Research On Weibo Public Opinion Communication Based On User Characteristics

The maturity of internet technology and the development of social media have attracted a large number of internet users,leading to an increasingly complex and ever-changing network environment.When a major public opinion event occurs,users will publish information representing their personal opinions,opinions,emotions,and attitudes.These contents are constantly disseminated and fermented on media platforms,becoming a hot topic of widespread attention on the internet.Analyzing and judging the characteristics of public opinion user communication,as well as the process and methods of public opinion dissemination,can timely take effective measures to avoid the adverse development of public opinion situations and negative impacts on society and government image.At the same time,it can also serve as a reference for relevant institutions such as the government to guide public opinion trends,improve the ability to handle public opinion events,maintain social stability,and create a clear and healthy online environment.This article takes Weibo public opinion as the research object,studying and analyzing the characteristics of user communication and the evolution process of public opinion communication.Firstly,a Bi LSTM+Attention+Emotional Adverb Dictionary sentiment intensity classification model is proposed,which solves the shortcomings of sparse sentiment features in Weibo texts and insufficient mining of contextual semantic relationships.The experimental results show that the accuracy,recall,and F1 values of the classification model are higher than those of other comparative models,and can accurately depict actual public opinion trends;Secondly,it analyzes the information of public opinion users,constructs the public opinion user clustering feature index system from three dimensions of emotion,attribute and behavior,and proposes an improved K-Prototypes algorithm.The improved model increases the difference between target data and improves the clustering accuracy.The improved clustering algorithm has higher RI,JC and FM in mixed cluster analysis than other comparison models.The model clusters public opinion users into four groups,They are respectively attention oriented,edge oriented,authoritative,and expert oriented;Finally,using weighted user communication features to calculate user communication influence,an improved SEIR public opinion communication model based on the HITS algorithm was constructed by taking user influence into account in the SEIR communication model;Simulation experiments were conducted on the SEIR and H-SEIR models using Matlab software platform,and the simulation results of the two propagation models were compared and analyzed.It was found that the improved propagation model more accurately portrayed the evolution process of public opinion.