Public Opinion Analysis and Popularity Prediction for COVID-19 Hot Search Based on Weibo

Since the end of 2019, a new type of coronavirus pneumonia (COVID-19) has broken out in Wuhan, and various topics about the development of the epidemic have spread in full swing on the Sina Weibo. In this paper, the web crawler is used to capture the relevant Weibo and popularity of the hot searches during the COVID-19 outbreak, and the Weibo related to the epidemic are extracted by the Bayesian text classification method. Then, the potential Dirichlet model (LDA) was established to obtain the public opinion topic model, and ten public opinion topics were obtained to analyze the public opinion changes with the development of the epidemic. According to the topic model and the influence of daily time point on the popularity of Weibo, a multiple linear regression model is established to predict the popularity. Real-time analysis of changes in public opinion concerns provides reference for decision-making on epidemic prevention and control and information release.