Research On Evaluation Algorithms Of Social Network User's Influence

In recent years,the rise of social networking platform has had an underestimated impact on people's access to information and lifestyles.The rapid growth of the number of social platform users has also led to the explosive growth of the total amount of information.The evaluation of user influence has gradually become one of the research hotspots in the direction of social networks.The speed of information dissemination,the scope of coverage and the depth of influence can comprehensively reflect the size of user influence.Therefore,calculating the influence of nodes in the network and analyzing the transmission mechanism of the messages in the network can effectively find out the key nodes in the information transmission network,so that it can make early warning and public opinion analysis of public opinion and is of great significance for network security.This thesis studies and analyzes the status quo of the current social network analysis and development,outlines the main focus of research on social networks at home and abroad,and analyzes the advantages and disadvantages of a variety of evaluation algorithms for user influence in detail.Considering that PageRank algorithm based on link analysis can only quantitatively evaluate user influence from the perspective of network structure,this paper first establishes the index system of user influence based on the characteristics of Weibo and user behavior.Secondly,the defects of the traditional PageRank algorithm are analyzed in detail from the perspective of Markov chain.Finally,the PageRank algorithm based on weighted transfer matrix is proposed based on the factors influencing the user forwarding and comment.At the same time,this paper puts forward a new way to solve this problem by introducing the fuzzy comprehensive evaluation model into the user influence assessment of social network.Fuzzy comprehensive evaluation algorithm can choose evaluation index freely,meanwhile it can solve non-deterministic problem or difficult to quantify problem.In the process of application of the model,taking into account the weight of the traditional fuzzy comprehensive evaluation model and the problem of fuzzy matrix synthesis,this paper compares the different points when the AHP and the entropy method determine the weights and their respective applicable environments A new non-linear fuzzy synthesis operator is proposed.This article tests on a real Sina Weibo data set to assess the impact of users involved in a topic Weibo.By comparing with the traditional PageRank algorithm in the aspects of timeconsuming algorithm,convergence speed and accuracy,the validity of PageRank algorithm based on weighted transfer probability matrix is verified.However,the improved fuzzy comprehensive evaluation model can dynamically adjust the weights,and considering the multiple evaluation factors,the accuracy is also improved.Compared with the improved PageRank algorithm,the algorithm is more efficient and consumes less time with similar accuracy.