Vers une méthodologie pour l’extraction et la classification automatiques des collocations terminologiques verbales en langue médicale

This study deals with the description of an efficient methodology for the development of a new supervised machine learning system for the extraction and classification of semi-compositional and compositional « verbal terminological collocations » - as defined in Costa/Silva (2004) - which appear in medical language. In particular, we describe the phases of preprocessing of the specialized corpus, automatic extraction of base (noun) + collocate (verb) and formulation of the ground truth for the classification of terminological collocations in order to train and validate an effective automatic system for the French language, which is optimized on the basis of the adopted theoretical premise.