Similarity coefficient of collaborative filtering based on contribution of neighbors
Qiaoqiao Li, Zhe Lin, Zhang Fei · 2016
Collaborative filtering is one of the most popular recommendation techniques which makes recommendations to a user based on other users with similar tastes (i.e. neighbors) to the target user. Although most of existing similarity measure between users are symmetry, a user with less uncommonly rated items is able to provide more accurate recommendation to another user. Therefore, the contribution on the recommendation of a user to another may be different. In this study, a similarity coefficient is proposed to measure measures the contribution of a user on recommendation. Experiments suggests that our coefficient improves the existing similarity measures to provide more appropriate recommendation than the original ones.