Exploitation de la morphologie pour l'extraction automatique de paraphrases grand public des termes médicaux

The medical area conveys very specific terms (p. ex., blepharospasm, appendicectomy), which are difficult to understand by people without medical training. We propose an automatic method for the acquisition of paraphrases, which we expect to be easier to understand than the original terms. The method is based on the morphological analysis of terms, syntactic analysis of texts, and text mining of non specialized texts. An analysis of the results and their evaluation indicate that such paraphrases can indeed be found and show easier understanding level. According to the setting of the method, precision of the extractions ranges between 90 and 7.4%. Such resources are useful for several NLP applications (p. ex., information retrieval, text simplification...). MOTS-CLÉS : domaines de spécialité, terminologie médicale, composition, analyse morphologique, paraphrase, compréhension.