptions.

Yes it is some manual labor to make individual file-names and descriptions. I process my images with Lightroom, which for me is more of an organizing tool with the added features for processing. With the Commons:LrMediaWiki-plug-in tool it is possible to prepare images for upload to Commons. The information-template and the artwork-template are implemented. The images can e labeled, i.e. label "Kulturhistorisches Museum Magdeburg". Put a # in front of a label and it will become a category when you upload it. So you can do all your cataloging beforehand. You can group descriptions and use drag and drop for identical information.

You may be familiar with other up-load tools like Pattypan.

I am currently uploading images I took in museums in India. There is a great community there and they like to make photo excursions. But it saddens me when the result is categories like this: Category:Bronze sculptures at the Thanjavur Palace. This was a huge category with some 600 or so images. All with titles like "Artifacts and sculptures at Art Gallery in Thanjavur Palace" and descriptions like "Artifacts and sculptures at Art Gallery in Thanjavur Palace". In uploading my images I manually searched visually all the individual sculptures and grouped them in separate categories. Up to now, I have identified 82 individual sculptures. That is with the bronze ones. With the Stone ones it has been 32.

It would be so much more fun doing museum photography, if the uploaders wouldn't put this burden onto others. Wuselig (talk) 22:24, 10 June 2026 (UTC)

no extra program/app installed unless absolutely necessary. my devices are all stripped to the minimum. because
1. security risk
2. i dont want to develop a reliance on some niche tools. then when it is no longer maintained my workflow would be disrupted. that has happened to Commons:Commonist for example.
many issues ultimately result from uploadwizard design, for example:
1. it's best to avoid working on description etc. within uploadwizard, because that work is not temporarily saved, so any crash/problem will cause all work to be lost.
2. uploadwizard has small and fixed thumbnails, so users cant see clearly what the image shows. -> not easy to work on description.
3. scrolling to next image is tricky and annoying. there's too much useless empty space in uploadwizard interface.
filename is useless. it is no more useful than accession number (Q1417099): object identifiers used in galleries, libraries, archives, and museums.
it is spam. i mostly just use a crap phone to photo some interesting looking things that quite often happen to have no other photos uploaded earlier than i do.
for curious shutterbugs, each of the following leads to a lot of unnecessary time spent:
1. i often dont know what i photoed.
2. i also dont have enough local knowledge, of history, languages, etc.
as such, it's best to leave enough clues (including but not limited to city, coords, timestamp, museum label) for knowledgeable locals, who can identify what i photoed much faster than i do.
and i can similarly work on southeast asia (hong kong, singapore, etc.) that i'm familiar with.
those filenames are mostly chronological, so despite common's lack of sorting by time taken, people can still easily deduce where i was and what i saw. however, that only works if the same set of files have the identical prefix.
commons is in urgent need of better sorting, including by time up to the precision of a second.

RoyZuo (talk) 15:51, 14 June 2026 (UTC)

Usage of Wikidata entities

I'm working on a gadget that will warn Wikidata admins about SDC usage before items are deleted. I'm hoping to avoid querying item by item. Is there an easy way to extract a list of all Wikidata entities used by SDC?

I could run a set of queries like

   SELECT DISTINCT ?y WHERE {
     ?file wdt:P180 ?y . 
   }

but that's expensive and will only include the set of usage patterns that I explicitly query for. Bovlb (talk) 16:17, 10 June 2026 (UTC)

I’m not sure what you want to achieve. You can replace all three parts of a triple with variables, so you could write SELECT DISTINCT ?object WHERE { ?subject ?predicate ?object } (or with anonymous variables: SELECT DISTINCT ?object WHERE { [] [] ?object }), querying all Wikidata items used in SDC, but that’d mean a huge number of items, and would probably time out. Maybe you want to query whether multiple, but not all, items at a time? VALUES can help: SELECT DISTINCT ?item WHERE { VALUES ?item { wd:Q42 wd:Q5 … } [] [] ?item. } (But actually, why not item by item? Deletion also happens item by item.) —Tacsipacsi (talk) 18:53, 10 June 2026 (UTC)

It's not clear to me that all objects are necessarily Wikidata items, which is why I didn't go straight for dumping all objects. For example, many objects are statements.

Certainly I could do batch queries when I'm working with a group of items, but that complicates the workflow on my side.

I'm thinking about downloading the dump and just scanning for "wd:", but downloading 27GB daily for a few MB of information is less than ideal.

Unfortunately, deletion does not always happen item by item. For some reason, Wikidata is continually faced with the situation that a user creates dozens or even hundreds of new items that need to be bulk deleted. I'm trying to streamline the process of finding items that ought to be rescued. Bovlb (talk) 19:11, 10 June 2026 (UTC)

@Bovlb: You might also want to have a look at phab:T334352 where some work in a similar direction is happening. Lydia Pintscher (WMDE) (talk) 11:27, 12 June 2026 (UTC)