Analyse morphologique non supervisée en domaine biomédical. Application à la recherche d'information

In the biomedical field, using of specialized terms is key to access information. However, in most Indo-European languages, these terms are complex morphological structures. The presented work aims at identifying the various meaningful components of these terms and use them to improve biomedical Information Retrieval (IR). We present different approaches combining automatic alignments with a pivot language, Japanese, and analogical learning that allows an accurate morphological analysis of terms. These morphological analysis are used to improve the indexing of medical documents. The experiments reported in this paper show the validity of this approach with a 10% MAP improvement over a standard IR system. MOTS-CLÉS : morphologie, terminologie biomédicale, alignement, apprentissage par analogie, indexation morphosémantique, recherche d’information biomédicale.