Echo chamber detection and analysis A topology- and content-based approach in the COVID-19 scenario

Social media allow to fulfill perceived social needs such as connecting with friends or other individuals with similar interests into virtual communities; they have also become essential as news sources, microblogging platforms, in particular, in a variety of contexts including that of health. However, due to the homophily property and selective exposure to information, social media have the tendency to create distinct groups of individuals whose ideas are highly polarized around certain topics. In these groups, a.k.a. echo chambers, people only hear their own voice, and divergent visions are no longer taken into account. This article focuses on the study of the echo chamber phenomenon in the context of the COVID-19 pandemic, by considering both the relationships connecting individuals and semantic aspects related to the content they share over Twitter. To this aim, we propose an approach based on the application of a community detection strategy to distinct topology- and content-aware representations of the COVID-19 conversation graph. Then, we assess and analyze the controversy and homogeneity among the different polarized groups obtained. The evaluations of the approach are carried out on a dataset of tweets related to COVID-19 collected between January and March 2020.

Automated summary

Social media play a significant role in fulfilling social needs and serving as essential news sources, but they also have the tendency to create highly polarized groups of individuals around certain topics. This article focuses on studying the echo chamber phenomenon during the COVID-19 pandemic, considering relationships between individuals and the content they share on Twitter. The approach involves applying community detection strategy to distinct representations of the COVID-19 conversation graph and analyzing the controversy and homogeneity among polarized groups.