Enriching image collections with Iconclass: can it be automated? A Datalab case study

At the end of 2023 the Badisches Landesmuseum (BLM) and Allard Pierson Amsterdam (AP) launched the Datalab - “a temporary research and development space for practice-based research, improving collection data and experimenting with machine learning techniques”.

In a first case study we started a concordance between the German keywords used for subject access by the BLM and Iconclass concepts. The aim is to enhance the interoperability of the BLM dataset and datasets which use Iconclass for their subject access.

The related, much broader question that was raised, was whether this type of concordance-building - linking an existing set of keywords to Iconclass concepts - can be automated, or at least accelerated by the use of Artificial Intelligence.

Although it is far too soon to offer a definitive answr to this question, some conclusions can be drawn from the practical experience gained in the experiment.
Hence it seemed useful to include the resulting case study in the Iconclass Bibliography. It is also posted on the website of the Datalab