museumofmodernart / exhibitions Goto Github PK

Hello.

First, sorry about issuing those pull requests before I absorbed the README properly.

Second, we're wondering if you have plans to add Exhibition IDs, Constituent IDs across the Exhibitions dataset? We're ingesting everything into a new database over here, and wondering how to handle missing things if we ever need to do an update (and there's a new row or two to integrate, but they don't have IDs...).

Hello,

We're looking at all your bits! And, we found a new ID for Andrew Carnduff Ritchie:

Q19766219
https://m.wikidata.org/wiki/Q19766219

Hello,

This dataset relates artists and exhibitions. This allows you to answer these questions:

• Which artists did exhibit their works at the same events at the MoMa?
• With what frequency was their oeuvre exhibited together?
• Which artists have not yet been (or maybe once) exhibited together?

The answer can be visualised in a force-directed graph which is another rendition of an adjacency matrix.

Instead of artists, the same could be done with artworks too:

• Which artworks were exhibit at least once together at an event?
• Which artworks were exhibit many times together?
• Which artworks have never been exhibited together?

So. I noticed that the dataset does not include references to the proper artworks which were on display at the time. I was wondering: is it possible to compile a dataset relating artworks and exhibitions?

I expect this not to be a trivial exercise since data points are far more granular and it requires far more work to assert the presence of all the data points per exhibition. And the resulting dataset will probably be far larger then either yet available datasets.

Generating an adjacency matrix is an intensive task. As a dataset increases linear in size, the associated workload and required CPU resources increase at exponential rate. As such, even if the entire set was available, it would be hardly realistic to include everything. Small datasets (a few dozen artworks & exhibitions) yield optimum results. Maybe a small sample such as the curated highlights of the collection, or - better - a consistent timeframe (ie. artworks produced between 1950 to 1960) could be a viable alternative?

Thanks!