Research And Application Of Network Public Opinion Analysis Based On Supernetwork

Along with the rapid development of informationization, the Internet has been recognized as the forth media following the traditional media such as newspaper, radio and television. It has made increasing influence on people's work, living life and public opinion.Therefore, the Internet has become the barometer of public opinion and provided an important platform for public to give their opinions. Although the Internet lowers the threshold of public opinion, it has also become the hardest hit to make and spread rumors.Many factors has made different influence on the network public opinion, so it is very important to choose the appropriate model to analyze the network public opinion.This thesis focuses on the research and application of network public opinion analysis that based on supernetwork. It aims to make a reasonable analysis model of the network public opinion, and to explore the appropriate network public opinion guidance strategy on the basis of the analysis model. The main content is included as following three aspects:Firstly, this thesis makes research on the evolution of network public opinion and builds a framework for analyzing network public opinion. The process of the evolution of network public opinion can be briefly summarized as that the net-citizen expressed their opinion under the action of internal driving forces like psychological factors and external driving forces like environmental information factors. Three kinds of nodes are created during the process, they are net-citizen nodes, driving nodes and opinion nodes. New concepts are cited such as subnetwork and superedge. A subnet is made of the homogeneous nodes while a superedge is lined by heterogeneous nodes. Combined with introducing the nodes, subnets and superedge, the framework has set the foundation for further research of public opinion analysis.Secondly, this thesis constructs a key factor recognition model based on supernetwork after making subnetwork layer analysis, sorting the superedge and identifying the key factor.It is fundamental to make subnetwork layer analysis. First, the environment subnetwork is analyzed from both range and depth aspects. Then, the analysis of the psychological subnetwork is combined with hesitant fuzzy linguistic term set. Next, the vector space model is used to analyze the keywords subnetwork. On the basis of subnetwork layer analysis, the improved superedge sorting algorithm is proposed combining with the hyperlink-induced topic search algorithm (HITS). Lastly, a key factor recognition model is constructed to identify the key factors of the subnetwork layers, such as public opinion leaders, key environmental information, mainstream psychological situation and mainstream public opinion.Finally, this thesis explores the application of network public opinion analysis based on supernetwork. It uses a comprehensive evaluation method and a numerical example to verify the rationality and reliability of the key factor recognition model. Then, some network public opinion guidance strategies are provided based on the supernetwork according to the results of the model analysis.