Research On Recommendation Algorithm Based On User Feedback In Social Network Environment

As the data generated by online social networks grows exponentially,the analysis of social network data is becoming more and more important in the research work of the recommendation algorithms.However,most of the current recommendation algorithms have the following two problems: At First,the sparse data seriously affects the recommending quality of the recommendation algorithm;secondly,most recommendation algorithms ignore the trust relationship between users.Therefore,in order to furtherly alleviate the impact of the data sparse on the quality of recommending,this paper introduces social network data into recommendation algorithm.This paper proposes two algorithms: the one is that User Feedback Trust based Recommendation Algorithm,and the one is that User Feedback Trust based Recommendation Algorithm combining Circle.1.In order to alleviate the influence of data sparseness on the accuracy of recommending to some extent,this paper introduces user trust degree into recommendation algorithm,and proposes a recommendation algorithm which is User Feedback Trust based Recommendation Algorithm.It measures user trust degree by analyzing trust relationships and ratings records between users.Apply it to the recommendation algorithm.The effectiveness of the algorithm is verified by experiments on the public CiaoDVD dataset and Epinions dataset.2.In order to furtherly improve the coverage of the algorithm which is User Feedback Trust based Recommendation Algorithm,this paper proposes the algorithm which is User Feedback Trust based Recommendation Algorithm combining Circle.This article believes that the "authority" of users is only reflected in a certain field.In order to provide recommendations to users by finding experts in various fields to furtherly improve the coverage of recommendation algorithm,this paper proposes a method of circle division based on user's interest and trust relationship.Circles are created to find a highly trusted group of users in each circle,which are the "domain experts" in each circle.The effectiveness of the algorithm is verified by using the public CiaoDVD dataset and Epinions dataset.