Rating Recommenation Research Based On Matrix Factorization

The emergence and popularity of network brings a large amount of data and information to users,meeting the needs of users in the information age.However,with the rapid development of Internet e-commerce,the amount of online data has increased dramatically,the number and types of products appearing more and more,and many users cannot get effective information rapidly in the face of huge amounts of data.Under the background of such information overload,the recommendation system came into being and attracted the attention of experts and scholars in many fields.Recommendation based on matrix factorization is an important method in recommending system.The main challenges are the inaccurate score prediction and the inadequate use of social relations.Relevant research shows that local low rank matrix factorization methods can make a better rating prediction,but how to determine the anchor points of the local low-rank matrix remains an unresolved issue.In addition,social recommendation algorithms that integrate social relationship information further explore how to deal with the data sparsity,scalability and quality of prediction problems,and how to make better use of social relationship information is still a problem in many social recommender system.On this basis,the article briefly analyzes the relevant research on matrix factorization method,social recommendation and network representation learning.In the light of some shortcomings of these methods,the following two aspects are studied:1.We proposes a CLLLORA algorithm based on the heuristic anchor point selection.The basic idea is to generate candidate anchor-points by a clustering method,and then select respective anchor-points based on area density and anchor-points distance criteria.Finally,we can determine the location of the local matrix by selecting the anchor points.In this way,we solve the problem of how to select the local matrix anchors effectively and accurately,which improves the efficiency and accuracy of the local matrix factorization rating prediction.The experimental results on the three data sets demonstrate the effectiveness and efficiency of CLLORMA.2.A MERR algorithm is proposed based on the social relations and rating information of users.In order to overcome the shortcomings of social recommendations,we propose a new framework for which combines network embedding and probabilistic matrix factorization.We make use of social relation features extracted from social networks,on top of which we learn an additional layer that uncovers the social dimensions that explain the variation in people's feedback.Furthermore,the influence of different social network embedding strategies on our framework are compared.Experiments on three real datasets validate the effectiveness of the proposed solution.