Coûts de distance d'édition pour la Recherche d'Information XML

Structured information retrieval (SIR) on XML documents allows to retrieve focused parts of documents that match the user needs. These needs can be expressed throught content and structured queries, that as well as XML documents can be represented as trees. Our approach uses these trees through tree edit distance to estimate the relevance of XML elements. Tree edit distance is the minimum set of insert, delete, and replace operations to turn one tree to another. The effectiveness of tree edit distance strongly relies on these costs. In this paper we will study the estimation of these costs in the context of SIR. Our model was evaluated over the SSCAS INEX’s 2005 task as well as the INEX’s 2010 Datacentric track and our first results show the interest of such an approach. MOTS-CLES : Recherche d’information structurée, graphes, XML, distance d’édition, DTD.