Extracting Physical Events from Digital Chatter for Covid-19

Abstract

By June 3, 2021, the US experienced over 33 million total cases of Covid-19, surpassing 592,000 deaths. In response, the Centers for Disease Control and Prevention (CDC) advised masking, social distancing and avoiding mass gatherings. In this work, we seek to automatically identify physical mass gathering events including dates and locations from digital chatter, i.e., social media data. We also study spread and sentiment associated with such large gathering events, finding a moderate negative correlation between large public gatherings, overall sentiment, and reported Covid-19 case numbers post event.