Research On Group Recommendation Technology Based On Social Network User Interests

Social networks such as Facebook,Twitter,Douban,Sina Weibo that have become the most popular information dissemination platform in the world today.With the development of social networks and the continuous expansion of users,it is becoming increasingly difficult for users in social networks to find the information they may be interested in as the information is constantly updated.How to filter out unfamiliar information for users in social networks and provide recommended services that meet the needs of users' preference has become a research focus of academic and industrial concern.Currently,most recommender systems mainly focus on the recommendation of individual users.However,it is necessary to recommend groups formed by multiple users in many daily activities.In recent years,group recommendation systems have gradually become one of the hot topics in the field of recommendation systems.Group recommendation systems as an effective way to address the problem of group recommendation,due to the complexity of social network,the quality of traditional group recommendation system in social network is not high.In order to solve the above problems,this dissertation focuses on the influence of social relations,interest spread and potential trust on the recommendation method of social networks,proposes a group recommendation method based on the interests of social networks,and meets the needs of user groups in social networks.Achieve high-quality social networking group recommendation,the main work is as follows:1)Propose a group recommendation algorithm based on external trust network.The algorithm will extract the user groups trusted by the user groups and obtain the external group's evaluation of the items through the fusion strategy by using the true evaluation information of the trusted objects.The resulting external evaluations and traditional group recommendations are then weighted to obtain the new group recommendation results.The data set uses Filmtrust,a data set that has trust data in addition to movie data.Experimental results show that the proposed method can effectively reduce the recommended error.2)Propose a group recommendation system based on estimation of social network trust.At present,most social networks,the trust relationship is not directly given,usually only get the structural information between users,we have some ways to estimate the trust relationship between users,not only that,but also consider the issue of the spread of trust,And find out the relationship between the radius and the user's influence,and finally recommend the group to get a good recommendation.The dataset uses Douban movie,which has social network data in addition to movie data.Experimental results show that the proposed method can effectively improve the recommended performance.