Analysis Of The Evolution Of Online Public Opinion Emotions And Themes In Sudden Public Health Incidents

Since the COVID-19 broke out in 2019,public health emergencies have become the focus of social attention.Unlike other emergencies,public health emergencies will pose a serious threat to human life and social order,often accompanied by characteristics such as uncertainty,destruction,and focus,which can easily attract widespread public attention and participation.At the same time,social media,as an important channel for public information exchange,emotional expression,and maintaining social relationships,has promoted the formation of an online public opinion environment.This thesis takes the social media data of COVID-19 as the research object,explores the evolution of online public opinion on public health emergencies from the perspective of emotional analysis and theme mining,reveals the focus issues and emotional trends of public opinion users,and provides methods and decision-making references for the government and relevant departments to understand public opinion,master public opinion situation,and scientifically guide online public opinion.First,take "epidemic situation" and "COVID-19" as keywords,use Python programming to crawl relevant comment forwarding data in Sina Weibo platform,and preprocess the data and predict emotional orientation.In terms of emotional tendency prediction,this article fully utilizes the advantages of large-scale pre trained language models and graph convolutional network(GCN)models to propose a Chinese sentiment classification method based on Chinese-BERT-wwm-GCN,and compares this method with four baseline models: Text CNN,Bi LSTM,Text GCN,and Chinese-BERT-wwm.The experimental results show that the overall performance of the sentiment classification model in this thesis is superior to the baseline model mentioned above,and it is then applied to predict the emotional tendency of network public opinion in sudden public health events.On the basis of the above work,this thesis analyzes the evolution of online public opinion of COVID-19 epidemic from the perspective of theme evolution and emotion evolution.Firstly,based on the theory of public opinion lifecycle and temporal evolution trends,public opinion events are divided into four stages: initiation,outbreak,recurrence,and regression.Secondly,combining with the LDA topic model,we build a public opinion user theme map,mine the hot topics of public concern,and effectively reveal the distribution characteristics of user themes in different stages of public opinion development.Furthermore,based on the aforementioned Chinese-BERT-wwm-GCN sentiment classification method,sentiment tendency prediction is performed on collected text data.Combining with the forwarding and commenting relationships of public opinion users,a sentiment evolution graph of public opinion users is constructed,dynamically displaying the emotional evolution patterns of public opinion users at each stage,providing reference suggestions for network public opinion response and guidance in public health emergencies.From the research results,in the public opinion event of the COVID-19,the overall emotional attitude of public opinion users tends to be negative,and the emotional distribution is uneven in the public opinion life cycle.Among them,in the initial stage,public opinion users paid high attention to the national epidemic prevention and control policies,and the proportion of users with this theme was about 40%.The distribution of user emotions is relatively average.At the outbreak stage,the public paid more attention to the early symptoms of COVID-19 infection and the impact of COVID-19 outbreak on personal life.Most public opinion users exhibit negative emotions,which continue to intensify as the public opinion lifecycle evolves.In the iterative stage,the public opinion user group pays attention to a wide range of topics,and the proportion of each topic group is relatively average.The netizens’ emotions on topics such as the late symptoms of COVID-19 infection and the prevention and control of the epidemic during the Spring Festival fluctuated greatly,and negative emotions accounted for the highest proportion at this stage.In the regression stage,the public’s focus shifts to exploring the effectiveness of national epidemic prevention measures,restoring social order,and a positive attitude towards overcoming the epidemic.The emotional distribution of public opinion users is relatively average and shows repetitive fluctuations.Finally,based on the above empirical results,propose response and guidance strategies for public opinion regulatory authorities on how to respond to online public opinion in public health emergencies from the perspectives of public opinion themes and emotions.Firstly,in the face of sudden public health emergencies,new technologies such as statistical analysis and data mining should be fully utilized to enhance the precise identification and early warning capabilities under various themes of public opinion,improve public opinion response mechanisms for different themes,actively guide public opinion direction,and strengthen information disclosure and sharing of public opinion events.Secondly,in terms of emotional guidance,it is necessary to make full use of network data and text mining methods to accurately identify emotional tendencies in public opinion events,prioritize the handling and guidance of negative emotions among netizens,and expand the guiding role of positive emotions among public opinion users.It is also important to advocate for multi-agent participation and pay attention to the guidance and cultivation of healthy emotions among netizens.