Vocabulary Matching for Book Indexing Suggestion in Linked Libraries - A Prototype Implementation and Evaluation

In this paper, we report on a technology-transfer effort on using the Semantic Web (SW) technologies, esp. ontology matching, for solving a real-life library problem: book subject indexing. Our purpose is to streamline one library’s book description process by suggesting new subjects based on descriptions created by other institutions, even when the vocabularies used are different. The case at hand concerns the National Library of the Netherlands (KB) and the network of Dutch local public libraries. We present a prototype subject suggestion tool, which is directly connected to the KB production cataloguing environment. We also report on the results of a user study and evaluation to assess the feasibility of exploiting state-of-the art techniques in such a reallife application. Our prototype demonstrates that SW components can be seamlessly plugged into the KB production environment, which potentially brings a higher level of flexibility and openness to networked Cultural Heritage (CH) institutions. Technical hurdles can be tackled and the suggested subjects are often relevant, opening up exciting new perspectives on the daily work of the KB. However, the general performance level should be made higher to warrant seamless embedding in the production environment—notably by considering more contextual metadata for the suggestion process.