ABSTRACT
In 2020, COVID-19 became one of the most critical concerns in the world. This topic is even still widely discussed on all social networks. Each day, many users publish millions of tweets and comments around this subject, implicitly showing the public's ideas and points of view regarding this subject. In this regard, to extract the public's point of view in various countries at the early stages of this outbreak, a dataset of Coronavirus-related tweets in the English language has been collected, which consists of more than two million tweets starting from 23 March until 23 June 2020. To this end, we first use a lexicon-based approach with the GeoNames geographic database to label each tweet with its location. Next, a method based on the recently introduced and widely cited Roberta model is proposed to analyse each tweet's sentiment. Afterwards, some analysis showing the frequency of the tweets and their sentiments is reported for each country and the world as a whole. We mainly focus on the countries with Coronavirus as a hot topic. Graph analysis shows that the frequency of the tweets for most countries is significantly correlated with the official daily statistics of COVID-19. We also discuss some other extracted knowledge that was implicit in the tweets. © 2022 World Scientific Publishing Co.