Research On Collaborative Filtering Recommendation Algorithm Based On Clustering And Expert Opinions

With the rapid popularization of Web 2.0 technology,the data and resources on the Internet are in exponential growth stage,which will cause the user to face the problem of information overload.Collaborative filtering uses the user rating matrix to compute the similarity of the user,and recommends to the target user according to the contiguous user's preference.However,there are some problems in collaborative filtering,such as cold start,data sparsity scalability and so forth.Aiming at these problems,this paper makes use of the advantages of the hybrid recommendation algorithm,and improves the collaborative filtering algorithm.The main work includes:(1)the related technologies of recommendation system are summarized and analyzed,and the research status of collaborative filtering algorithm is analyzed.Based on the analysis of the disadvantages of collaborative filtering algorithm,this paper discusses how to use the existing technology to alleviate the disadvantage of collaborative filtering.(2)to solve the problem of cold start and recommendation accuracy,a collaborative filtering recommendation algorithm based on user characteristics and expert trust is proposed.By introducing the characteristics of users can fill out the use of the user registration information recommendation system to effectively alleviate the cold start problem;similarity between users and expects can be compared by introducing expect trust;by calculating the similarity matrix of experts-users,can effectively reducing the sparsity of data sets,and improve the accuracy of prediction.It can be seen from the experimental results that the improved algorithm can effectively alleviate the cold start problem and improve the accuracy of the system.(3)to solve the problem of data sparsity and scalability,a collaborative filtering recommendation algorithm based on singular value decomposition and K-means++clustering is proposed.The user into multiple clusters,and then find the neighbor set in with the target user and similar clusters,which can alleviate the sparsity of data,and also reduces the amount of computation;the singular value decomposition was used to reduce the dimension of the user item matrix,and fill the sparse matrix,to complete the establishment of the model in the offline state.The experimental results show that the algorithm can effectively improve the sparsity and scalability.