Hybrid Recommender System Based on Collaborative Filtering Employing Personal Values-Based User Model

Abstract

In this paper, hybrid recommender system based on personal values-based collaborative filtering is proposed. Though collaborative filtering is one of the well-known technologies in recommender systems, it is known that their accuracy falls by biased ratings in datasets. This problem is expected to be solved by collaborative filtering based on personal values-based user model. However, user and item coverages might fall because its effect depends on users’ characteristics. Hybrid recommendation method combining existing and personal values-based collaborative filtering is therefore proposed to solve these problems. The experimental results show accuracy and coverages are improved by hybrid recommendation.