User talk:Multichill/Archives/2022/June

Getty Center

Hi Multichill, can you explain your revert at Getty Center. You say my edit is vandalism - that's not ok for me. Please check the articles and get the information: The J Paul Getty Museum is part of the Getty Center but the Getty Center is not part of the J Paul Getty Museum. The Getty center houses different institutions and the Museum is one of it. Thanks and greets --Rlbberlin (talk) 13:30, 18 June 2022 (UTC)

Sorry, that qualification might have been a bit too harsh. Let's call it counter-productive. You removed instance of (P31) which is the identity of an item. By doing that it becomes unclear what an item is and it causes thousands of constraint violations. You also removed the link between the Getty Center (Q29247) and J. Paul Getty Museum (Q731126) which will also cause problems downstream.

I understand that you want to improve the model. The world isn't black and white, the model is, so trying to fit things in is a bit of a puzzle. It's quite common for a museum to have other things too like a restaurant, a research library and maybe some institute (for example Centre Georges Pompidou (Q178065)). Modeling that is another step of refinement. You need to do careful modeling for that. Just removing some (source) statements is not the way forward. Multichill (talk) 10:19, 19 June 2022 (UTC)

Long time no speak

It must be ten years or more - there seems to be a problem with an IP adding crap Userː105.111.101.102 I notice you have an A against your user name - I have left a comment at admin noticeboard but notice you are more active than the admin board...hope you are able to help ̃ JarrahTree (talk) 13:21, 19 June 2022 (UTC)

That IP also messed up one of the items I created ( Q27032840 ). Blocked and reverted. Multichill (talk) 13:29, 19 June 2022 (UTC)

Thank you JarrahTree (talk) 13:42, 19 June 2022 (UTC)

About inventory numbers

Hi Multichill,

Regarding your revert, are there specific guidelines prohibiting two collections on a single number? I was first a bit disappointed when I saw QS had added to your existing entries but after second thought I found it's quite ok this way, as it is actually the same number which is passed along by the museums to which the painting is assigned. Do you strongly prefer having two distinct statements? You added quite some new items which are all lacking a lot of basic data. I started doing some by hand, but given the scale QS is the only reasonable solution, and I guess we have to do with what it can handle. Otherwise maybe you can have your bot work on tidying all entries as you like them, instead of reverting a single one?

Regards,--Nono314 (talk) 20:30, 23 April 2022 (UTC)

Hey Nono314, I was about to leave you a message, but you beat me to it. I'm working on paintings stolen during WW2. I started with the Linz Collection (warning, big page) also known as Führermuseum (Q475667). I noticed a lot of overlap and connecting inventory numbers for Musées Nationaux Récupération (Q19013512) so I did a basic import for that collection too. I was amazed that it wasn't on Wikidata already. Makes me wonder what else is missing for France.

I'm far from done. I'm just doing the second pass of the Linz Collection to add a lot of missing data and also have to do that for Musées Nationaux Récupération (Q19013512). Still have to see what data is up for grabs.

Thanks for adding more data to these items. I prefer distinct inventory number (P217) for consistency and ran into the same issue with Quickstatements. Ended up having a bot clean it up. I'll do that for these too at some point.

I expect a lot of duplicates. These will become more easy to find after more data (including images) have been added. It almost looks like auction houses and museums are trying to hide the provenance of these paintings. Multichill (talk) 20:49, 23 April 2022 (UTC)

Hello Multichill, yes I figured out later you had actually started on the German side after first stumbling on some of your stub MNR entries. The consensus used to be not having Musées Nationaux Récupération (Q19013512) as collection (P195) but rather the actual museum collection, which was surprisingly absent from your import. Granted, having already everything scraped in local databases, it was not too difficult for me to add them with their ids through a join. I also fixed creators not solved by the bot and added Commons images when available, also quite some location (P276). Still missing: dimensions and mediums ; I might submit a PR to your importer. Do you have some plans to improve past collections with dates to help sort out this kind of things? Note, I only found a couple duplicates among MNRs, ids helped your bot figure out most of them. I see you have now also uploaded images: b&w is not looking too bad, though modern color images would be nice too :)

Indeed, heck a lot is still missing for France! Just 2 numbers: Joconde has around 75,000 paintings online vs. 13,000 on WD, and even for the Louvre their online collections feature almost 10,000 paintings vs. less than 6,500 on WD (up by ~10% through recent MNR additions). I have always felt like you were avoiding France for your imports, and you're probably the largest contributor in that domain, so there is a huge space for growth. Nono314 (talk) 21:04, 25 April 2022 (UTC)

@Nono314: Multiple collections seem to confuse people. I always use The Night Watch (Q219831) as an example of a painting being in multiple collections and of course I had to restore it again. I'm really getting of that one user constantly making a mess of paintings. I wrote a bot in a similar case to add start and end time to each collection. Could probably do the same in this case to get a nice provenance overview and only the current collection(s) as preferred.

Anyway, was still planning to do a second pass on the MNR paintings. Let's see what can still be added. I assume a lot

Regarding the images. Could use a hand with https://w.wiki/56Um . When the painter has been added, the robot will upload it to Commons:Category:Paintings in the Sonderauftrag Linz. The images are not very good, but at least makes visual identification a lot easier to merge duplicates.

Your feeling is correct. I stayed away from France because other people seemed to be working on it. I always assumed that if people started working on a collection, they would also finish it. Any idea why this didn't happen? Would love to get these complete. It's not that I don't like the collections, I really loved visiting Louvre and Orsay again last Summer and will visit some more museums in Paris in a couple of weeks. Maybe complete some of the collections also on Wikidata. For my imports artdatabot is doing all the work. Per collection I have a generator that just grabs the metadata for all the paintings, for example for Linz or Rijksmuseum. Multichill (talk) 21:36, 25 April 2022 (UTC)

Users tend to mess with the data to get the display they want, that's why the risk is higher if a bunch of collections appears without apparent order.

I never meant you didn't like the collections :) But I guessed you were more focused on countries that are more opendata-friendly, i.e northern ones. One notable exception being the municipal museums of Paris that got fully uploaded as a partnership with frwiki after going open. Louvre and Orsay did get mass uploads quite early on, and there was probably less data available online back then. This was huge for the time and I wouldn't blame anyone for not following up. Speaking for myself, I usually only do small to medium batches (a couple hundreds at most) as my workflow involves semi-automated steps, and thus I tend to focus on significant paintings from notable artists. I've been keeping an eye on your git repo for a few years now, but have sticked to QS for my usage. Nono314 (talk) 19:45, 26 April 2022 (UTC)

@Nono314: coming back to the provenance. I proposed a Linz DB ID property to make that part easier to maintain and I wrote some provenance documentation which also includes a part about WW2 and use of preferred ranks. Basically this query should be empty. Here we should add all the data qualifiers and give the current collection(s) a preferred rank. A bit too much work to do by hand so I'll have to dust off one of the bots to do that. I have to think about what date qualifiers I'll add. Multichill (talk) 13:10, 30 April 2022 (UTC)

Please restrain from wrongly insulting me "that one user constantly making a mess of paintings"--Oursana (talk) 21:10, 22 June 2022 (UTC)

P180

[1] Since when, why, and where it was discussed? Wostr (talk) 23:52, 18 June 2022 (UTC)

You assume that this was discussed somewhere. That assumption is incorrect. I often add missing constraints to properties. This is a wiki, I assume you are familiar with Be bold?

The constraint is merely a suggestion (see suggestion constraint (Q62026391)). depicts (P180) is mostly used on art here. Artworks should generally have more than one depicts statement. On Commons depicts is used over 8 million times and should also generally have more than one depicts statement. So to me it seems that this gives a useful suggestion. Multichill (talk) 10:30, 19 June 2022 (UTC)

'Be bold' is an en.wiki guideline and it's really annoying that people refer to the en.wiki rules on independent projects, it happens too often, it does not constitute any argument, and fortunately on some projects it ceased to be an editing guideline. Ad rem: as it may be true for using P180 as a statement, it seems to me it's not while using P180 as a qualifier, IMO this constraint should be limited with constraint scope (P4680)constraint checked on main value (Q46466787). Wostr (talk) 20:51, 19 June 2022 (UTC)

A come on, it's a principle that is translated in many language. It's just that the version at Wikidata:Be bold is very short. Sure, scope added. Multichill (talk) 21:03, 19 June 2022 (UTC)