Analyzing the relationship between relevance and extremist discourse in an alt-right network on Twitter

Abstract

Nowadays, Twitter is used by several political extremist groups to establish close communities on which the opinions are amplified following an echo-chamber effect. However, few literature analyses the effect of the use of an extremist discourse in relation to the relevance of these users on their online network. With the aim of analyzing this effect, this work studies the relationship between the use of indicators of extremist discourse from users belonging to an alt-right network on Twitter and their relevance on it. The network of alt-right users is created using the retweets of 96 accounts where the user relevance is measured by five different types of centrality metrics, including in-degree, eigenvector, k-shells, betweenness, and closeness. Both the linguistic indicators and the tone were analyzed using LIWC and VADER software. The network analysis outcomes show that user relevance on the network is indeed related to the use of an extremist discourse. Finally, this relationship is also tested on different corpus of texts and about different topics, being found that this relationship is more clear on retweets made by the users and when discussing about hate speech topics.