A Collaborative Filtering Systems based on Personality Information

Abstract

As one of the most successful recommendation approaches, collaborative filtering (CF) has been attracting an increasing amount of interest in academia. However, the cold-start problem, where historical data is too sparse since new users have not rated enough items, limits the success of collaborative filtering in certain application domains. In this paper, an innovative item-personality-based collaborative filtering system based on personality information is proposed to resolve the cold-start problem. By incorporating human personality into the collaborative filtering framework, we improve the conventional user-based collaborative filtering method. By a linear combination of both user-based CF and item-personality-based CF, the predicted rating can be obtained. An experiment is conducted to evaluate the effectiveness. The results show that the proposed item-personality-based CF can considerably alleviate cold-start problem.