Structural-Based Relevance Feedback in XML Retrieval

Contrarily to classical information retrieval systems, the systems that treat structured documents include the structural dimension through the document and query comparison. Thus, relevant results are all the document fragments that match the user need rather than the whole document. In such case, the document and query structure should be taken into account in the retrieval process as well as during the reformulation. Query reformulation should also include the structural dimension. In this paper we propose an approach of query reformulation based on structural relevance feedback. We start from the original query on one hand and the fragments judged as relevant by the user on the other. Structure hints analysis allows us to identify nodes that match the user query and to rebuild it during the relevance feedback step. The main goal of this paper is to show the impact of structural hints in XML query optimization. Some experiments have been undertaken into a dataset provided by INEX1 to show the effectiveness of our proposals.