Showcase example

[ajs6f]

As per discussion on this call we are going to assemble a "showcase" example for use on a call with curators and collections managers.

[ajs6f]

@SamiNorling That's a really good point, and I should clarify my position; I don't want to see social relationships modeled in Linked.Art now. I do want to use them (expressed in any convenient form, no matter how unscholarly or inexact) to demonstrate why Linked.Art is interesting and useful. I don't have confidence that we can generate a really useful model without empirical feedback (doing hard examples) or without stealing as much as possible from extant work (reusing ULAN relationships).

[SamiNorling]

ThreeDegreesOfOKeeffe.xlsx

Attached is a spreadsheet with the output of a query to ULAN SPARQL endpoint to grab three degrees of relationships starting with Georgia O'Keeffe. If this is the example we go with, any artist listed here is fair game for any museum partner to generate sample Linked Art JSON-LD representing their artworks.

If we go this route, the "social network" provides the framework for our selection of artworks to model, as well as an easily-understood and (IMO universally interesting) example of linked data, should we decide a general LOD overview is necessary for our audiences before diving into the Linked Art model. Once we have a chunk of artwork data collected (and we could focus in on a specific set of artists if it would help, since the list right now is pretty large), then the main demonstration, along with the model, would focus on any interesting facts/trends/visualizations/etc. we can distill from consuming the data in bulk.

[SamiNorling]

Updated Three Degrees of O'Keeffe spreadsheet: ThreeDegreesOfOKeeffe_v2.xlsx

Clarified column labels re: IDs, and includes URI prefixes: ULAN entities - http://vocab.getty.edu/ulan/ ULAN relationships ontology - http://vocab.getty.edu/ontology#

[azaroth42]

This query results in 59 photographs by Stieglitz of Georgia O'Keeffe that we have in our photography collection. Warning: artistic nudity

These are interesting, model wise, on two fronts -- the artist's relationship to the subject, and the role of O'Keeffe as being depicted, rather than being the artist.

Further, this object depicts O'Keeffe standing in front of one of her watercolors, "Blue I." which was sold by Christies: https://www.christies.com/lotfinder/Lot/georgia-okeeffe-1887-1986-blue-i-5994689-details.aspx

And ... note the orientation of the painting in the background ... it's landscape. Christies has it depicted as portrait, which is then used in pretty much every other copy of the image I can find online. Given that she painted it in 1916, and the photograph was 1917, why would she have it oriented in a way that she wasn't intending it to be seen?

[lizneely]

Hello! This is Liz Neely jumping in from the Georgia O’Keeffe Museum. As some of you know, we’ve been working on a project to deliver data from our collections as linked open data using Linked Art as a target model (learning from the AAC). The project includes data from our fine arts collection, our archives, a collection of the artist’s personal property (and the two historic homes we manage), and her personal libraries. We are doing this work with Design for Context and Charlie is doing the lifting with the data.

I am thrilled with the prospect that GO’K could be a showcase example and would love to support that in any way possible! I also think she is a great example (though I am a bit biased!)

One thing I wanted to bring up is thinking about the role of ULAN as a key glue, but perhaps more in connecting links than in limiting ourselves to what ULAN itself has entered – we’ve been thinking about this, and the 3 degrees of O’Keeffe that Sami pulled illustrates the point of why we need to use ULAN to get beyond ULAN data.

For example, in the 3 degrees of O’Keeffe data using ULAN, Ansel Adams is listed as a 3rd degree connection to O’Keeffe. But, just looking at our collection data (and at the objects in the AAC), Adams photographed O'Keeffe as early as the 1930s and as late as the 1980s, they traveled to Yosemite together, he took installation photos of her exhibitions, they have both been in exhibitions together, there have been books about their friendship. So, the data tell a different story than ULAN does.

As a showcase to a subject expert, just using ULAN and showing Adams as a 3rd degree to O’Keeffe would quickly discredit the value we try to express. Additionally, every artist has important connections that aren’t limited to artists in ULAN (In our data (mind you, as a single artist museum we might be an outlier), we are almost evenly split between ULAN, LCNAF and local sources for authorities. Wikidata may help us deal with some of those local sources.)

Could it be that instead of the connections centering around ULAN, that instead ULAN 'learns' from the data of the linked data collections themselves and from the other authorities, which might tell a more nuanced story? (linked data FTW!)

I’ll be joining your call next week and would love to hear your thoughts. Thanks – and happy Friday!

Liz

Liz Neely Curator of Digital Experience Georgia O’Keeffe Museum Santa Fe, NM

[ajs6f]

@lizneely, thanks for joining in! I agree that relying solely on ULAN would give us a very sparse and unreal picture of the world. I think we gravitated towards it because it is a Getty product and because so many of our sites use it, but indeed to present a compelling example we would need more relationships than it can supply. Whether ULAN can "learn" from our LD work is an interesting question, but I suspect it has as much to do with how the ULAN team intends to manage that system as much as anything we do in Linked.Art. Getty folks, of course, would know much more than I.

Can you tell us a little more about the linked data you are currently producing? What vocabularies (aka join points) are you using for that?

[SamiNorling]

Hi @lizneely, glad to see you jumping in as well!

We investigated ULAN relationships pretty much just as a way to help us set a scope for collecting artwork data (since we started with the scope as ALL OF OUR COLLECTIONS), and also because it could provide an interesting intro to linked data example, if that ends up being needed. That potential intro example becomes even stronger if we can include the linked data that GOKM is publishing. One of the greatest benefits of publishing linked data is that, if you do it right (i.e., using/reconciling with official entities like ULAN records), it ultimately connects local data with much more rich, comprehensive, and potentially more authoritative data on the web (a.k.a., GOKM's linked data about O'Keeffe).

Sami

[zeroexp]

I'm curious - after looking at the AAC this example page on Georgia O'Keefe: http://browse.americanartcollaborative.org/actor/ulan/500018666.html which shows relationships of O'Keefe to Stieglitz as well as teachers and colleagues as well as works pulled in from multiple collections. Is this not something that already showcases the value of linked data in many of the ways described here? In what way is this showcase trying to go beyond something already seen here at this link at AAC?

[zeroexp]

This query results in 59 photographs by Stieglitz of Georgia O'Keeffe that we have in our photography collection. Warning: artistic nudity

These are interesting, model wise, on two fronts -- the artist's relationship to the subject, and the role of O'Keeffe as being depicted, rather than being the artist.

Further, this object depicts O'Keeffe standing in front of one of her watercolors, "Blue I." which was sold by Christies: https://www.christies.com/lotfinder/Lot/georgia-okeeffe-1887-1986-blue-i-5994689-details.aspx

And ... note the orientation of the painting in the background ... it's landscape. Christies has it depicted as portrait, which is then used in pretty much every other copy of the image I can find online. Given that she painted it in 1916, and the photograph was 1917, why would she have it oriented in a way that she wasn't intending it to be seen?

@azaroth42 ---- this is just an aside to this whole conversation because, I think this is a neat question. Wondering if @lizneely noticed this observation and has any guidance on the orientation of this painting as O'Keefe herself directed. Putting on my artist hat, I can think of many reasons why an artist might change the orientation of a painting to suit - in this case - the photograph being made because of the way the shape surrounded her head behind her. As an abstract artist, sometimes I am critically aware of what orientation I think a piece should be in and then other times, I realize that I've painted some of my best work upside down. Also, I have made some pieces that I intended to be displayed in one orientation and then someone else came along to buy it and really insisted that they wanted to hang it in a different orientation. Some artists are more protective of their vision at different times than others for various reasons. I love that you questioned this detail and I'm interested to see if @lizneely has a perspective.

Interesting - I just pulled both of them up side by side in an ImageSnippets folder and rotated the Blue I image, is it possible she was in the middle of painting this watercolor series when this photo was made? There are some slight differences that would almost make you question slightly if it was the same 'blue' as Blue I....I did not look at the dates yet. Just find this quite curious to examine.

[lizneely]

Hello – Thanks for the interesting discussion! I certainly didn’t mean to make it sound like I thought you all hadn’t already seen the limitations of only branching out from one authority! It’s just something that has been on my mind a lot and thinking about how cool it would be for authority links to branch to other authorities (and associated information) and to be data-informed and evolving as a living generative graph.

@ajs6f – We are currently using vocabularies (glue points) on people/organization data and classification/type data. We have place data on our mind, but that is more complex and deserves its own project (connected with our viewshed work with the Land Trust). As you would suspect, people/organization comes up in a variety of categories (maker/creator/designer/manufacturer, depicted, correspondent, owner, exhibitor, exhibition venue, etc.) as does type/classification (type of artwork, genre type in archives).

For these areas, the following are the authorities (in the case of person in terms of priority, i.e. if there is no ULAN entry use LCNAF, if there is no ULAN nor LCNAF use Wikidata.)

People/Organization: 1/ ULAN 2/LCNAF 3/Wikidata 4/Local Source (and we’ll be working on minimizing these through research. For the ones for which we can find data, we’ll create wikidata/wikipedia entries.(Edit-a-thon!))

Type/Classification: AAT (behind the scenes we also use Nomenclature for our historic house collections – but will have to address this more when it goes LOD. Nomenclature is not currently serving our LOD – I convert those things to appropriate AAT terms.) In Archives we use AAT for in the Genreform area to distinguish and allow for query on different types of archives (photographs, correspondence, etc.)

Place (future): TGN

And to the comment of making the authorities work for us -- when we saw that the ULAN bios weren't going to do the trick for us, we (and by we, I mean Charlie!) simply used the ULAN (or LCNAF) to grab the Wikidata intros for bios and link to the full Wikipedia page. (In the cases where Wikidata didn't have ULAN/LCNAF IDs, I just added them (and felt like I was helping the world of knowledge in the process, because I am a dork.))

[lizneely]

@zeroexp As to your note - I had corresponded with Rob separately before seeing your query, sorry about that.

As for orientation, Georgia O’Keeffe correspondence with Steiglitz reveals she preferred the orientation as portrait and not landscape as he had exhibited it at his gallery 291 (and in the photograph that Rob shared). (Source: Catalogue Raisonne (CR)) and I believe (though I didn’t dive deeply) that the exhibition installation photo at 291 is also in the CR (attaching a snapshot of the page.)

These CR conclusions are not all yet input into our data set, but it should be! And then we should link directly to the correspondence (though that could be in our collection or in the Beinecke -- I didn't check and hopefully in the future with linked data it won't matter. :) ) (whoops..sorry, the below photo is upside down..)

[lizneely]

Hi -- This message is meant for Jonathan @ Moma regarding exhibition data (I didn't see your name to tag you) or anyone interested in Exhibition data.

As I mentioned in today's call, we are actively researching and cataloging data for past exhibitions, venues, and O'Keeffe objects in those exhibitions. So far we have about 186 exhibitions, 1917-1948; and then about 50 from recent times (1997-now) and are working through those gaps over the next few months. A lot of those older exhibitions center around NYC and other east coast venues, and I wonder if there is any interesting crossover with your work. I'm attaching a spreadsheet (since that's the easiest thing for me to put together.) The GOKM_ID and Object IDs are local, but I can provide that info if it would be useful-I kept the Object IDs to show the # of objects cataloged to each. (These IDs will also be part of the object URL when we publish, i.e. http://data.okeeffemuseum.org/object/{GOKM Object ID}/)

Let me know if you think this is anything interesting to pursue.

As for international crossover, GOK has had several international shows (and we are planning more), but they tend to be monographic (which a couple exceptions such as the exhibition with Australian female modernists.)

She did travel extensively, and therefore our archives have international ephemera. I'll look at the artist relationships with this lens as it might generate more crossover with collections.

GOK_Exhibitions_In_Progress_2019_5_8.xlsx

[karinanw]

Hello, I'm circling back to this thread about the showcase. Philadelphia Museum of Art has been working on linked data in preparations for the Duchamp Research Portal and may be able to provide some connections to O'Keeffe. Marge Huang is ready and able to share spreadsheets or whatever you like... just let us know what would be most helpful.

Thanks!

[ajs6f]

Great, thanks @karinanw! I think our last discussion settled on using a new Github repo to start collecting this material. @azaroth42, have you had a chance to create that yet? We should probably start coagulating our data.

[SamiNorling]

@karinanw @ajs6f Rob created this repository for gathering example data: https://github.com/linked-art/showcase1.

As for IMA, I'm looking for some interesting connections with O'Keeffe, and have my basic object transformation in a good place where as soon as I know what objects would make a good example, I can bulk transform to JSON-LD. Hopefully I'll be able to add some data early next week!

[ajs6f]

Thanks @SamiNorling! In that case, @karinanw, you could make a Philadelphia Museum of Art folder in the data directory of that repo and Marge could put data there. I'm not sure what form might be best without knowing more about what you have.

[JonathanLill]

@danieltbrennan Dan, I haven't been monitoring this thread so I'm sorry for the delayed response.

The MoMA project you linked to we refer to as the Thomas Walther project since the project grant focused on the donation of his photo collection. Unfortunately the archives wasn't involved in that project and they never published their data in any raw form. But I'm in touch with some of the staff who developed it and am going to see if I can acquire the data. Hopefully that won't be too difficult and I'll be able to share it with linked.art.

Jonathan

Dropping a note here mid-discussion on other possibilities for the artist-network (ULAN) example, I'm reminded that 20th century photography is particularly rich in well-documented figures/relationships - so that might be another area in which to look for examples.

I'm reminded of this (pre-linked data) project:

https://www.moma.org/interactives/objectphoto/#connections?dateBegin=1900&dateEnd=1950

[JonathanLill]

@lizneely I sent you that email a month ago but wasn't monitoring this thread until today so I hadn't seen you post your spreadsheet. It looks great. I thought I'd share here the spreadsheet I sent to you in case it's of interest to anyone else.

In response to Liz, I produced a report of all exhibitions Georgia O'Keeffe was included in at MoMA, attached. (this is based on my own exhibition data, which is currently not absolutely identical to the information on moma.org or on github and itself not absolutely complete or up to date). This does not include data on traveling shows MoMA organized unless they also appeared in New York but some of these probably traveled to other venues.

The first worksheet lists the 63 exhibitions O'Keeffe has appeared in in my dataset, pretty much all MoMA shows. The second worksheet shows all other artists who were exhibited in those shows, over 1300 names, and the number of shows they were in together. There's a lot of permanent collection rotations listed here which boosts the number of total artists.

So that frequency ratio of "# of times one artist was in show with O'Keeffe"/"total number of O'Keeffe shows" can be used as an "exhibited with" kind of proximity measure and used to build visualizations or other things. And it would interest me to see someone look over the entire dataset and see how many times O'Keeffe was exhibited with any other artist compared to their total number of shows and also the frequency they appeared at MoMA compared to the total number of exhibitions in their history or whatever.

Best,

Jonathan

OKeeffeData.xlsx

Hi -- This message is meant for Jonathan @ Moma regarding exhibition data (I didn't see your name to tag you) or anyone interested in Exhibition data.

As I mentioned in today's call, we are actively researching and cataloging data for past exhibitions, venues, and O'Keeffe objects in those exhibitions. So far we have about 186 exhibitions, 1917-1948; and then about 50 from recent times (1997-now) and are working through those gaps over the next few months. A lot of those older exhibitions center around NYC and other east coast venues, and I wonder if there is any interesting crossover with your work. I'm attaching a spreadsheet (since that's the easiest thing for me to put together.) The GOKM_ID and Object IDs are local, but I can provide that info if it would be useful-I kept the Object IDs to show the # of objects cataloged to each. (These IDs will also be part of the object URL when we publish, i.e. http://data.okeeffemuseum.org/object/{GOKM Object ID}/)

Let me know if you think this is anything interesting to pursue.

As for international crossover, GOK has had several international shows (and we are planning more), but they tend to be monographic (which a couple exceptions such as the exhibition with Australian female modernists.)

She did travel extensively, and therefore our archives have international ephemera. I'll look at the artist relationships with this lens as it might generate more crossover with collections.

GOK_Exhibitions_In_Progress_2019_5_8.xlsx

[JonathanLill]

@ajs6f @edgartdata

I'm late to this conversation but wanted to mention that I contacted Tom Scutt at the Paul Mellon Center a month ago about getting the raw data from the Chronicle250 project. He says that the outside vendor they hired reconciled the artist names to ULAN and I suppose otherwise structured and cleaned up the data but he was waiting for them to send him a copy of the final version. All he had at the moment was a rough spreadsheet but promised to send it all along when he received it.

Jonathan

RA could be a good one. Chronicle250 has a handy index of 250 years of its exhibition catalogs: https://chronicle250.com/index/exhibitors Maybe the French Royal Academy of Painting and Sculpture? Here is a partial list of its members: https://en.wikipedia.org/wiki/List_of_members_of_the_Acad%C3%A9mie_Royale_de_Peinture_et_de_Sculpture