Wikidata talk:WikiProject sum of all paintings/Archive/2015

From Wikidata
Jump to navigation Jump to search
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

Paintings

Painting types now are mostly from Commons, which are "oil on copper", "oil on panel" and "oil on canvas". Now and then there is some confusion about gouache, fresco, and watercolors. Since we tend to view everything as a painting behind our screens, I think all of these should be paintings. Jane023 (talk) 15:08, 29 December 2014 (UTC)

Do not forget tempera and acrylic paint. Concerning gouache and watercolours, if these are painting, should also graphics (like linocut) be paintings?--Ymblanter (talk) 16:41, 30 December 2014 (UTC)
Hi there. Linocuts are not paintings. However, acrylic, tempera, gouache, watercolors and fresco are paintings. Missvain (talk) 17:09, 30 December 2014 (UTC)
Here is a tip my old art history professor taught me - when you think of paintings, think of brushes being used to paint on something. Linocuts, screen printing - while they all use paint, it's not "painting" :) Missvain (talk) 17:10, 30 December 2014 (UTC)
+1 for your old professor! Multichill (talk) 21:06, 30 December 2014 (UTC)
Pastel drawings, are they paintings? Some of them are done only with pastels and fingers, some use brushes also.--Giftzwerg 88 (talk) 09:11, 29 March 2015 (UTC)

Create pages for missing countries

Hi everyone, I started Spain based on en:List of most visited art museums in the world. We should probably do that for other countries too. Who feels like helping? Multichill (talk) 11:29, 1 January 2015 (UTC)

Focus on all paintings without collection

We currently have quite a few paintings without collection. The number is a bit off because autolist is acting up. Sarah and I think it's a good idea to focus on reducing this list to zero. This way when we're don people can focus on one collection, get it to a certain standard and than move on to another collection. @Spinster: @Jane023: want to help? vele handen maken licht werk! If we work with several people on this we can quickly reduce the numbers. If you encounter an item that has collection (P195) and keeps showing up in autolist: Do some sort of edit to make it show up in Special:RecentChanges. That will trigger an update. Multichill (talk) 16:22, 1 January 2015 (UTC)

Also, remember that while most works have a collection, there might be a few that don't, and that's ok :) So the list might not be emptied...but we'll get close! Missvain (talk) 00:48, 2 January 2015 (UTC)
Thanks for the help Missvain. We just dropped below 1000 paintings. We started with over 3000 paintings so the number seems to be heading in the right direction. A lot of paintings were in a private collection some time of their life so I expect the number of paintings without collection to be very low if not zero. Multichill (talk) 17:25, 2 January 2015 (UTC)
@Coyau: @Shonagon‎: @Ymblanter‎: are you also helping out? We're approaching the 800 so now we're getting to the fun part! Multichill (talk) 13:08, 3 January 2015 (UTC)
I am around but not doing more than 10 per day. What do we do if the source says "private collection" but does not specify which one?--Ymblanter (talk) 13:15, 3 January 2015 (UTC)
If we all do 10 each day we're don in two weeks! Just add collection (P195) -> private collection (Q768717). We can always add a qualifier if we dig up more information. Maybe you can focus on the ones that have a link to ruwiki? Multichill (talk) 14:16, 3 January 2015 (UTC)
We seem to have dropped below 800.--Ymblanter (talk) 14:52, 6 January 2015 (UTC)
Made some statistics at Wikidata:WikiProject sum of all paintings/Top collections to get an idea of the progress. Still very rough! Multichill (talk) 22:13, 6 January 2015 (UTC)

Statistics brainstorm

Hi everyone, what kind of should we collect? Be creative. @Husky: I think you had some ideas about this. Multichill (talk) 17:09, 3 January 2015 (UTC)

For portraits, all person items in a depicts field should have an image in their wikidata item. Which leads me to my next question...Jane023 (talk) 10:52, 15 January 2015 (UTC)
In terms of visualized statistics I'd like to see
  • a simple graph that shows our progress (monthly data points: how many paintings are there on Wikidata?)
  • a bar graph that shows by decade how many paintings have images on Wikidata, from the start? This will show the 20th Century copyright gap really nicely, since we will have very little images starting from the 20th Century and almost none since 1944. Spinster (talk) 17:11, 29 May 2015 (UTC)

How to model details of paintings

example detail of group portrait

I have illustrated Wikipedia articles with details of group portraits and it follows that these should be used as images for the items of those people. How do you specify these details? Should I use "part of" for this? Thx, Jane023 (talk) 10:52, 15 January 2015 (UTC)

@Jane023: I don't really undesrtand the issue. What do you need to specify ? Data about the file itself (that it is a detail of a larger painting, etc. should rather be stored on Commons). --Zolo (talk) 07:41, 29 March 2015 (UTC)
@Zolo: When you link an image to a person's item, it follows that you specify what the image is in qualifiers. In this case, it is a portrait painting by Frans Hals, but only a detail, not a whole painting. It would be nice to be able to filter these out of a list of paintings by Frans Hals linked on Wikidata. Jane023 (talk) 07:58, 29 March 2015 (UTC)
I don't really see what sort of list you are trying to get. It could only get you a list of files depicting a painting by Frans Hals used in Wikidata, not a list of paintings by Frans Hals. Given that the same work can be depicted in several files, this can always be different. Ideally, statements should be enabled directly on Commons files so that a machine could automatically check that the image is a file by Frans Hals. Sadly it is unclear whether it will ever happen, but even so, I doubt it is worth the trouble to describe files directly in Wikidata.

Qualifiers

I stumbled upon several painting items and it seems there is a inconsistency about whether to use collection (P195) with inventory number (P217) as the qualifier, or the other way around. Can someone please shed some light on this? Thanks. —Wylve (talk) 05:40, 28 March 2015 (UTC)

There is no general rule for this. The third solution is to have both as separate properties without qualifiers. For now it is only important to add both properties, no matter which way round. Everything else can be done esch way we like it by bot at a later stage. The solution with qualifiers is necessary when a piece of art is divided in parts that are part of different collections or if one piece has several numbers, e.g. when collections are merged.--Giftzwerg 88 (talk) 14:54, 28 March 2015 (UTC)
Please always add collection (P195) as qualifier to inventory number (P217) claims so we know in what collection the inventory number is valid. Portrait of Margaretha Cornelia van de Poll (Q17323941)  View with Reasonator View with SQID is a good example (plenty where that came from). Multichill (talk) 15:01, 28 March 2015 (UTC)
@Wylve: collection (P195) can be useful as a qualifier of the accession number (but the scheme may require some refinements. I think the inventory number systems are common to all French state-run museums, whiile some are specific to a single institution). At the same time, and even if that sounds a bit redundant, adding collection (P195) as a standalone property is also useful. For instance, it enables to qualify it with start time (P580) / end time (P582). --Zolo (talk) 07:38, 29 March 2015 (UTC)

Destroyed or missing paintings

Nativity with St. Francis and St. Lawrence (Q1433194) has handled this problem with collection (P195) + Perdido (Q7167677). But I think it is better to take the last collection and use qualifiers p582. We need some more qualifiers to describe a certain fate of the painting:

  • Painting destroyed by fire, war, vandalism, catastrophic events. Painting for sure does not exist any longer
  • Painting missing. Painting probably still exists, but it is unknow where it is. It was accidentaly dislocated or looted during war times or for unknown reason can not be found at it´s usual place.
  • Painting stolen.
  • Painting has a certain location but it is unknown or disputed who the legal owner is

--Giftzwerg 88 (talk) 15:42, 28 March 2015 (UTC)

@Giftzwerg 88: Ha, that was a bot import. I updated the item, what do you think? significant event (P793) seems to be quite useful to document all these things. Not sure if we need more qualifiers for this. Multichill (talk) 15:57, 28 March 2015 (UTC)
Your solution is not that clear. It seems the painting has moved to an "unknown" collection. But it is as far as we know not part of any collection at all, so it needs "no value" for collection. I think P793 is usefull.--Giftzwerg 88 (talk) 16:16, 28 March 2015 (UTC)
It can be part of collection (maybe some drug lord's private collection (Q768717)?), we just don't know it. With "no value" we say that we're sure that's not part of a collection. Multichill (talk) 16:19, 28 March 2015 (UTC)
So both solutions are not accurate. One day WD will be the database for all lost and found paintings, the ultimate weapon to hunt down thieves, dealers of illegal antiquities and pieces of art and also the buyers in all countries, just because you can find every stolen painting with a simple query on WD along with a precise description the legal owner and a picture! We just need to have a solution that is precise and easy to handle.--Giftzwerg 88 (talk) 17:55, 28 March 2015 (UTC)

Project scope

Seeing the topics discussed here, it seems that this project has basically taken over Wikidata:WikiProject Visual Arts. I think the two projects should be merged, or their respective scope clarified. --Zolo (talk) 07:44, 29 March 2015 (UTC)

The scope of this project is deliberately narrow. It only focuses on paintings. The Visual Arts project focuses on all sorts of art like statues, public art, installations, etc etc. The ontology are probably be better suited over at the visual arts project. I wouldn't mind at all if some moderation is applied and some topics (like the previous one) moved over there, as long as pointers are left here. Multichill (talk) 10:24, 29 March 2015 (UTC)
I'm not sure if a separation between paintings and other genres of visual arts is that useful. It appears to be quite artificial partly causing double work. --Marsupium (talk) 10:08, 3 April 2015 (UTC)
Feel free to propose the termination of the WikiProject Visual Arts if you feel that you have to do that, we're definitely not going to expand the scope of this project. Multichill (talk) 12:57, 3 April 2015 (UTC)
At least on the front side an expansion is taken into account. Why do you oppose to it? --Marsupium (talk) 01:34, 25 May 2015 (UTC), 01:55, 25 May 2015 (UTC)

Sum of all painting details

Hi everybody, especially @Multichill, Shonagon, Dschwen, Husky: @Jane023:. So great to see so many of you in the last couple of days. I was in the Mauritshuis this afternoon, blown away by the beauty and the detail in so many of the paintings, and it struck me -- could we make a visual search, like Shonagon's Krotos or Hay's Tools, but for details in paintings? Perhaps with the Lyon hackathon as target date to get a demo done?

So that one could, say, search for hats or carts or dogs, and the search would return images of those things for all the relevant paintings we have in Wikidata with images on Commons -- but show just the image of the detail thing -- ie just the dog, or just the hat, or just the cart -- with the user able to refine the search eg by time or painter or place or to a particular type of dog. I think this would be awesome, and something that we could be in a unique position to pull off, since it's Wiki where the images from all different collections get pulled together.

A couple of key things could make this possible:

  • First, if we could set up an IIIF-standard image service running on tools for master images on Commons that are a image (P18) for one of the sum of all paintings. (See International Image Interoperability Framework and Phabricator T89552 for more about IIIF.) This could presumably be very similar to the zoom endpoint already created by Dschwen. The advantage of IIIF is it has a very sane URL syntax for specifying a sub-part of an image.
  • Second needed would be a good tagging gadget, to be able to specify a region of an image (a simple rectangle would be enough), and then identify it with a thing that had a Q-number on Wikidata. (Or edit an exisitng visual tag -- eg adjust the boundaries, or change or make more specific the Q-number identification. This could then be included as a depicts (P180) in the Wikidata entry for the painting, together with a string to specify the IIIF geometry of the sub-part of the image.
  • Then with an appropriate WDQ search, it should be possible to pick out not just say all the Jan Steen images that include a dog, but to visually browse what those dogs look like, and use that visual search to jump straight to a painting.

What does anybody think? Jheald (talk) 16:15, 13 April 2015 (UTC)

It sounds like a good idea, in fact you sound thrilled by this idea, but we need to do one step after another. First of all we need to do the basic properties for a million paintings. We have not even got a complete list of collections. The thing you want to do needs to be done on Commons, and we need images of the artwork first. I somehow also have in mind a detailed description of every artwork as an unstructured text (or better even for all images in all languages on commons), this would be usable for the visualy impaired as an alternate text for creenreaders this also is usefull for search engines. I think the day is not so far away when we can use computers to process the kind of search you want to do without the help of thousands of editors and generating clickable images.--Giftzwerg 88 (talk) 17:45, 13 April 2015 (UTC)
Yes of course all of that is important. But it is no reason not to get at least a tech demo of this in place as well. The more different things people can do, and the more impressive visual applications we can show off (even for however small a subset of a million paintings), then the faster we will attract a community and the faster everything else will get done as well. If I'm right that we may be technically in a place already very close to being able to achieve this, I think it could be a stunning way to be able to get people involved, to create together a resource that no other platform can currently offer. Jheald (talk) 18:42, 13 April 2015 (UTC)
This sounds like a pretty neat idea indeed. Actually we already have the visual annotation tool on Commons, so maybe we could just reuse that, or even import the annotations from there to Commons. I know Jane has done a tremendous amount of annotations on Commons already. I'm not quite sure how to model the depicts (P180) property. Maybe using a qualifier that uses a to-be added IIIF property? For the visual look, i think we could refer to Rijksmuseum's Rijksstudio interface, which is basically this idea. Husky (talk) 19:13, 13 April 2015 (UTC)
Yes this would be a good idea. I think we might even be able to import some of the crowd-sourced tags that are on the BBC Your Paintings website, if we can develop a proper mapping of tags to items that can be added to the "depicts" property. See the tagger explanation here Jane023 (talk) 14:34, 14 April 2015 (UTC)
Yes, a great idea. Just for information there is a W3C recommandation about media fragments, including images, which is quite simple. How to implement an annotation tool, as I am not a developer, I really do not know. Shonagon (talk) 00:50, 15 April 2015 (UTC)

Proper way to indicate 'current country' of a painting

What do you all think is the proper way to model the current country the painting is located in? One obvious way is to use country (P17), for example, Mona Lisa (Q12418) ==> country (P17) -> France (Q142). However, this might be a bit ambiguous because that property might be confused with country of origin (P495).

One other candidate would be located in the administrative territorial entity (P131), but AFAIK that property can't be used with countries.

If we have consensus over this it would be trivially easy to do queries like "give me all paintings in the Netherlands, and we might do some batch work to add the property using the collection and location information. Husky (talk) 12:43, 16 April 2015 (UTC)

Nice idea, but we also need a 'country depicted' for landscapes, because people might confuse the two, like I just did. Jane023 (talk) 13:45, 16 April 2015 (UTC)
What's wrong with location (P276) -> Salle des États (Q10375063). Multichill (talk) 16:30, 16 April 2015 (UTC)
Because that doesn't give you the country? Of course, you could traverse the 'tree' up from those other items to get the country, but that seems pretty cumbersome. Husky (talk) 18:31, 16 April 2015 (UTC)
@Husky: Do you think this is cumbersome? I don't think you're current botrun is a very good idea. You shouldn't have run that without proper consensus. Multichill (talk) 20:48, 18 April 2015 (UTC)
Right, i've never seen a nested query like that before. Thanks for showing that. I still think the property should be added to the individual paintings as well. country (P17) does say "state of this item". Husky (talk) 09:39, 19 April 2015 (UTC)
From an information modelling point of view, I would also vastly prefer NOT to have the country, but the most detailed and specific location instead... Spinster (talk) 17:12, 19 April 2015 (UTC)
Why not have both? Husky (talk) 17:21, 19 April 2015 (UTC)
Isn't the point of structured data is that it can be hierarchical? We don't need to put a "[located in] country" statement because we have a statement for the institution and the institution has a location property where we should put the city and the city has a location field where we would put the country. Isn't it a duplication of effort to write the same overarching property "country" into every level beneath it - city, institution, painting? Wittylama (talk) 11:54, 9 June 2015 (UTC)
As country (P17) does not have any clear meaning, we may not really know what to do for Mona Lisa, the US Embassy in London or the Battle of Waterloo. Does it mean we should break it into smaller, better defined properties ? I don't think so. The point of this property is to allow some large-scale data-sorting, plausibility checks, etc. From a logical standpoint, it should never provide any information that could not be somehow inferred from other statements. Considering that, I think things should be kept as simple possible, and when we don't know what to do, we just do not need to use any "country" property. --Zolo (talk) 17:40, 19 April 2015 (UTC)

Photograph collections?

I've been thinking today about the Canadian Copyright Collection which WMUK and the British Library had digitised - 3000+ photographs all now on Commons (mostly in master tiff + cropped jpg versions).

We have fairly reliable metadata for them, including Commons creators in many cases, in a form I could easily transform for upload and link across to Commons. Given it's a finite and well-documented set, would this work for a sum-of-all-paintings approach on Wikidata, or would there be notability issues? Andrew Gray (talk) 20:45, 16 April 2015 (UTC)

Yes, we would have notability issues. Please take a look at commons:Commons:Structured data. That's more suitable for this. Multichill (talk) 08:06, 17 April 2015 (UTC)

authority control

Some time ago I imported items for all the paintings in the Rijksmuseum (Q190804). For the last months I've been busy connecting the paintings to the right painters and during that proces I decided to take in on step further: I wanted every painter in the Rijksmuseum to have a link to RKDartists ID (P650) or Union List of Artist Names ID (P245) (and preferably both). To keep track of progress I created User:Multichill/Rijksmuseum creators RKD. After I linked some more painters I run a bot that updates the page. This turns out to be very useful. I found a lot of duplicate items and incorrect links. The fact that items about painters here on Wikidata are linked to RKD and ULAN also makes this data much more useful for (re)user and make it much more usable as linked open data (Q18692990). I would propose we aim, as part of this project, to get all our painters (that is, painters used as creator (P170) on a painting) linked to some form of authority control. This is a good way to increase the quality of the data and is actually already the case for most our paintings with creators (about 26.000 right now), but still quite a few to do (900 paintings, so I guess about 700 painters).

In the future we can even take this further by linking concepts like painting (Q3305213) to controlled vocabulaires like Art & Architecture Thesaurus ID (P1014), locations to GeoNames ID (P1566) or Getty Thesaurus of Geographic Names ID (P1667) and even individual paintings to RKDimages ID (P350) or Cultural Objects Names Authority ID (P1669). But let's stick to the painters for now. What do you think? @Spinster: and @Magnus Manske: I promised to ping you on this. Multichill (talk) 21:17, 18 April 2015 (UTC)

I totally approve! Painters are good because we have so many of them on Wikipedia and on Commons. I am also slowly realizing that our problem with sculptors and other artists is industry-wide and not just a problem of Wikipedia projects. Apparently paintings (being movable) have historically been more easy to attribute than e.g. sculptures, which tend to be large and unwieldy. One caveat though, as I have also been working on this: not all painters are in both databases, and oddly the ULAN even has painters born in NL that RKD misses, and RKD has painters born in the US that ULAN misses. I really think we need to work with them somehow, too. Jane023 (talk) 09:12, 19 April 2015 (UTC)
I also meant to suggest that we propogate the names used by RKD for all Benelux painters to all EU languages (as I think this will help prevent doubles from being created). Jane023 (talk) 09:15, 19 April 2015 (UTC)
I think this makes sense (although I'd like to cover all notable artists while we're working on this anyway - not just the painters - otherwise we'll have to perform the same trick twice or more). On a side note: I think it would help so much if we would also have all name variations of artists on Wikidata - that would make them much more findable and would avoid many cases of duplication that we have now. Any way to achieve this (apart from adding these one by one manually, which I now do every now and then and which is very cumbersome)? RKDartists has many name variations, so does ULAN, and many are mentioned all over Wikipedia articles too. Spinster (talk) 17:17, 19 April 2015 (UTC)
The database of painters is a necessity and a result of the attempt to create the sum of all paintings. To straighten out duplicate items for painters that are not in the authority control databases, we need date of birth and place of birth if possible (also of place & date of death if available). If an artist has many aliases, I think it is useful in many cases to use birth name (P1477) to help identify the artist, as this name won´t change in different languages.--Giftzwerg 88 (talk) 20:01, 19 April 2015 (UTC)
BTW to do an effective search for aliases, it would be nice to have a search engine that searches all Wikimedia projects at once and shows the results ordered by projects.--Giftzwerg 88 (talk) 20:06, 19 April 2015 (UTC)
I'm happy to see you all agree. I'll update the project page at some point. I'm focusing on painters used in paintings to avoid scope creep. Of course this practice applies to other painters and artists too, but that's out of scope for this particular project.
Both ULAN and RKDartists seem to be opening up so we might have two good sources to import aliases from. Multichill (talk) 17:22, 21 April 2015 (UTC)
...and as an active uploader of painting files on Commons of obscure painters (and commons category creators) not in current collections or image donations, I am one of the people expanding your backlog :) Jane023 (talk) 08:54, 23 April 2015 (UTC)

New property (qualifier) proposed - 'after a work by'

While editing some paintings here, I noticed I missed a qualifier for an artwork's creator - after a work by. I've submitted a property proposal for it and would be interested to get some feedback there. More explanation in the property proposal itself. Thanks! Spinster (talk) 14:25, 21 April 2015 (UTC)

4.3 Describe paintings on Wikidata

As far as I can see, it is enough to identify all paintings just by inventory number (P217) and collection (P195). But inventory number (P217) is not mentioned in this section.--Giftzwerg 88 (talk) 20:33, 23 April 2015 (UTC)

Van Gogh Museum (Q224124)

I just finished adding all the paintings in the Van Gogh Museum (Q224124) to Wikidata. I went the extra mile also uploaded all the pictures to Commons. We're getting quite a nice collection of Van Gogh paintings. Maybe someone feels like getting his paintings complete? It's a bit of a puzzle because many Wikipedia articles are about several similar paintings, not about the individual ones. You can use part of the series (P179) and has part(s) (P527) to connect these. The article en:List of works by Vincent van Gogh gives a nice overview. On Almond Blossom (Q1432536) you can see how to connect the F and the JH numbers. Multichill (talk) 21:36, 29 April 2015 (UTC)

Thanks so much for this valuable work! I see Carole Hensen has been working on the enwiki list of works for years. Great to see this chunk of his work now available on Commons and Wikidata. I will try to look into the dataset and see what I can do. Jane023 (talk) 08:29, 30 April 2015 (UTC)

Frick Collection (Q682827)

A lot of the paintings were already available here. Cleaned it up and added the remaining ones. @Jane023:: feel like helping? Could use a hand with adding the missing painters. Multichill (talk) 22:21, 4 May 2015 (UTC)

Great work! So nice to have all of these. ~I am working on adding the creators. Thx. Jane023 (talk) 08:52, 5 May 2015 (UTC)
✓ Done Jane023 (talk) 09:55, 5 May 2015 (UTC)
Great Jane! Thanks for the help. I'm currently importing Brooklyn Museum (Q632682) based on the paintings we have on Commons. I still have to refine the code a bit so creators are added and backlinks are made from the images back to Wikidata.
Maybe you feel like helping with Metropolitan Museum of Art (Q160236) and Museum of Modern Art (Q188740)? Quite a few paintings without inventory number to figure out in the MET and MOMA. Could use a hand there. Multichill (talk) 12:08, 5 May 2015 (UTC)
Nice, you can also spot missing data through Wikidata:Frick Collection created by user:Jura1. --Zolo (talk) 11:11, 6 May 2015 (UTC)

How to model the "form factor" of the painting?

Hi, I came across several paintings which are painted on paper scrolls as well as silk scrolls. Is there a property that I can use to indicate that they are painted on scrolls and not just ordinary paper and silk? —Wylve (talk) 17:35, 18 May 2015 (UTC)

Use instance of (P31)scroll painting (Q19969434). --Marsupium (talk) 02:13, 25 May 2015 (UTC)

New subproject

Hi all, I have been working on Frans Hals paintings and have taken the model I built to another painter, Pieter de Hooch. It's so nice to see that most of his paintings were on Commons already in some form! Anyway, to explain more about how I am doing this with external open data I started a subproject here: Wikidata:WikiProject sum of all paintings/Hofstede de Groot. Theoretically we can use all the data and text (from the German or English) to create painting articles on Wikipedia, so this could be a great way to get people to contribute to all three projects: Wikidata, Commons, and their language version of Wikipedia. My model (which is all in Excel) converts the painting's data to a form both suitable for making a Wikidata item as well as creating a list of files that can be displayed in a Commons gallery or Wikipedia list of paintings. It would be nice if there could be a way to make tools to automate some of this (double and triple) work. Jane023 (talk) 14:00, 30 May 2015 (UTC)

Why this project?

Hi everyone,

I am working on a blog about this project, to bring more awareness and to provide an easy explanation on why we are doing it, how people can help, and what we want to see happen with it. I'm trying to articulate why this project exists and how to explain it to people (simply saying "to get the metadata and do cool things" is not sufficient). What are our expected goals with this project? If we can come up with a few things, that would be great. Please share your thoughts! Missvain (talk) 01:34, 8 June 2015 (UTC)

I'd like to "+1" this question! I'm often trying to come up with a good way of describing both the eventual value, and also the practical (easy entry-point) activities that this project offers. But I find it difficult... Europeana is very interested in supporting the project but I can't quite work out a way for it to do that that is of practical benefit [to both groups]. Equally, I'd sure like to cross-promote Missvain's blogpost on Europeana if possible. Wittylama (talk) 11:48, 9 June 2015 (UTC)
Some time ago, we got started on https://www.wikidata.org/wiki/Wikidata:WikiProject_sum_of_all_paintings/Benefits_for_museums ... feel free to edit/add. Spinster (talk) 19:07, 9 June 2015 (UTC)
that's a good start with some ideas on it, thanks! Missvain (talk) 18:37, 14 June 2015 (UTC)
User:Multichill - any thoughts? Missvain (talk) 18:37, 14 June 2015 (UTC)
The Wikimedia mission is to collect the sum of all knowledge. We try to contribute to this by bringing together the sum of all paintings so we have a central entry point if we want to know something about or related to a paintings. No such thing exists at the moment so we're creating it. Multichill (talk) 19:44, 14 June 2015 (UTC)

Updated scope

Hi everyone, I updated the scope. It's not an actual scope change, just a change of wording. The main scope is now "The WikiProject sum of all paintings is a WikiProject to get an item for every notable painting". And the criteria for this is either of these:

  • The painting is in a notable collection of a museum, library, archive (i.e. a painting in the Metropolitan Museum of Art)
  • The painting is made by a notable artist (i.e. Van Gogh painting in a private collection)

I hope you agree. Feel free to improve on the wording. I'm wondering if we need a third line for edge cases with paintings of anonymous painters which are published about in reliable sources. Probably to small of an edge case to bother. Multichill (talk) 15:01, 4 July 2015 (UTC)

Yes fine, there are tons of paintings now labelled "anonymous" that have been attributed to big names in the past. Not every artist has his or her own "Rembrandt Research Project" to forensically analyze these. I think any panel painting from before 1700 is well worth including, and notable 19th-century forgeries are probably also worth including, but edge cases, like you say. Jane023 (talk) 15:29, 4 July 2015 (UTC)

Google Art Project

I noticed some people working on Google Art Project paintings. To assist in this process I created two tracking categories on Commons:

Might take some time for these categories to fully fill up. @Coyau:, @Shonagon:, @Poulpy: this might be useful for you. Hope this helps. Multichill (talk) 15:23, 4 July 2015 (UTC)

Duplicate Paintings

Per request by Multichill, I created a list of duplicated painting items. I put the list on Wikidata:WikiProject sum of all paintings/Duplicate paintings. --Pasleim (talk) 22:38, 23 July 2015 (UTC)

Wonderful, thanks Pasleim! There are still lots more dupes out there that don't have their inventory numbers set (and some don't even have the collection right because the collection name has changed since the original WP article was created). But this is a great list to have and thankfully not too terribly long. Jane023 (talk) 06:27, 24 July 2015 (UTC)
@Pasleim: awesome! Thanks for making this. I see some interesting puzzles in the list. Is it possible to schedule it to update say, once a week? Multichill (talk) 22:01, 24 July 2015 (UTC)
@Pasleim: I see you scheduled a bot to update the list. Great! I'll see if I can reduce this list a bit.
I was wondering if you could do the same logic for catalog code (P528) and catalog (P972)? Maybe limit to items in the tree of work of art (Q838948) if it contains too much out of scope items. Multichill (talk) 15:31, 1 August 2015 (UTC)
@Pasleim: went through the whole list. Should be much shorter now. I see quite a few lines like:
Maybe you can filter these out in the future? Multichill (talk) 20:01, 1 August 2015 (UTC)
fixed it--Pasleim (talk) 19:23, 4 August 2015 (UTC)

How to deal with series?

Yes, I noticed that Joconde uses "Musée de Luxembourg" numbers for paintings that are not in that collection which is weird, and I merged two that both referred to the same series. I think we need to discuss the proper way to deal with series (create the parent item with "has part" (1), "has part" (2), etc, and for each series member "is part of" parent item. The only time you can avoid this is when there are two pendants and the one is a "pendant of" the other. Jane023 (talk) 07:58, 25 July 2015 (UTC)

@Jane023: The approach I've been using that seems to work:
Is this something we can agree on and can document it somewhere? Multichill (talk) 15:31, 1 August 2015 (UTC)
Hmm interesting approach. But I think we need to define a series as also a set. Sometimes we don't know the order or the original installation (especially common in altarpieces that have been split up). Yes I think this needs to go in Item structure here. Jane023 (talk) 15:51, 1 August 2015 (UTC)
We're not doing mathematics here so a slight slide of definition (set (Q36161) vs sequence (Q133250)) shouldn't be a big problem in my opinion. Multichill (talk) 20:01, 1 August 2015 (UTC)

100.000 paintings on Wikidata

WikiProject sum of all paintings has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. We reached the milestone of 100.000 paintings on Wikidata. Thanks for all the help and keep up the good work! Let's see how long it takes to reach 200.000 and if we can continue improving the quality along the way. Multichill (talk) 15:50, 13 August 2015 (UTC)

Yay! Jane023 (talk) 10:49, 14 August 2015 (UTC)

Error reports to museums

Do we have any place to report data issue to the source institution ? --Zolo (talk) 06:34, 21 August 2015 (UTC)

What is a Painting?

The BM collection http://collection.britishmuseum.org/sparql has almost 600k objects of type "visual representation" with query: 

select * {?x crm:P2_has_type/skos:broader/rdfs:label "visual representation"}

This is too broad: includes photographs, medals, plaquettes, mosaics.

WikiProject sum of all paintings has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.: Which narrower terms of "visual representation" are paintings? There are 68 

select * {?x skos:broader/rdfs:label "visual representation"; skos:prefLabel ?label}

--Vladimir Alexiev (talk) 15:07, 26 August 2015 (UTC)

The answer is that it depends who you talk to. We are nt restricted to oil paintings, though personally almost everything I do is an oil painting. We also accept gouaches, watercolours, pastels, and pen paintings. If you dig into museum collections you will notice that they aren't in agreement with each other either. --Jane023 (talk) 05:25, 27 August 2015 (UTC)
I've put the BMT concepts on google sheets and made tentateive judgement y/n. Please comment there, or if you want to edit directly request edit access. --Vladimir Alexiev (talk) 10:59, 27 August 2015 (UTC)
What Jane said. Paint on a surface, that's what we need. We're not very picky about what the paint is made of and what the surface is. This page gives a nice overview.
Just add the type the museum is using and make sure it's somewhere in the tree. Multichill (talk) 20:12, 30 August 2015 (UTC)

Multichill's tool

@Multichill: made a tool that matches painters across Mix-n-Match authority lists and to paintings, and allows you to create a WD item for the painter with basic data: http://tools.wmflabs.org/multichill/painters/. This uses Quick Statements to create them. It took me a couple minutes to create 3: Sidney Goodman (Q20873014), Seymour Rosofsky (Q20873006), Charles Lewis Fussell (Q20873005).

As with any other tool, the responsibility about data accuracy is yours.

Eg I got a row "Elizabeth Osborne (7 paintings)":

  • RKD artists (P650): 61040 - Amerikaans; schilder
  • ULAN (P245): 500012687 - American painter, born 1936
  • BMT (P1711): 75942 - This is wrong. If you look in BM data, that's just someone mentioned on a ring inscription "ELIZ.OSBORNE OB 19 MAY 1737 AE:71"
  • People Australia (P1315): 585047 -

Some enhancements could be useful:

  • Add checkboxes which of the authority IDs should be created (in the case above, I'd uncheck "BMT")
  • Attach the paintings to the newly created painter

--Vladimir Alexiev (talk) 15:07, 26 August 2015 (UTC)

I agree. No matter how you shake the dice, some handwork is necessary. Maybe such a checkbox is a good idea, but I think adding them by hand as you go through the item is easier. We can't automate everything, because the names are tricky, as you point out. The main thing to avoid is creating a double entry, so a thorough search of commons and other wikiverse projects beforehand is also necessary. --Jane023 (talk) 05:28, 27 August 2015 (UTC)
In the above case, I want to create the item, with RKD, ULAN, Australia IDs, and only skip the BMT id. I can remove the respective Quick Statement, but many people wouldn't know it.
"Thorough search of commons": wait a sec: if there's a WD item, the tool will find it and not offer me to create a WD item, right? Why do I need to search in Commons? --Vladimir Alexiev (talk) 07:53, 27 August 2015 (UTC)
No, commons data is only on Wikidata because a Wikidatan put it there. Only Wikipedia pages have Wikidata items. There was never any systematic interwiki links on Comons, it was always done by hand. Jane023 (talk) 15:46, 27 August 2015 (UTC)
@Vladimir Alexiev: you get taken to quick statements so you can edit the statements to add. I could have just added "doit" to the url and quick statements would have started at the moment you clicked the link So if the BMT is incorrect, remove it in quick statements. I'm not going to implement check boxes. A bot runs almost daily to add the creator to the paintings, that covers the second point.
@Jane023: We imported all creators and only had leftovers in Commons:Category:Creator templates without Wikidata link. Looks like there are still some useful painters in there. I asked Amir to import them. Multichill (talk) 20:26, 30 August 2015 (UTC)

Nice to have creator pages when they exist, but there are lots of painter categories without creator pages, and there are also lots of paintings not catagorized, but out there in various creator spellings. Searching usually picks something up for important artists. I often surpised what is there and what is not there. --Jane023 (talk) 21:51, 30 August 2015 (UTC)

Subclasses of painting

based on WD:Bistro#Terminologie dans la peinture. Most paintings have instance of (P31) painting (Q3305213). However, that covers fairly different things, in particular, it includes both works on a mobile support and murals or frescoes. fr:Tableau (beaux-arts) (and Spanish cuadro, Dutch Schilderij ?) on the other hand exclude murals. So maybe it would make sense to use painting (Q20900710) as a P31 value instead ? Obviously it doesn't preclude from using more precise terms. --Zolo (talk) 13:24, 4 September 2015 (UTC)

To me all this seems quite simple. In my eyes painting (Q20900710) could simply be merged back into painting (Q3305213) and its subclasses like panel painting (Q55439) and mural (Q219423) should preferably used because they are more specific. This will lack a possibility to specify the movability of the object. Though, movability is relative and I do not think that it is necessary or desirable to indicate it. In many cases murals were transferred to canvas or otherwise moved and there are also paintings on panel or canvas permanently fixed to walls. (That's why I do not like painting (Q20900710)subclass of (P279)painting (Q3305213).) However, it is useful to specify the materials, for that we have several subclasses of painting (Q3305213) though. Not necessary to change anything I think. Cheers, --Marsupium (talk) 00:32, 5 September 2015 (UTC)
On a more structural level, the difference may be that a "tableau" is a readily identifiable object (Mona Lisa is wood panel and the paint of it), while "painting" can be more fuzzy (when we say that the Sistine Chapel ceiling is a painting, what exactly does it include beside the paint itself). --Zolo (talk) 05:40, 5 September 2015 (UTC)
It is of course better to use more precise subclasses of painting (Q3305213). I just do not see a need to use "tableau" as a intermediately precise class. --Marsupium (talk) 20:47, 6 September 2015 (UTC)

45 pairs of painters with matching days of birth and death

Courtesy of the new SPARQL query service:

http:// tinyurl.com/pssbzgj

-- hit 'execute' to run the query. (You might also need to hit return in the browser URL bar, to make the query appear).

Having just sorted out a similar block of matched items that had nationality:UK, the easiest way to deal with them seems to be to paste the date of birth into the URL for Reasonator's date search, eg

https://tools.wmflabs.org/reasonator/?date=1593-05-19

and then find the two items with the matching death dates and open up Reasonator pages to look at them.

Note that some of the matches will be pairs of items that genuinely ought to be merged; but some may be the result of a different painter having given incorrect dates due to name confusion (or sometimes even copied to a painter with a completely different name). So I do think it's well worth looking at the two merge candidates in Reasonator -- just because the names and dates appear to match, it doesn't necessarily mean they should be merged, it might be that the right dates have been put on the wrong item.

I'm a bit worn out having done the UK matches, so if anyone would like to have a go at these, that would be great.

(I did try to wrap some more around the query to look up the item numbers and names, but for some reason that seriously slowed it up. (See account at Wikidata:Project_chat#Identical_data_sets if anyone wants a look, or can fix it). With this first beta-release of the service, it seems very easy to run into the 30-second execution time limit for queries. But it can do some quite impressive things!) Jheald (talk) 20:49, 9 September 2015 (UTC)

See also Wikidata:WikiProject Data cleaning/matched birth and death days for matches to investigate in other groups of humans. Jheald (talk) 21:37, 10 September 2015 (UTC)
✓ Done -- 3 pairs remaining (see page linked above): one pair needing merging on sv-wiki, one pair to merge on es-wiki; and one pair of apparently different people (though their birth & death dates not referenced, and not verified). Jheald (talk) 21:30, 16 September 2015 (UTC)

Size of a painting

A painting frame

Finally, units are here! Now it's time the address something I've been thinking for quite some time. We have height (P2048) and width (P2049), but how do we deal with frames? Paintings without frames are easy. Here it's just the size of the painting. But if we have a painting with a frame and height (P2048), does that include the frame or not? I propose we create 4 items to serve as qualifiers:

  • height (P2048)
    • inner height, the height of the painting excluding the frame
    • outer height, the height of the painting including the frame
  • width (P2049)
    • inner width, the width of the painting excluding the frame
    • outer width, the width of the painting including the frame

Do you like this approach? Any suggestions for better labels? Multichill (talk) 18:47, 11 September 2015 (UTC)

Yes this is good. Displaying this data in a meaningful way with Listeria for example is still a ways away, but this is something. --Jane023 (talk) 18:54, 11 September 2015 (UTC)
Actually, I just realized something. The dimensions without qualifiers is the painting without a frame. If a frame gets added the inner width and height are smaller than the actual paintings because there might be a bit of overlap. Also frames might change over time so we need to qualify it with a date. Changing of frames is something that is pointed out in the Rijksmuseum collection manual. We can also record the weight of a painting. Multichill (talk) 19:06, 11 September 2015 (UTC)

Some project top-lists

Most common properties (full list) for items with instance of (P31) painting (Q3305213) (based on 50,000 instances with lowest Q-numbers):

Most common values for some properties, from SPARQL queries (over all paintings):

Enjoy! Jheald (talk) 17:49, 13 September 2015 (UTC)

Interesting! I like the way you think. Thanks for posting --Jane023 (talk) 12:17, 14 September 2015 (UTC)

And now some queries to show which classes of items the properties most often connect (no longer now limited to paintings -- though there are so many painting-items now, they do tend to dominate). In each case, follow the link then hit "Execute" to run the query:

Jheald (talk) 14:13, 16 September 2015 (UTC)

Deprecated statements

Qualifier reason for deprecated rank (P2241) has just been created, allowing a reason to be supplied if a statement is deprecated.

The idea is that reasons would need to come from a controlled vocabulary -- ie to be items that would belong to particular class of allowed reasons.

It's all happened a bit sooner than I expected, without much discussion of what sort of reasons ought to be on the list. (And how routinely we should expect them to be applied). So it would be good to think, for the paintings project, what are the sort of things we might want to note or highlight as reasons that a particular statement should not be trusted ?

For reference, there are currently 101 statements about paintings that are deprecated,

summary by property: tinyurl.com/ovzf3fv and detailed list: tinyurl.com/nokz46a

and 73 about painters

summary by property: tinyurl.com/psbzjbp and detailed list: tinyurl.com/nteewoy

Are these deprecations self-explanatory, or would it be useful to add reasons? (eg, because not all applications may spot the quite subtle deprecated rank). And if so, is there a shortlist we can make of acceptable reasons? Jheald (talk) 17:21, 16 October 2015 (UTC)

One that is fairly common (and which I used yesterday) is deprecation of the inventory number due to deaccessioning (from sales, theft, war loot, whatever). Artworks pass from collection to collection and the order of accession number may be random so that deprecation of all older ones may be necessary. I try to use start and finish dates to help but to avoid confusion deprecation works best. --Jane023 (talk) 08:56, 17 October 2015 (UTC)
@Jane023: I think we should use "rank = preferred" for the current inventory number and "normal" for former IDs. As clients can filter out data so that they only keep those with the best rank, it should not cause major issues. In Wikidata context, "deprecated" actually means something like "an important source said that, but it is wrong and has always been". If we use "deprecated" for "no longer true", then it is difficult to distinguish it from "was always wrong". --Zolo (talk) 08:37, 28 October 2015 (UTC)
Yes but that can be difficult for use in lists when I just want to get a list of all paintings in a collection. For theft and destruction I suppose it's OK to have these show up, but when a painting has been sold to another museum it shouldn't show up in a list of paintings from the former owner. Or maybe it should? I think it's confusing for the reader without some highlight to indicate it is no longer in that collection. For the same reason I haven't been sure what to do with long-term loans. What do you think about this? --Jane023 (talk) 08:50, 28 October 2015 (UTC)
When using the #property parser, only the best ranked values show up. For Listeria generated list, maybe user:Magnus Manske can tweak things so that it does the same. --Zolo (talk) 12:37, 28 October 2015 (UTC)
That should already be the case? --Magnus Manske (talk) 15:36, 28 October 2015 (UTC)

I created deaccessioned object (Q21441777), and also withdrawn identifier value (Q21441764) for eg identifier values that have been withdrawn and now give dead links.

Possibly another qualifier beyond reason for deprecated rank (P2241) is needed to apply these, if statements are not actually deprecated; but I hope they may usefully help document things all the same.

Collection / applies to part

I have just run a query which finds that we have over 4000 collection (P195) statements, where applies to part (P518) is being used to indicate something that is either an art museum (Q207694) or an art collection (Q7328910), the holdings of which are considered part of a larger collection.

For example, the statements on Q18602432  View with Reasonator View with SQID and Sigismund (Q18579295)  View with Reasonator View with SQID.

It seems to me that this is a misuse of applies to part (P518), which is usually used to indicate that the statament applies to part of the subject of the statement, rather than part of its object. (E.g. a normal use of the applies to part (P518) qualifier might be on, for example, a creator (P170) statement that applies to the frame of a picture: part of the subject of the statement (the picture), not the statement object (the maker)).

Instead, if we have more precise information about the object of a statement, the normal approach would be to use the more precise item rather than the general item, and then to make sure that the precise collection was part of (P361) the general collection, and that the general collection has part(s) (P527) the more specific collection.

Another alternative is to consider there to be one collection, but different works in it are located in different places. The Tate in the UK considers its collection in this way, so that a work might have collection (P195) Tate (Q430682), but location (P276) Tate St Ives (Q2577210).

Either of these two approaches seems IMO okay, but use of applies to part (P518) for this seems to me definitely not right. Jheald (talk) 13:28, 27 October 2015 (UTC)

I agree I think, but maybe you can link to corrected statements to be clear? I may have created things like what you are description (but I think I would have used "part of" rather than "applies to part"). --Jane023 (talk) 10:16, 28 October 2015 (UTC)
Hi @Jane023: I've looked into this some more, and the problem is entirely with paintings that are part of the collection and the sub-collections of the National Museum, Sweden -- date uploaded as part of a bot run by L_PBot last year.
This query tinyurl.com/pdmqcep summarises how qualifier applies to part (P518) is being used on collection (P195) statements.
The qualifier is being used entirely appropriately on 1472 Polyptych (Q3907484)  View with Reasonator View with SQID and Melun Diptych (Q2589776)  View with Reasonator View with SQID.
The remaining incidences are all on items in the collections of the Nationalmuseum (Q842858)  View with Reasonator View with SQID, where it is being used to indicate a sub-collection. It looks like we need to contact User:Lokal Profil / User:André Costa (WMSE) to see what he thinks would be the best way to take this forward. Jheald (talk) 15:33, 28 October 2015 (UTC)
My bad. I'll change them over so that the current qualifier replaces the object and make sure all of these are part of (P361):Nationalmuseum (Q842858). I'll just have to hunt down any implicit assumptions in my code about which value collection (P195) is supposed to have and any wdq links I might have sent them. /André Costa (WMSE) (talk) 18:59, 28 October 2015 (UTC)
Bot is now replacing the applies to part (P518) in the Nationalmuseum objects. As a bonus there is also a patch to make these things easier in the future. /André Costa (WMSE) (talk) 20:30, 3 November 2015 (UTC)

I used "applies to part" on a creator statement here: Family portrait in a landscape (Q21280514). Is this correct usage? --Jane023 (talk) 09:25, 30 October 2015 (UTC)

Hi Jane. That looks exactly right, and very useful. Jheald (talk) 12:29, 30 October 2015 (UTC)

IIIF

With luck in future it may even be possible to add coordinates as qualifiers to the values of the "depicts" statement, so that then a user could jump straight to a zoom of the relevant part of the image. (If we can talk somebody into plumbing in an IIIF endpoint for Commons, to allow easy access to parts of images). Jheald (talk) 12:29, 30 October 2015 (UTC)
Sounds like a good idea, but only if you could get it working with the "add a note" option on Commons. I was pretty upset that the image viewer strips all of the notes as I had been a big fan of that option (especially on maps, but also many paintings). --Jane023 (talk) 16:32, 30 October 2015 (UTC)

IIIF meeting in Ghent, Tuesday 8 December

By the way, there is a one-day international meeting to present IIIF taking place in Ghent on Tuesday 8 October, http://iiif.io/event/2015/ghent.html, where they are going to present what it does and how it works, and how IIIF services are making it possible for people to work in an integrated way with content being brought together from multiple repositories.

I don't know how many places are left, but I do think it could be very useful if somebody could go along, to get an idea of what it's all about, and what it makes possible, (and why making Commons accessible through a working IIIF endpoint, plus listing the content available on each Wikipage media-viewer-style as a IIIF manifest, would be a good thing to provide.)

Pinging @Jane023, Multichill, Spinster:, and maybe you know the best way to get the word out to the wider NL and BE Glam-Wiki communities. Jheald (talk) 12:57, 25 November 2015 (UTC)

IIIF moving forward

Tracked in Phabricator
Task T89552

Thanks to the efforts of Dschwen (talk • contribs • logs), we're now very close to having Commons images accessible through a usable IIIF endpoint, that can be used eg to serve up specified details of images.

So from eg: File:Jean-Baptiste_Perronneau_-_Madmoiselle_Huquier_-_WGA17215.jpg one can specify 

http://tools.wmflabs.org/zoomviewer/proxy.php?iiif=Jean-Baptiste_Perronneau_-_Madmoiselle_Huquier_-_WGA17215.jpg/pct:4,80,28,20/full/0/default.jpg

or from File:Godward_Idleness_1900.jpg the detail 

https://tools.wmflabs.org/zoomviewer/proxy.php?iiif=Godward_Idleness_1900.jpg/pct:65,81,35,15/full/0/default.jpg

The full IIIF image API syntax is given at 

  http://iiif.io/api/image/2.0/

URLs have the general form 

  {scheme}://{server}{/prefix}/{identifier}/{region}/{size}/{rotation}/{quality}.{format}

So as well as specifying the crop with the {region} part of the URL being eg pct:x,y,h,w specifying x (from left), y (down from top), height, h and width w, of the crop region (as a percentage of the whole image), the syntax also allows the crop to be

  • reduced or blown up to a particular horizontal or vertical size;
  • rotated by 90, 180 or 270°, and/or flipped;
  • and rendered in colour or grayscale.

One can also ask for the particular options supported by the server with a call like 

https://tools.wmflabs.org/zoomviewer/proxy.php?iiif=Godward_Idleness_1900.jpg/info.json

All in all it's a nice syntax for fundamental operations, that more complex software can then build on top of.

One snagging point at the moment is that the IIIF calls can only be used on images which someone has first viewed through the Commons zoom-viewer, eg
https://tools.wmflabs.org/zoomviewer/?f=Godward_Idleness_1900.jpg

But with luck User:Dschwen should be able to rectify that soon, so that then it will make sense to propose a new property "relative position within image" to qualify image details with position values such as "pct:65,81,35,15".

All now fixed! Jheald (talk) 21:18, 15 December 2015 (UTC)

Relative postion within image

Wikimania 2016

Only this week left for comments: Wikidata:Wikimania 2016 (Thank you for translating this message). --Tobias1984 (talk) 11:45, 25 November 2015 (UTC)

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata_talk:WikiProject_sum_of_all_paintings/Archive/2015&oldid=2113173658"