Research on Sentiment Analysis Model Due to COVID-19 Using Social Media

Abstract

Since the outbreak of the disease in early 2020, it has received worldwide attention and become a hot topic of discussion on social media. This paper crawled more than eight billion tweets from the first inflection point of the global outbreak of the epidemic, extracted several prominent hot topics related to the epidemic through two objective and intuitive means, namely LDA model and generated words, and then compared and analyzed the degree of subjectivity and positivity of their tweets for 'lockdown', 'mental health' and other topics. The experimental results showed that the number and active degree of global Twitter users' tweets on the above epidemic hot topics were most correlated with the number of newly diagnosed patients after 12 days. Comments are generally more positive to lockdown than to mental health. For all tweets about the epidemic in India and Wuhan, the positive degree of tweets judged as objective was relatively stable, while the positive degree of subjective tweets fluctuated greatly, which verified the rationality and effectiveness of the model for subjective and objective classification. Among the tweets judged as subjective by the model, the positive component of sentiment analysis was more. Most tweets about the epidemic in India and Wuhan were positive, and the fluctuation degree of subjective and objective curves verified the rationality and effectiveness of the model for subjective and objective classification.