A Systematic Evaluation of Concept-based Cross-Lingual Information Retrieval In . . .

The paper describes experiments and results of the MuchMore project, which is concerned with a systematic comparison of concept-based and corpus-based methods in cross-language information retrieval (CLIR) in the medical domain. Primary goals of the project are to develop and evaluate methods for the e#ective use of multilingual thesauri in the semantic annotation of English and German medical texts and subsequently to evaluate and compare the impact of such semantic information for the purpose of CLIR. In particular we describe work on semantic annotation with both domain-specific (UMLS, the Unified Medical Language system ) and general language semantic resources (EuroWordNet (Vossen, 1997)). Central to the approach is the use of linguistic processing (part-of-speech tagging, morphological analysis, phrase recognition and grammatical function analysis) for an accurate semantic annotation of relevant terms and relations in both the queries and the documents (Vint