Research On Graph-based Social Network User Impact Ranking Method

With the rapid development of human society and the advancement of science and technology,human society has gradually developed from the era of industrial economy to the era of information economy.Social media appeared in the era of Web2.0,and rapid development,our daily life is full of various networks.Social media is abstracted as a social network,which can be seen as an online communication between people and attracts more users to participate in it.Through online communication and publishing information in social networks,users will identify influential users,increase the breadth and depth of information dissemination,and play a very important role in information dissemination,network public opinion dissemination control,and advertising dissemination.How to effectively measure the actual influence of users in social networks,and further sort the influence of social network users has become a key issue that requires in-depth research.The main research work of this article includes:(1)Improve the LR(LeaderRank)value equalization and bias to the old webpage by introducing the network motif in the network topology and introducing the average performance feature factor and the time decay factor in the user's own characteristics.(2)Introduce a distance metric to balance the correlation factor and diversity factor to further improve user influence ranking,and seek the k subset of the size of the nodes with the largest edge weight sum of induced subgraphs.The correlation factor is the LR value of the improved node pair,and the diversity factor is the symmetry difference of the node pair.The distance metric is used to balance the user satisfaction and the user experience.(3)A highly parallelizable method with approximate guarantees is proposed to further improve the diversity graph ranking.Using the distance metric,a centralized linear time approximation algorithm is proposed to solve the diversity graph ranking,and based on the algorithm,the framework is further used in a highly parallel method to solve the diversity graph ranking and obtain the diversity graph ranking.Approximate guarantee.In response to the above problems,we use the Twitter network as a research case to rank the influence of social network users.By referring to the classic literature,we use different experimental methods to experiment with existing methods under relevance and various diversity measures.As well as the result analysis,the experimental results clearly show the effectiveness and efficiency of our improved algorithm in ranking the influence of social network users.