Attributes Coupling based Item Enhanced Matrix Factorization Technique for Recommender Systems

Abstract

Recommender system has attracted lots of attentions since it helps users alleviate the information overload problem. Matrix factorization technique is one of the most widely employed collaborative filtering techniques in the research of recommender systems due to its effectiveness and efficiency in dealing with very large user-item rating matrices. Recently, based on the intuition that additional information provides useful insights for matrix factorization techniques, several recommendation algorithms have utilized additional information to improve the performance of matrix factorization methods. However, the majority focus on dealing with the cold start user problem and ignore the cold start item problem. In addition, there are few suitable similarity measures for these content enhanced matrix factorization approaches to compute the similarity between categorical items. In this paper, we propose attributes coupling based item enhanced matrix factorization method by incorporating item attribute information into matrix factorization technique as well as adapting the coupled object similarity to capture the relationship between items. Item attribute information is formed as an item relationship regularization term to regularize the process of matrix factorization. Specifically, the similarity between items is measured by the Coupled Object Similarity considering coupling between items. Experimental results on two real data sets show that our proposed method outperforms state-of-the-art recommendation algorithms and can effectively cope with the cold start item problem when more item attribute information is available.