The Evolution Analysis And Modeling Of Public Opinion Based On Online Social Network

Complex network theory abstracts the complexity of problems in reality,which makes it possible to study the nonlinear relations between entities.In recent years,more and more scholars have made breakthrough achievements in the field of information spreading.At the same time,the development of technology has changed the ways people interact with others from face-to-face communication and text messaging to online social media.Due to the low threshold,fast speed and timeliness of online social platforms,the information published on these platforms can attract the attention of a large number of individuals in a very short time.Under the influence of these characteristics,there have been public opinion events that have had a great impact on the society on online social networks for many times.In this thesis,based on the information data of public opinion events on Sina Weibo platform,we conduct empirical analysis and models of the evolutionary process of public opinion,in an attempt to explore the evolution mechanism of public opinion on social networks.The researches mainly include the following contents.First,this thesis collects data from the public domains of Sina Weibo platform.In order to distinguish from the research of other scholars on single public opinion event,this thesis proposes the concept of Public Opinion Event Group.According to the obtained data,9 public opinion event groups that had great influence on the society were selected as research objects.In the part of empirical analysis,we construct public opinion propagation networks and the user's fans relationship network,and analyze them by using the theories of temporal networks and multilayer networks.In this thesis,we find that the evolutionary process of public opinion exhibits the characteristics of multiple phase transitions,clustering and explosion at the macro level,and the non-Markov characteristics at the micro level.In addition,this thesis takes the number of users participating in public opinion event groups as an index to quantify the activity of users.We find that highly active users are also important in the topology of the user's fans network,and they can influence the evolutionary process of public opinion.Secondly,this thesis makes an empirical analysis of public opinion data of online social networks,and based on the empirical analysis results,proposes an information propagation model based on individual differences and an information propagation model based on temporal.In the information propagation model based on individual differences,the propagation probability is a function of the importance of nodes in the topology.In the experimental results,the clustering and explosiveness of public opinion event propagation process are reflected.In the information propagation model based on temporal,the propagation probability is a function of time.In the experimental results,the non-Markov characteristics of the behaviors of users related to public opinion events are reflected.