New algorithms for composite retrieval,

ZABALA, PAULA; MÉNDEZ-DÍAZ, ISABEL; FEUERSTEIN, ESTEBAN

Valparaiso

Congreso; CLEI 2016; 2016

Pontificia Universidad Católica de Valparaíso

Internet users constantly make searches to findobjects or results of their interest, generally through terms orphrases. Traditional search offers only solutions that take intoaccount just the individual characteristics of the results, and notthe relations they have with the rest of the universe. Typically,we are given an ordered list of the results related to the searchcriterion, which implies the need of changing several times theterms of the query to get to a solution that is more adequateto the intended search goal. As a solution to this problem,Composite Retrieval proposes that the results to a query may begrouped in sets of items (bundles), related through some similaritycriterion, but at the same time are complementary. In this work wepropose heuristic algorithms for Composite Retrieval which areevaluated experimentally, showing performance improvementsover the previous results presented in the literature.