Detection of echo chambers in Reddit

Abstract

In today’s technology climate, people can easily connect with anyone in the world, and to quickly access any kind of information online. While this is beneficial, it does come at a great cost as well. When given the freedom to express their thoughts, people usually concur with views that are consistent with their own, ignoring and dismissing other views that are different. This results in the phenomenon called “echo chamber”, in which people are trapped in an enclosed chamber where their own ideology is reinforced, or echoed again and again.

Echo chambers can be detrimental as it diminishes mutual understanding between groups of people with different ideologies, causing social tension. Thus, it is more important now than ever, to know how to detect echo chambers. However, there are limited studies and research done in this domain.

This report presents a novel and structured approach in detecting echo chambers in a widely popular social media platform named Reddit. It is a hybrid approach consisting of topological and semantic aspects, which makes use of graph community and stance detection respectively, to detect echo chambers.