Abstract de la publi numéro 13478

In this paper we present our structured information retrieval model based on subgraphs similarity. Our approach combines a content propagation technique which handles sibling relationships with a document query matching process on structure. The latter is based on tree edit distance (TED) which is the minimum set of insert, delete, and replace operations to turn one tree to another. As the e ectiveness of TED relies both on the input tree and the edit costs, we experimented various subtree extraction techniques as well as di erent costs based on the DTD associated to the Datacentric collection.