Collaborative Filtering Recommendation Algorithm Based on Users of Maximum Similar Clique

Abstract

In order to improve the performance of Collaborative filtering (CF), a new method of producing the nearest neighbor for active user is proposed in this paper. Inspired by the conformist of E-commerce consumers, we build the user model of maximum similar clique and we use it to improve the method of producing the nearest neighbors for target users. A collaborative filtering recommendation algorithm MCQ-CF based on user model is present. The experiment results show that the algorithm MCQ-CF has good performance for accuracy and stability.