Technology Research, Social Network-based Collaborative Filtering Recommendation

With the wide popularity of Internet,traditional trade and business' activities have been revolutionarily changed by e-commerce, which also causes the reform from model of commodity-oriented to customer-oriented. it is a trend for enterprises to provide personal service to meet the different needs for different people. E-commerce recommendation system appeares in such a situation, which can effectively retain customers, increase enterprise sales, improve service quality and enhance the competitiveness of enterprises.Recommendation system has a wide range of applications and good development prospect in the area of electronic commerce. It gradually becomes an important part of e-commerce and attracts a large number of researchers' attention. Nowadays, there are three main recommendation technologies in the recommendation system, including content-based,collaborative filtering and hybrid recommendation technology. Collaborative filtering is one of the most successful applications of recommendation technology. However,due to data sparsity and cold start issues of collaborative filtering and the growing of data scale in the E-commerce, e-commerce Recommendation system faces many challenges.With the full understanding of the principles and problems of the collaborative filtering recommendation, further useful exploration and research are made in this paper. we propose a collaborative filtering recommendation method based on community detection ,which brings methods of community detecting into collaborative filtering.First,we focuse on the community detection algorithm and propose two algorithms based on core node and k clique about community detection ,which effectively find communities in the network.Second,we select a part of user communities from the user network projected by user-item network as the candidate neighboring user set for the target user, thereby reducing calculation time and increasing recommendation speed of recommendation system.Finally, to make up for defect of few rating information, we add prerating mechanism into collaborative filtering to solve the problems arising from data sparity, rasing precision of recommendation system.This paper has a perfect combination of social network technology and collaborative filtering technology,which can greatly increase recommendation system performance.Finally, we carry out experiments by MovieLens data set to test two performance indexes which including MAE and RMSE. Results demonstrate that algorithm proposed in this paper is better than the algorithms based on the Pearson similarity and Cosine similarity.