you’re so not a book

As a “friend of UCAI,” I was recently informed that the project to create a Union Catalog of Art Images won’t continue past its research and development stage to spawn a fully fledged service. For those of you not familiar with this Mellon-funded initiative, UCSD and various partner institutions investigated how to bring copy-cataloging to the slide library – what if visual resources curators had authoritative data available for download and integration into their local catalog, instead of duplicating the work of describing their slides or digital images? The project wound up clustering about 750,000 records from six partner institutions into a database – one of the core issues of the UCAI team consisted in fine-tuning the clustering algorithm so it would “know” which records describe the same art works represented by the images, and unite it with its brethren in a fashion reminiscent of the RLG Union Catalog.

The proposition of creating a joint resource to realize the benefits of copy cataloging seems intuitive, but I think what UCAI has shown is that at least for now, images just aren’t books. The descriptions among the partner institutions varied widely, which turned clustering into a formidable challenge. Seeing the clusters from the test database in the many presentations UCAI staff gave at ARLIS, VRA and other conferences impressively drove home the point that standardized description equals interoperability and all the benefits it entails. In the end, I believe that this area is where the lasting impact of UCAI will be felt: UCAI and its staff were instrumental in galvanizing the visual resources community around further standardization of their practice. They played an active role in the development of Cataloguing Cultural Objects (CCO) (a little more on CCO also here on this blog), and were instrumental in the development of an XML schema for the VRA Core (authored by UCAI staffer Esme Cowles), which will be unveiled on the VRA website soon. Widespread implementation of these emerging standards will eventually result in a descriptive landscape and technological infrastructure which makes efficient data sharing possible in the visual resources community, and that’s when it’ll get really interesting again.