[Cru09] Image retrieval and relevance feedback

Chapitres de Livre : Titre du livre: "Encyclopedia of Database Systems", January 2009, Springer, pp. 1384-1389, (isbn: 978-0-387-35544-3)

Auteurs: M. Crucianu

Mots clés: relevance feedback, active learning, multimedia retrieval, scalability

Résumé: Relevance feedback is a means for refining a query in an information retrieval system by asking the user to specify how relevant each result of the query is. An image retrieval session relying on relevance feedback is interactive and iterative. The session is divided into several consecutive rounds; at every round, the user provides feedback regarding the current retrieval results, usually by qualifying the returned images as either "relevant" or "irrelevant"; from this feedback, the system attempts to better identify the target of the user and to return improved results. A relevance feedback mechanism must maximize the relevance of the results while minimizing the amount of interaction between the user and the system.

Commentaires: Liu, Ling; Özsu, M. Tamer (Eds.)

Equipe: vertigo


@inbook {
title="{Encyclopedia of Database Systems}",
chapter="{Image retrieval and relevance feedback}",
author="M. Crucianu",
note="{Liu, Ling; Özsu, M. Tamer (Eds.)}",