Research On Recommendation Algorithm Based On Double Regularization And Social Relationship

In recent years,with the explosion of the Internet information,users have been unable to obtain the information they need effectively and accurately in the face of huge amounts of data.In this context,the emergence of the recommendation system to a certain extent solve the problem,the core of the recommendation system mainly through the analysis of the behavior of users,mining of user preferences and to push for the user information.Collaborative filtering recommendation technology is widely used because it is difficult to implement and requires only users rating on items information and then recommends to users.However,the algorithm still faces data sparsity,cold start and other issues.Previous research results show that the fusion of social relations recommendation has important research significance.On this basis,in view of the existing algorithms of data sparseness and cold start problems still exist,which led to the decrease of the recommendation accuracy.To solve above problems,the concrete research content of this article is as follows:(1)The research of integrated social information recommendation models has received extensive attention.However,the most existing social recommendation models based on the matrix factorization technique ignore the impact of information between items on users' interests,resulting in a decline of the recommendation accuracy.To overcome this problem,which first introduces the attribute information and manifold learning method to calculate the similarity.Then,the matrix factorization model is constrained by the regularization of item association relations and user social relations,this paper proposes a double regularization matrix factorization recommendation algorithm.The experimental results on real data sets show that the proposed method can effectively alleviate the problems such as cold start and sparsity in the recommender system and improve the recommendation accuracy compared with that of existing methods.(2)The traditional social recommendation algorithm ignores the social relations of trust relationship's influence on the recommendations,and most of the algorithm when calculating the user preferences similarity not to distinguish the item category,it will lead to recommend lower accuracy.In order to solve the above problems,this paper proposes a classification based on social relations and user preferences domain matrix factorization of the recommendation algorithm.Which first introduces user preferences domain classification method to calculate the similarity.Then,the trust relationship and the distrust relationship in the social relationship are modeled,and finally the user's social relationship feature matrix is constrained to construct a social recommendation model based on the user's social relationship and preference domain classification,this paper proposes a matrix factorization recommendation algorithm based on social relationship and user preference.and the experimental verification method on Epinions and Ciao data sets can effectively improve the accuracy of the recommendation.