Fan Up The Flames: Mining Aggregation Patterns Of Cyber Political Trolls And Predicting Information Cascade Through Dynamic Rewiring Mechanism

Advances in information technology empowers cyber political trolls' use of social media to attack targeted opponents and win political arguments. Preventing vicious polarization and exacerbation of the general public is crucial, thus it is essential to understand how these trolls launch strategic moves and how their interactions at the aggregate level create information cascade. This study analyzes how different types of cyber political trolls proceed coordinated attacks in their argument promotions and how they accomplish so over time. We propose a data-driven method to identify aggregate patterns of cyber trolls and analyze how different categories of trolls work together to foster intended political arguments. Our method leverages graphic knowledge representation to predict information cascade by modeling the rewiring dynamics of interactions among trolls. This method supports timely detections of misinformation before it creates negative social impacts, and this helps social media prevent dissemination of false consensus about social-political events.