Expansion de requêtes par apprentissage

We propose in this paper a learning query expansion approach using association rules. The query expansion problem is modeled as a supervised classification problem which aims at identifying the appropriate set of association rules to expand a given query. A training data set is generated using a GA based exploring algorithm of the association rules space. Classification is made by the method of the decision tree and the Random Forest method . The experiments are conducted on the French texts SDA95 collection of CLEF evaluation campaign 2003. The results show an improvement in task performance IR. MOTS-CLÉS : règles d’association, expansion de requêtes, apprentissage supervisé.