Research On Multi-Dimensional Recommendation Of Social Network Based On Link Prediction

With the rapid development of Internet technology,a variety of social networking sites and applications emerge in an endless stream,making people's social and various information needs guided and released.However,the surge in the number of users and information content on social networking sites has made it difficult for ordinary people to expand social circles and obtain effective resources.Therefore,social network recommendation is a hot research field with practical applications.However,there are some deficiencies in the current research.For example,common indicators of friends recommendation are not effective when the data is sparse.The common methods of content recommendation neglect the network structure and limit the range of recommended results.But there are a large number of users,resource nodes,and complex network relationships in social networking sites.Therefore,this article considers the relationships between users and resources in social networks comprehensively,adopts similarity-based link prediction methods,and the similarity index in the existing link prediction method is improved,and the related recommendation of the social network is finally realized,including user recommendation and resource recommendation.The research content of this paper can be summarized as the following:(1)The scope and advantages and disadvantages of the theory and existing methods of social network recommendation,link prediction are discussed;(2)The multi-dimensional features of social networks are analyzed.The social network is characterized from two perspectives:node and relationship,and the relationship between user-user,user-resource,and resource-resource dimensions is analyzed to provide a basis for link prediction similarity calculation;(3)According to the characteristics of social networks,appropriate improvements have been made to the existing similarity indicators in the link prediction,so that the new method is more suitable for the recommendation of social network.The method proposed in this paper is implemented on the Douban data set,which a content-based social networking site.The experiment starts with user recommendation and resource recommendation,and sets up different control groups.The experimental results were evaluated from the precision,recall and F1-measure.The results show that the algorithm has a good effect on accuracy,recall rate and F1-measure value.