Vitriol on social media: Curation and investigation

Abstract

Our online discourse is too often characterized by vitriol. Distinct from hate speech and bullying, vitriol corresponds to a persistent coarsening of the discourse that leads to a cumulative corrosive effect. And yet, vitriol itself is challenging to formally define and study in a rigorous way. Toward bridging this gap, we present in this paper the design of a vitriol curation framework that serves as an initial step toward extracting vitriolic posts from social media with high confidence. We investigate a large collection of vitriolic posts sampled from Twitter, where we examine both user-level and post-level characteristics of vitriol. We find key characteristics of vitriol that can distinguish it from non-vitriol, including aspects of popularity, network, sentiment, language structure, and content.