Title: A hybrid framework for similarity-based recommendations

Authors: Nikos Karacapilidis, Lefteris Hatzieleftheriou

Addresses: Industrial Management and Information Systems Lab, MEAD, University of Patras, 26500 Rio Patras, Greece. ' Industrial Management and Information Systems Lab, MEAD, University of Patras, 26500 Rio Patras, Greece

Abstract: By exploiting the concept of fuzzy similarity measures, this paper presents a hybrid recommendation framework that builds on the strengths of knowledge-based and collaborative filtering techniques. Following a multi-criteria approach, the proposed framework is able to provide users with a ranked list of alternatives, while it also permits them to submit their evaluations on the existing items of the database. Much attention is given to the extent to which the user evaluation may affect the values of the stored items. The applicability of our approach is demonstrated through an already implemented web-based tool, namely CityGuide, which provides recommendations about visiting different cities of a country. Issues related to the robustness of our framework and the selection of the appropriate similarity measure are also discussed.

Keywords: recommendation techniques; recommender systems; web services; fuzzy similarity measures; knowledge-based filtering; collaborative filtering; alternatives ranking; user evaluation.

DOI: 10.1504/IJBIDM.2005.007321

International Journal of Business Intelligence and Data Mining, 2005 Vol.1 No.1, pp.107 - 121

Published online: 05 Jul 2005 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article