Design And Implementation Of Community Collaborative Filtering Method Based On Topological Potential

Recent years have witnessed an explosion of information that the exponential growth of the Internet confronts us with an information overload. People have to spend a lot of time to search for the information they need. At present, one of the most powerful tools to solve the problem of information overload is personalization recommendation system which excavates potential interests and makes personalized recommendation for every user by establishing the binary relation between users and products.Collaborative filtering technology is most important and widely used in recommendation systems. The goal of collaborative filtering system is to suggest new items or to predict the utility of items for users based on the users'previous likings and the opinions of the other like-minded users. But there still exist many problems, such as the sparsity of data, the "cold start-up" problem, scalability of the algorithms and so on.To deal with these problems, this paper presents an efficient community collaborative filtering method based on topological potential, in which community discovery technology in theory of complex networks is used. The users in the same community tend to have closer relationship between each other and likely have common interests and hobbies. This paper uses community discovery method based on topological potential to detect the community structure in users'network, and to find the user groups of common interests.This paper brings in a multi-relational data mining method to construct user similarity network, which makes the similarity calculation more comprehensive and reasonable. Experimental results show that compared with other collaborative filtering methods, the proposed method can get higher accuracy and efficiency, and reduce the impacts of sparsity and "cold start-up" problem on system in a certain extent.