Proposal of a Recommendation Method by Direct Setting of Preference Patterns Based on Interrelationship Mining

Abstract

In this paper, we introduce a recommendation method by directly setting of users’ preference patterns of items for similarity evaluation between users. Yamawaki et al. proposed a recommendation method based on comparing users’ preference patterns instead of directly comparing users’ rating scores. However, Yamawaki et al.’s method for extraction of preference patterns has limitation of representation ability and this method is not applicable in the case that a user has “cyclic” preference patterns. Our method for extraction of preference patterns is based on partial pairwise comparison of items and it is applicable to represent “cyclic” preference patterns. A tourist spot recommender system is implemented as a prototype of recommender system based on the proposed approach and the implemented recommender system is evaluated by experiments.