Research On Social Network User Influence Evaluation System Based On Topic Correlation Analysis

The concept of social network appeared in the 1920s and 1930s.With the progress of science,researchers gradually put forward regular network,random network and other models to describe some systems in nature.With the emergence of the Internet,scholars have found a new network model,namely complex network.It is a common research method to apply the theory of complex network to the analysis of social network.Many scholars have proposed centrality evaluation algorithms for network nodes based on graph theory,such as PageRank,leaderrank and weighted leaderrank.These algorithms can find out the key nodes in the network to a certain extent,but the main factor to evaluate a node is the topology of the network.Some scholars add node attributes to the evaluation algorithm,but they do not solve the problem of the lack of topic differentiation.At the same time,these algorithms also tend to have a large number of fans.In view of the above problems,this paper makes innovation from two aspects of algorithm improvement and system design.First of all,this paper takes Zhihu as the specific research object,through the analysis of the characteristics of social networking sites and user structure,proposes an improved algorithm named top leaderank(TPR)with weight leaderank algorithm.On the basis of related algorithms,this algorithm adds topic relevance to the evaluation algorithm,and uses the methods of topic extraction and topic classification to analyze the content published by users,focusing on the characteristics that the influence of a user under different topics will change in a social network.At the same time,in order to solve the problem that the current algorithm is biased towards users with a large number of fans,we propose a topic related expert influence score(TES)calculation method.Based on the idea of peer review,the algorithm combines the TPR score of users with the approval relationship between users.Improved the relevant algorithm does not take into account the impact of different users for the same answer to agree with the differences in the impact of evaluation problems.Secondly,based on the expert influence score algorithm proposed in this paper,a comprehensive user influence evaluation system is designed.The system includes four modules:data collection,data processing,data storage and influence calculation.In the system,we propose a design method that combines the influence calculation and data collection module,so that the influence evaluation can guide the crawler to collect data.This design method effectively optimizes the current problems of data collection difficulty and data content in social networking sites.Finally,the algorithm and system are verified by experiments.From the experimental results,the proposed algorithm can give some users who provide high-quality content but have fewer fans a higher score.At the same time,under different topics,the user’s score will be different according to whether the content is related to the topic.At the same time,the evaluation system designed in this paper has higher efficiency and topic relevance in data acquisition,and the operation of the whole system is relatively stable.