Recherche de passages pertinents dans les fichiers logs par enrichissement de requêtes

The question answering systems are considered the next generation of search engines. This paper focuses on the first step of this process which is to search for relevant passages containing the responses. Such a task can be difficult because of the complexity of data, logs files in our case. Our contribution is based on enrichment of queries using a learning method based on the notion of "lexical world" and a novel term weighting function. This original weighting function, implemented within the enrichment process, aims to assign a high weight to terms that might be relative to the context of the answer. Experiments conducted on real data show that our protocol of primitive query enrichment allows to extract relevant passages. MOTS-CLÉS : Recherche d’Information, Système question réponse, Recherche de passage, enrichissement de requête, apprentissage de contexte