Research On The Topic Of Public Opinion Emotion Based On The LDA Model

In today’s Internet information age,which takes speed and efficiency as the standard,Sina Weibo,as a popular social media information exchange platform,provides instant communication and interaction services across time and space for network users.Therefore,netizens in the era of "we media" can timely comment on social public events and express their personal views,thus public opinion analysis arises at the historic moment.Major public events in society will form different emotional trends on the Internet after fermentation of time and heated discussion by netizens.Mining and analyzing them by using big data analysis technology is conducive to accurately guiding online public opinion to develop in a positive direction and improving the comprehensive governance ability of the government.In this paper,a research framework combining text sentiment analysis based on deep learning method and topic mining based on the LDA model is proposed to realize the evolution analysis of emotion theme of network public opinion on social public events.Firstly,deep learning neural networks are used to classify Chinese texts.Text＿CNN,Bi LSTM and CNN-BILSTM models are studied respectively.The optimal emotion classification model was screened by design experiment and result analysis.Then,in view of the characteristics of online media text data with high dimension,sparsity and unclear theme,especially sina Weibo comments text data are mostly short texts with sparse semantic features and insufficient co-occurrence information,this paper proposes the TF-COLDA model for topic mining.In order to solve the problem that the traditional LDA theme model is difficult to extract high quality and representative theme when dealing with social media platform web text.In the face of the global outbreak of the COVID-19 in 2020,this paper uses the above framework of network public opinion evolution analysis to conduct case studies,mining the hot topics and emotional expressions that netizens are concerned about at different stages of public opinion development,and mastering the direction of network public opinion and netizens’ attitudes about the COVID-19 on Sina Weibo platform.It is conducive to the relevant management departments to obtain the development trend of public opinion and the emotional changes of the people,so that corresponding countermeasures can be taken to maintain the normal social order and promote the stable development of society.