Social Network User And Resource Recommendation Based On Node Similarity

With the continuous development of information technology and the Internet,various information services are becoming more and more popular.Social networks have evolved from traditional communication tools to platforms that meet the social needs of people's communication,learning and entertainment.The rapid development of positioning technology and sensors has generated a large amount of user behavior data,making the recommendation system a hot topic for academic and industrial research.Faced with people's growing social needs,this article focuses on the ability to get effective information and friends' recommendations in social networks.In the research recommendation method,more attention is paid to the features and structure between nodes,and the similarity of nodes is used as the basis of social needs.The combination of graph theory skills,fuzzy mathematics and probabilistic methods are used to link the three-degree network and have common topics.The comprehensive influence and user similarity of the viewpoint-oriented population are portrayed and the recommendation list is generated;the recommendation algorithm for the tripartite graph is finely divided by the introduction of the “entry and exit degree”,supplementing the missing weight,using the material diffusion algorithm and the heat conduction algorithm.Establish a tripartite graph resource recommendation model to effectively conduct personalized friend recommendation and resource recommendation.In the social network that focuses on acquaintances,this paper calculates the influence of user friends in the third degree of influence on the target users from two aspects.One is to create a connectivity factor based on the connected subgraphs between two users,and the edge between the two users.Give reasonable weights,take into account the distance between users,use graph theory to get the structural influence formula;the second is to calculate the closeness of the topic according to the topic frequency membership degree,and get the fuzzy influence formula.Combined with structural influence and fuzzy influence,the comprehensive influence formula in the three-degree connection network is constructed,and the recommended scheme for the target users is obtained.Through the analysis of micro-blog text,we can use graph sorting algorithm to distribute views directly to micro-blogs with clear viewpoint characteristics;for micro-blogs without viewpoint characteristics,we can calculate the probability of users publishing a topic viewpoint through context relations and graph sorting algorithm;and we can use the user viewpoints with common topics to construct differences and get new calculation among users.Similarity formula is used to obtain user recommendation scheme based on semantic analysis.Faced with the information deficiency and unbalanced transmission of resources in the traditional tripartite graph recommendation algorithm,this paper defines the initial resources by the number of node tags and tags;introduces the concept of access degree to refine the problem;constructs one-way complete bipartite graph and directed star graph,and establishes a directed graph model integrating the relationship among users,items and tags;and combines the active period of the network.A comprehensive weighting formula is given based on the time-decay property of labels,which converts the calculation of user similarity and label similarity into the weighting problem of one-way complete bipartite graph.The resource recommendation scheme of weighted tripartite graph network structure is obtained by using the multiplication principle and addition principle in statistics.