Improved collaborative filtering on recommender based systems using smoothing density-based user clustering

Abstract

Recommender systems improve the user satisfaction of internet websites by offering personalized, interesting and useful recommendations to users. The most famous recommender system algorithm is collaborative filtering. However, the collaborative filtering algorithms need large amount of training data in order to generate the recommendation and the processing of large amount of user profiles dataset causes scalability problem. Furthermore, another problem faced in collaborative filtering algorithm is data sparsity. Existing approaches to these problems mostly ends up with loose of accuracy. In this paper, we propose a smoothing based hybrid recommender system by combining density-based clustering and user-based collaborative filtering to address accuracy and sparsity problem. The experimental results have shown that the proposed method improves the accuracy of recommender system.