Research On The Construction Of User Sentiment Evolution Map In Sudden Network Public Opinion

The frequent occurrence of online public opinion events in recent years has greatly affected the order of cyberspace and social stability,and brought a test to the government’s public opinion governance work.Network public opinion contains public perceptions and sentimental tendencies about public opinion events.Therefore,research on users’ sentiments under unexpected network public opinion has gained more and more attention.By exploring the evolution of users’ sentiments under unexpected network public opinion,we can provide references and suggestions for the government’s management of network public opinion.This study starts from the perspective of sentiment analysis and the evolution of network public opinion.Taking the "Tangshan beating incident",a hot topic on Weibo,as an example,the three-stage life cycle of network public opinion is divided into the outbreak period,the spreading period and the proliferation period according to the trend of network public opinion dissemination.The TF-IDF algorithm was used to extract keywords from the text of the comments,and the LDA topic model was used to analyze the topic evolution of public opinion.Against the Chinese sentiment lexicon of Dalian University of Technology,the new word sentiment features were manually marked and expanded to the sentiment lexicon ontology library,and then the fused sentiment lexicon was used to classify the comment data for sentiment.Combined with the sentiment classification results,user sentiment evolution maps were constructed from three dimensions,namely,the life cycle dimension,the geographic dimension and the community dimension of network public opinion,respectively.It is found that the network public opinions triggered by hot topics conform to the rule of network opinion life cycle evolution,and there are correlations among topics at different stages.Users’ sentiments are in negative sentiments throughout the whole opinion cycle,and there are differences in the evolution of users’ sentiments in different geographical areas,while the same community has similar sentimental tendencies,but negative sentiments still dominate.Based on the results of sentiment evolution,the following suggestions are made for the management of unexpected network public opinion: to make reasonable adjustments to the development of public opinion in different periods according to the time;to strengthen the joint management of public opinion coverage according to local conditions;and to actively guide the positive direction of users’ sentiments with a human-oriented approach.