A network-based approach for collaborative filtering recommendation

Abstract

Collaborative filtering used in recommender systems produces predictions about the interests of a user by collecting preferences or taste information from many users. A substantial part of collaborative filtering centers on similarity computation. The most popular similarity metrics are Pearson correlation, cosine constrained Pearson's correlation, Spearman rank correlation and mean squared difference. In this paper, we propose a new metric to compute the similarity between two users, based on Jaccard similarity, inter1-link similarity and intero-link similarity. Extensive experimental results and comparisons with other existing recommendation methods based on MovieLens dataset show our proposed recommender system is more effective than traditional collaborative filtering algorithms in terms of accuracy. © Springer International Publishing Switzerland 2013.