COVID-19 Concerns in US: Topic Detection in Twitter

Abstract

COVID-19 pandemic is affecting the lives of the citizens worldwide. Epidemiologists, policy makers and clinicians need to understand public concerns and sentiment to make informed decisions and adopt preventive and corrective measures to avoid critical situations. In the last few years, social media become a tool for spreading the news, discussing ideas and comments on world events. In this context, social media plays a key role since represents one of the main source to extract insight into public opinion and sentiment. In particular, Twitter has been already recognized as an important source of health-related information, given the amount of news, opinions and information that is shared by both citizens and official sources. However, it is a challenging issue identifying interesting and useful content from large and noisy text-streams. The study proposed in the paper aims to extract insight from Twitter by detecting the most discussed topics regarding COVID-19. The proposed approach combines peak detection and clustering techniques. Tweets features are first modeled as time series. After that, peaks are detected from the time series, and peaks of textual features are clustered based on the co-occurrence in the tweets. Results, performed over real-world datasets of tweets related to COVID-19 in US, show that the proposed approach is able to accurately detect several relevant topics of interest, spanning from health status and symptoms, to government policy, economic crisis, COVID-19-related updates, prevention, vaccines and treatments.