OPTIMISATION DE REQUETES DANS UN SYSTEME DE RECHERCHE D'INFORMATION APPROCHE BASEE SUR L'EXPLOITATION DE TECHNIQUES AVANCEES DE L'ALGORITHMIQUE GENETIQUE
Lynda Tamine · HAL (Le Centre pour la Communication Scientifique Directe) · 2000
The thesis deals with the use of genetic algorithmes to tackle information retrieval issues. More precisely, our works focus on applying genetic algorithms for the design of adaptive information retrieval systems. First of all, we study various models in information retrieval and then highlight our contribution in this area. After this, we focus on genetic algorithms. We particularly show how to exploit their formal support and robistness in order to support query optimization approaches. Our specific approach consists in combining both relevance feedback evidence and genetic processing in order to refirmulate the querues. The proposed genetic algorithm is specifically devoted for information retrieval by designing enhanced genetic operators that tackle the multimodality relevance problem. The niching technique is applied to the whole population in order to involve a multi-optimization query through a cooperative approach. The experimental evaluation has been carried out using a TREC collection.