Tracking the Digital Traces of Russian Trolls: Distinguishing the Roles and Strategy of Trolls On Twitter
share
Online trolling has raised serious concerns about manipulating public opinion and exacerbating political divides among social media users. In this work, we analyse the role and behaviour of Russian trolls on Twitter through the lens of the social theory inspired by Tarde's ancient theory of monadology and its further development in Actor-Network Theory. Based on what the theory suggests, the social role or identity of individuals should be defined based on the traces they leave behind, which we define as a succession of timestamped items (e.g. tweet texts). To operationalise this, we develop a novel variant of text distance metric, time-sensitive semantic edit distance, accounting for temporal context across multiple traces. The novel metric allows us to classify roles of trolls based on their traces, in this case tweets, against a reference set. Through experiments aiming to identify the role of trolls into one of left-leaning, right-leaning, and news feed categories, we show the effectiveness of the proposed metric to measure differences between tweets in a temporal context. Through the qualitative analysis of tweet visualisation using the semantic edit distance, we find: (1) cooperation between different types of trolls despite having different and distinguishable roles; (2) complex and orchestrated interplay between left- and right-leaning trolls; (3) multiple agendas of news feed trolls to promote social disturbance and make Western democracy more fractious; and (4) a distinct shift of trolling strategy from before to after the 2016 US election.
READ FULL TEXT