Optimisation de la pertinence dans un SRI: Un problème multi-modal approché sous l'angle de la génétique

This paper presents a genetic relevance optimisation process performed in an information retrieval system. The process uses both genetic technique for solving multimodal problems witch is namely niching, and query reformulation techniques commonly used in information retrieval. Niching technique allows the process to reach different relevance regions of the document space. Query reformulation techniques represent domain knowledge integrated to the genetic operators structure in order to improve the converge conditions of the algorithm. Exprimental analyses performed using a TREC sub-collection validate our approach. MOTS-CLES : Recherche d’information, evaluation multi-requêtes, algorithme génétique, nichage