Thus, this study required a dataset of tweets related to COVID-19 published in 2020 on which aspect-based sentiment analysis could be applied. Chen et al.39 collected these data using 80 keywords, such as ‘Sars-cov-2’, ‘staysafestayhome’, ‘Coronials’, ‘Covid’, ‘pandemic’, and ‘Covid19’ to develop a repository associated with COVID-19. Classifying tweets related to COVID-19An initial challenge for this analysis was the presence of noise—that is, tweets that are unrelated to COVID-19. As test data, we utilized the free-form responses from a COVID-19 related survey conducted by our research team. Consequently, we utilized the Multinomial Naive Bayes classifier to distinguish COVID-19 tweets from non-COVID-19 tweets.

Source: CNN July 02, 2023 17:55 UTC