The Evolution And Response Strategies Of Micro-blog Public Opinion In Emergencies Based On Topic Analysis And Sentiment Analysis

In recent years,media technology and the Internet have entered a phase of rapid development,and the number of social network users has surged.Various social network platforms have become important channels for the public to disseminate and obtain real-time information,express views and opinions,and exchange opinions.Because of the convenience of releasing information and the interactivity of spreading information,micro-blog has become the most important information dissemination platform bearing the evolution of public opinion.It is precisely because of these characteristics of Weibo that there have been frequent and largescale incidents of public opinion crises on Weibo in recent years.Especially after the covid-19 epidemic,Weibo has also become an important channel for the government to release information,and the management of public opinion has become particularly important.If the relevant parties can not take reasonable and effective measures to guide and control some major public opinion events,it will lead to the outbreak of public opinion,affecting social stability and the credibility of the government.The governance of public opinion needs to accurately identify and judge public opinion,and to explore the evolution of public opinion topics and public emotions on microblogs can identify the main topics and emotions of public concern.At the same time,it can identify the characteristics of the two changes over time,so as to provide effective guidance for the government’s public opinion regulation,which is of great practical significance.This paper makes an overall analysis of the typical evolution process of public opinion on microblogs.By establishing an analytical framework for the evolution of public opinion on the Internet,combining word segmentation techniques and Naive Bayes classifier,a model is established for sentiment analysis,and the emotional orientation of each text was calculated.On the basis of the above,the research calculate the emotional heat according to the number of comments and the variance of emotional tendency,and divide the stage of public opinion according to the emotional heat,so as to lead the emotional differences into the division of the stage of public opinion.The research process of this paper includes the static analysis of the evolution parameters of public sentiment,subject matter and emotional heat,and the dynamic study of the characteristics of each stage of public sentiment evolution from the perspective of time trend.This article analyzes the evolution of public opinion on microblogs based onevent that,it is difficult for the public to buy vegetables during the outbreak of the epidemic.First,the original research data were obtained from Sina’s microblogs,the text data are preprocessed by the methods of noise reduction,word segmentation,word deprecation and feature,and then the evolution of public opinion is studied by the combination of theme analysis and sentiment analysis.In the emotion analysis stage,the emotion tendency was classified and predicted based on the naive Bayes principle,the emotion tendency of each comment was obtained.Then the study explored the change of public sentiment with time,and calculated the emotional heat.According to the results of the calculation,public opinion events can be divided into four stages:high-heat period,recurrent period,duration period,extinction period and incubation period,diffusion period,recession period.Based on Relevance-based LDA topic model and topic mining method based on TF-IDF value,this paper analyzes the categories and evolution of public issues in public opinion events,the results showed that the main concerns were delivery block and fresh delivery capacity,material supply,government decision-making,community co-governance and fresh platform enterprises.By comparing and analyzing the public sentiment and theme evolution of two events,it is found that the public sentiment is negative in the whole public opinion event,and there are emotional differences in the time window and dynamic fluctuations in the time trend.Public discussion is interactive and pluralistic in different stages of public opinion.Based on the evolution of public sentiment and subject analysis,it provides a reference for the supervision and guidance of public opinion.At the same time,based on the public opinion evolution perspective,the empirical analysis results also provide new ideas and methods for fresh supply chain management.