Research On Social Recommendation Based On Trust And Attention Degree Of Item's Characteristic Attributes

As an efficient information filtering method,the recommendation system is one of the effective methods to solve the problem of information overload and implement personalized recommendation.However,with the rapid development of information technology and the rapid popularization of a large number of electronic devices,the data in social media has exploded,increasing the problems of poor recommendation performance and low efficiency caused by "information overload".Social recommendation algorithm integrates social attribute information such as social tags and trust relationships in the traditional recommendation algorithm based on user-item binary model.Effectively improve the recommendation efficiency of the recommendation system.At present,there have been many studies related to social recommendation,and some social recommendation methods have been proposed,but there are still problems of data sparseness and cold start in these methods.This article analyzes social trust and user characteristics of project characteristics.Pay attention to improve the recommendation efficiency of social recommendations.The main work in this paper is as follow:The concept and classification of traditional recommendation system and social recommendation system and common recommendation algorithms are systematically described and analyzed.The common models of social recommendation and the existing problems in existing social recommendation models are analyzed in-depth.The differences between traditional recommendation systems and social recommendation systems are summarized.In order to solve the problem of insufficient trust relationship mining in the social recommendation system,this paper proposes a social trust-based recommendation method to mine users' trust relationships through auxiliary behaviors such as social behavior and social circles to improve the trust between users.The joint matrix factorization method is used to calculate the user's preferences,which improves the recommendation quality of the social recommendation system.In order to be able to accurately mine similar relationships between users and user preferences.This paper proposes a social recommendation model that integrates trust and the degree of attention of project characteristic attributes.A new heterogeneous network is constructed through the user-item scoring matrix and trust relationships,and similar relationships between users are mined in the new heterogeneous network.When calculating the user's prediction preference,the user's attention to the item's characteristic attributes is introduced to improve the user's recommendation efficiency.Finally,a comparison and analysis on the Filmtrust dataset with three other typical social recommendation models proves that the model effectively improves the data sparseness and cold start problems in the social recommendation system.