Thesaurus-based query and document expansion in conceptual indexing with UMLS
Résumé
UMLS is known as largest thesaurus in biomedical domain constructed by Library National of Medicine. In this paper, we aim to evaluate effect of the exploration of UMLS knowledge in medical domain information retrieval by mapping large text of collection ImageCLEFMed to UMLS concepts, and expanding queries and documents automatically base on semantic relations in the UMLS hierarchy. We get the encouraging result with the best enhancement on MAP of 66% compared to text only retrieval, and 34% compared to conceptual indexing baseline.