A multicriteria paradigm of relevance for the Web Information Retrieval problem

We consider the problem of ranking Web documents within a multicriteria framework and propose a novel approach for the purpose. We focus on the design of a set of criteria aiming at capturing complementary aspects of relevance, while acknowledging that each criterion has a limited precision. Moreover, we provide algorithmic solutions to aggregate these criteria to get the ranking of relevant documents, while taking into account the specificities of the Web information retrieval problem. We report on results of preliminary experiments that give first justification of the pertinence of the proposed approach to improve retrieval effectiveness.ou