—As of today, the ability of providing personalized user experience has been a critical factor to determine whether a company can be successful or not. Then both academic and industry have devoted a lot of energy to promoting its development. In this paper, with the purpose of generating recommendation ranked lists, we put forward a new recommender scheme based on item clustering and matrix factorization. First, we raise a novel clustering algorithm using distance to obtain latent factors, which gathers items similar to each others successfully. Using the latent factors got from clusters, we generate the item factor vector. In addition, learned from the idea of SVD(Singular Value Decomposition), we adopt matrix factorization to finish the matrix completion. By making a comparison with other algorithms, our approach performs better.
—Recommender systems, clustering algorithm, matrix factorization, latent factor.
Xu Wang, Xingjun Wang, and Zhixing Ding are with the Electrical Engineering Department, Tsinghua University, China (e-mail: email@example.com, firstname.lastname@example.org, email@example.com).
Xinxin Nie and Linghao Xiao are with the Computer Science Department, University of East Anglia, UK (e-mail: firstname.lastname@example.org, email@example.com).
Cite: Xu Wang, Xingjun Wang, Zhixiong Ding, Xinxin Nie, and Linghao Xiao, "A New Algorithm Based on Item Clustering and Matrix Factorization," International Journal of Engineering and Technology vol. 9, no. 2, pp. 160-165, 2017.