Research On Potential Topic Detection Of Network Public Opinion Based On Social Network Analysis

The rapid development and wide application of the Internet makes the network public opinion become the most important part of social public opinion.The characteristics of social networks and media propagation makes microblog become one of the most influential derivative sites of network public opinion.Most of the existing research is based on the results of the network public opinion evolution,which has been in a passive state to solve problem.And the needs of tracking in advance and real-time discovery of network public opinion isn't fully satisfied.The potential topic detection of the network public opinion can explore the core content of it timely,which can strengthen the potential topics research of network public opinion,which is helpful to grasp the rules and evolution mechanism of the network public opinion correctly,and it has a great significance in building a good environment for the network public opinion.First of all,this paper has carried on a thorough research about the achievements of the potential topic detection,having a deep understanding of the current situation about the studies on the topic detection of the network public opinion and making an in-depth analysis about the application of social network analysis method.Then,it has a systematic analysis of the microblog data,which includes the content,the user attribute and the user behavioral data.On this basis,the research combines the social network analysis and the microblog data characteristics to construct a model about the potential topic detection of the network public opinion as well as to illustrate the key indicators and methods used in the model:(1)building the user relationship network based on the microblog user behavior and giving specific weight according to the different user behaviors and concerns relationship;(2)identifying the community discoveries of the user relationship network by using community detection methods and calculating the between centrality and the eigenvector centrality of the nodes;(3)ranking the influence of the main community user nodes in descending order according to the influence formula of user nodes and screening the key nodes of community user;(4)mapping the key user nodes of the main community to the corresponding microblog to get the key microblog nodes;(5)ranking the weight of the keywords by TF-IDF method and selecting the alternative potential theme keywords for co-word analysis to get the result list of the potential topic detection.Finally,this paper will take the WeiZexi Event as a case study to verify the effect of the model.The experimental results show that the proposed potential topic detection method achieves good results.