Acquisition sur corpus d'informations lexicales fondées sur la sémantique différentielle

Semantic lexicons are an essential resource to let many natural language process-ing applications (automatic summarization, information retrieval, automatic transla-tion, etc.) penetrate the meaning of a text. The relevance of the information gathered bythose lexicons raises a problematic question: the meaning of a word likesoap, for ex-ample, varies considerably whether it is considered in a sanitary or televisual context.A linguistically motivated and cost-effective way of building semantic lexicons pre-cisely adapted to a certain domain of expression consists in “learning” word meaningsfrom their actual usage as observed in a representative collection of texts, or corpus. To answer this challenge, we propose in this document a three-stage methodol-ogy for the automatic acquisition of lexical semantic information from texts, based onthe linguistic principles of F. Rastier’s Interpretative semantics. Thanks to a statisticalanalysis of word uses, employing both classical and novel methods, we first man-age to bring together words belonging to a same domain (for exampledata, transfer,network for IT), then to build classes of words having a similar meaning (data andinformation). We finally propose a first method to put to light fine-grained meaningdistinctions between close words(data is more “concrete” than information), thusreaching a level of meaning refinement never before attained, to our knowledge, byautomatic means.