Topic Modeling And Sentiment Analysis Of Internet Public Opinion In Public Health Emergencies Based On Public Opinion Analysis Of "COVID-19"

Internet public opinion communication in public health emergencies has always been a hot research topic.Because the network public opinion in the sudden crisis situation is often mixed with a large number of emotional expressions and public emotions,if it can not be properly guided,it will inevitably bring a severe test to the emergency management and social reconstruction in the event.Therefore,it is the proper meaning of the current research to pay attention to the network public opinion communication in public health emergencies from the perspective of theme characteristics and emotional attributes.The study takes the "Corona Virus Disease 2019" epidemic,a typical public health emergency,as an example.First,by combining the two major search engines,Google Trends and Baidu Index,to conduct communication analysis,we obtained the range of the outbreak period that most attracted the attention of netizens in the early stage of the new crown pneumonia epidemic.The analysis data is obtained through network text collection.Then,the research method of combining LDA topic modeling and corresponding topic emotion analysis is adopted to carry out topic feature analysis and emotion analysis on the research data.Ten themes,such as "epidemic situation" "netizens’ prosocial behavior" "cluster cases" and "mourning" are obtained through theme exploration,which are the main theme types of public opinion field in public health emergencies.Through the secondary evaluation of these ten topics,six cluster analysis directions of strong relationship are obtained,and the social reasons behind these six cluster analysis directions are analyzed.At the same time,according to the results of topic analysis,this paper analyzes the overall public opinion sentiment and the classified public opinion sentiment under each topic based on the application of network sentiment dictionary.