The Digital Network for Collections tested a new approach toward the subject indexing of image databases together with the Media Library at HU Berlin’s Institute for Art and Visual History (IKB).

The test was carried out on a collection of art history images from the nineteenth century. Many of the approximately 100,000 objects in this collection had been digitized in recent years and made publicly accessible via the HU Berlin Media Repository.

During the digital cataloging process, the objects that can be seen in the images were assigned a reference data ID from the Wikimedia database (e.g., Neue Wache in Berlin -> Q152341). 

This makes it possible to clearly identify referenced objects and provides access to extensive information about them on Wikidata. The Data Inspector was developed to utilize this information for the substantive indexing of the image collection. The search and filter options do not work with the data in the image database and instead use the information available in Wikidata. For each of the objects found—which include paintings, sculptures, drawings, churches, squares, fountains, and more—the Data Inspector lists the images displaying them.

With one click, users can access the dataset in the IKB’s source database. There, they can retrieve detailed metadata, including information about the respective slide or photo as a physical object, and make use of numerous export and download options.

The application is intended as a laboratory application, making it possible to test the reference-data-based approach toward subject indexing on a significant use case (approximately 50,000 image objects with references to about 18,000 items).

Click here to access the Data Inspector.

We appreciate your feedback!
collections[at]berlin-university-alliance.de

Notes on design and usage

The search function is based on the labels, descriptions, and aliases of the referenced Wikidata items in the following languages:
German, English, French, and Italian.
Search terms are linked with a logical “AND” (a phrase search has not been implemented).