A Polarization Approach for Understanding Online Conflicts in Times of Pandemic: A Brazilian Case Study

Abstract

As society becomes digitalized, online social networks tend to be primary places for debate but can turn into a battlefield for imposing conflicting narratives. Automating the identification of online conflicts is a challenge due to difficulties in defining antagonist communities and controversial discussions. Here, we propose a polarization approach for understanding Twitter conflicts in Brazil during the COVID-19 pandemic, where a small group of polarizers influences a larger group of polarizees according to their ideological leaning. Polarizers are automatically identified by centrality metrics in following, retweet, and reply networks, and manually labeled as leftists, rightists, or undefined. We collected and analyzed the polarization of 21 potentially conflicted political events in Brazil. Our results show that polarizers adequately represent the polarization of events, the traditional media is giving way to a new breed of tweeters, and retweet and reply play different roles within a conflict that reflects their polarization level.