XFIRM: un Modèle Flexible de Recherche d'Information pour le stockage et l'interrogation de documents XML

Users looking for precise information do not want it to be drowned by noisy subjects, as it can be the case in long documents. Thanks to their structure, XML documents allow Information Retrieval (IR) systems to retrieve information units, that are not necessarily whole documents. Two main approaches have been proposed in the literature : database-oriented approaches and IR-oriented approaches, which extend traditional methods and allow to associate relevance values to the returned information units. This paper describes the XFIRM model, combining the advantages of both approaches. The model is based on a complete data representation and storing model. The XFIRM query language aims at allowing users to express their need with a set of keywords and/or in a more precise way with structural conditions. Thanks to the flexibility of the index structure, many information retrieval models could be supported. MOTS-CLÉS : XML, recherche d’information, langage de requête, modèle de représentation