A cross cluster-based collaborative filtering method for recommendation
Ming Zhou Gao, Fuyuan Cao, Joshua Zhexue Huang · 2013
As the clustering-based model has better scalability than typical collaborative filtering methods, it has become one of the most successful approaches for recommender systems. However, since clustering-based algorithms often result in nearby users being divided into different clusters, they only recommend items being rated by users belonging to the same cluster with the active user, and recommendation opportunities are missed for some users because of the loss of nearby users. In this paper, we propose a cross cluster-based method to take more recommendation opportunities by considering nearby users through merging of neighbors in user clusters. We define an associate degree to find the neighboring clusters. Experimental results on real data sets have shown that the proposed method can improve the accuracy of recommendation.