Création de clusters sémantiques dans des familles morphologiques à partir du TLFi

Building lexical resources is a time-consuming and expensive task, mainly when it comes to morphological lexicons. Such resources describe in depth and explicitly the morphological organization of the lexicon, completed with semantic information to be used in NLP applications. The work we present here goes on such direction, and especially, on refining an existing resource with automatically acquired semantic information. Our goal is to semantically characterize morpho-phonological families (words sharing a same base form and semantic continuity). To this end, we have used data from the TLFi which has been morpho-syntactically annotated. The first results of such a task will be analyzed and discussed.