RESUMO
Social polarisation processes have become a central phenomenon for the explanation of population behavioural dynamics in today's societies. Although recent works offer solutions for the detection of polarised political communities in social media, there is still a lack of works that allow an adequate characterization of the specific topics on which these divides between social groups are articulated. Our study aims to discover and characterise antagonistic communities on Twitter based on a method that combines the identification of authorities and textual classifiers around three public debates that have recently produced major controversies: (1) vaccination; (2) climate change; and (3) abortion. The proposed method allows the capture of polarised communities with little effort, requiring only the selection of some terms that characterise the topic and some initial authorities. Our findings show that the processes of social polarisation can vary considerably depending on the subject on which the debates are articulated. Specifically, polarisation manifests more prominently in the realms of vaccination and abortion, whereas this divide is less apparent in the context of climate change.