Addressing the pervasive problem of inadequate subject indexing for electronic images, CLiMB-1 suggested using computational linguistic techniques to mine scholarly texts for metadata terms. It developed a prototype toolkit for image catalogers. The CLiMB Cataloging Toolkit processed electronic texts to identify potential subject terms, matched them to terms in the Getty Art and Architecture Thesaurus, and allowed catalogers to select either natural language phrases or controlled vocabulary terms to be added to image cataloging records in several different formats. CLiMB-1 demonstrated the feasibility of this approach for extracting metadata, tested it with several collections and texts, and evaluated the Toolkit's usefulness for image professionals.

CLiMB-2's objectives are to firm up the Toolkit's capability for handling structured texts, such as exhibition catalogs; to expand its capabilities to cover unstructured texts, such as textbooks and scholarly journal articles; and to ensure its ease of use for catalogers. With partner institutions, CLiMB-2 will test the Toolkit in normal work settings with a range of image collections and texts to establish benchmarks for its use. It will also evaluate the impact of CLiMB-generated search terms on browsing and searching by end users.