Research On Recommendation Algorithms Based On Social Relationship Mining And Influence Analysis

With the popularity of the Internet,the amount of information on the network has increased dramatically,making it difficult for users to find the information they need from the vast amount of information,and the efficiency of using the information has decreased,which has brought about information overload.Recommender systems automatically explore users' interest points by exploring users' interest preferences,thereby recommending information that the users may be interested in,and solving the information overload problem.The collaborative filtering recommendation algorithms have the advantages of simplicity,high efficiency,and easy deployment,and have been widely used in recommender systems.However,the traditional collaborative filtering recommendation algorithms only use the users' ratings on items for recommendation,and face many problems such as data sparsity and cold start,which makes the recommendation results less accurate.This paper focuses on the problem of data sparsity in traditional collaborative filtering algorithms,and combines the social relationship between users into recommendation algorithms to improve the accuracy of recommender systems and solve the cold start problem.The research work and innovations of this paper are as follows:(1)The topological influence of users in the recommendation systems in social networks is analyzed,and a social recommendation algorithm based on collaborative filtering is proposed.Based on the matrix factorization algorithm,the users' topological influence and global influence are combined.The users' topological influence constraints the degree of similarity between the user and his/her trusted friends and the important users in the network.The global influence constrains the weight of a user's ratings in the prediction rating error function.Experimental results on two real data sets show that the algorithm can provide more accurate rating prediction for all users and cold start users in the system compared to other algorithms.(2)The indirect relationship between users is analyzed.Combined with implicit feedback,a collaborative filtering recommendation algorithm based on trust propagation is proposed.First,the calculation method of indirect trust value between users is proposed.On this Dasis,indirect trusted friends are searched for users in the system,and the influence of indirect trusted friends on target user preferences is introduced into the recommendation model.The above research results are introduced into the collaborative filtering recommendation algorithms,and a recommendation algorithm based on implicit feedback and regularization terms is proposed.The experimental results show that the algorithm has higher accuracy than other algorithms.This paper uses 12 figures,10 tables and 68 references.