Research And Implementation Of A Public Opinion Analysis Technology Based On Hierarchical Attention Mechanism

In recent years,with the development of the Internet,everyone can release information freely on the Internet.Especially with the development of Weibo,it has become an important promoter in more and more major events,and all kinds of public opinions occur and spread on it every day.How to effectively and timely understand the development of public opinion,social views on public opinion events and then guide public opinion has become an important topic.Netizens’ various comments will contain a large amount of emotional information.In order to better understand the emotional tendency of online public opinions and track the changing trend of netizens’ emotions,this thesis comes up with an emotion analysis model based on hierarchical attention mechanism.Based on traditional deep learning,this model innovatively integrates external dictionary information and subject word information to improve the accuracy of sentiment analysis.On the basis of sentiment analysis,three typical public opinion events,the Winter Olympic Games event,Sino-US trade event and the Zhang Yuhuan event are analyzed in different ways.The Winter Olympic Games event and Sino-US trade event is analyzed from the perspective of topic evolution,that is,sub-topics are selected from the breaking news,and then the heat change,mutual evolution process and emotional tendency of the sub-topics are analyzed.The Zhang Yuhuan event is analyzed from different angles,namely from the perspectives of multiple parties,netizens,media and legal practitioners,so as to achieve the purpose of a comprehensive analysis of the event.Through analysis,managers can timely understand public attitudes,so as to better guide the development of public opinion and provide services for social governance.Finally,a public opinion analysis system is constructed to display the information such as heat trend and emotion trend in detail.