Research And Application Of Collaborative Filtering Algorithm For Social Network

With the rapid popularization and the wide application of Internet and information technology, all kinds of social network are emerging on the Internet. In addition, social network has become an important application on the Internet that more and more people rely on to find their interested information. However, the rising of the number of social network user and the massive interactive information generated by frequent interactions among users result in the problem of information overload in social network. Personalized recommendation is the most effective way to solve the problem of information overload. Therefore, the personalized recommendation in social network has become a hot research topic, which has attracted the attention of scholars.Collaborative filtering algorithm is the most widely used personalized recommendation algorithm. Firstly, this paper proposes a collaborative filtering algorithm based on user trust degree and social similarity to solve the weakness of the collaborative filtering algorithm application in the field of social network. Secondly,this paper proposes a collaborative filtering algorithm based on social network user similarity clustering. Finally, above proposed algorithms are used to develop a recommendation system for social network. The main work of this paper is as follows:(1) A collaborative filtering algorithm based on user trust degree and social similarity is proposed to solve the weakness of the collaborative filtering algorithm application in the field of social network, which leads to the low accuracy of personalized recommendation in social network. Firstly, the algorithm calculates user trust degree as well as social similarity through constructed user network and calculates user similarity according to the user-item matrix. Next, the user similarity, user trust degree and social similarity will be merged to form a comprehensive value, which is used to produce neighbors. Finally, the optimal score prediction formula is used to generate recommendation results for the target user based on the user-item rating data of neighbors. The experimental results and analysis indicate that the proposed algorithm has higher recommendation accuracy than other algorithms in solving therecommendation issues of social network, but its efficiency will reduce with social network size increasing.(2) A collaborative filtering algorithm based on social network user similarity clustering is proposed to solve the problem that the efficiency of the collaborative filtering algorithm based on user trust degree and social similarity declining with the scale of the social network increasing. Firstly, the algorithm calculates the social network user similarity according to the attribute information and interactive behavior of users in social network. Next, the improved k-means algorithm is used to cluster the users according to the social network user similarity. Finally, a collaborative filtering algorithm based on user trust degree and social similarity is used to generate recommendation results on each user cluster. The experimental results and analysis indicate that the proposed algorithm not only significantly reduce the running time but also improve the accuracy of recommendation in dealing with large scale social network data sets.(3) A recommendation system for social network is designed and implemented.The collaborative filtering algorithm based on user trust degree and social similarity as well as the collaborative filtering algorithm based on social network user similarity clustering are chose to apply to the recommendation system according to the user scale of the recommendation system.