Multilingual access to library catalogues: Word sense disambiguation via classification systems

Abstract. Multilingual Digital Libraries ask for Cross-Language search engines able to disambiguate the search terms across languages. In this paper, we report on the development and first evaluation of a Cross-Language Information Retrieval (CLIR) system enriched with a Word To Category Module: we automatically extract a mapping from words in the meta data of the Catalog’s records to the associated Classification Categories and exploit it to help the CLIR system retrieve only books in different languages about the topic actually queried by the user.