Abstract
The COVID-19 pandemic is a worldwide catastrophe. In the absence of an effective drug, one effective measure to pull the pandemic to the end is herd immunity by taking vaccines, while the hesitation and anti-attitude from social media affect the vaccination. This makes it crucial to evaluate the text data about the COVID-19 vaccine from tweets. The period for data used in this study is 1 Aug to 31 Oct, 2020, since it is just before promoting the use when public reactions to the COVID-19 vaccine can influence their subsequent vaccination behavior. In this study, we used the latent Dirichlet allocation (LDA) topic model and sentiment analysis to explore public reactions to the COVID-19 vaccine. The results indicate that the public discussion could be divided into 11 topics, which could be further summarized into four different themes: (1) concerns about COVID-19; (2) concerns about vaccine development, production, and distribution; (3) how to control the COVID-19 before obtaining the vaccine; and (4) concerns about information of vaccine safety and efficacy. It can be concluded that to a large extent, public reactions to vaccines are dominated by positive sentiment. Specifically, the politicization of the vaccine approval process, suspension of vaccine trials, and measures to control COVID-19 tend to trigger negative public sentiment; whereas information related to successful vaccine development and availability enhances positive public sentiment. These findings help us understand public reactions to the COVID-19 vaccine, uncover potential factors that may influence vaccination behavior, and help policymakers understand public opinion about the COVID-19 vaccine and develop rational and effective policies.