Temporal Nuances of Coordination Networks

Current network-based methods for detecting coordinated inauthentic behaviour on social media focus primarily on inferring links between accounts based on common"behavioural traces"[19], such as retweeting the same tweet or posting the same URL. Assuming the goal of coordination is amplification, boosting a message within a constrained period, most approaches use a temporal window to ensure the co-activity occurs within a specific timeframe [9, 14, 19, 24]. Real-world application requires considering near real-time processing, creating performance requirements, which also highlight gaps in the semantics of coordination in the literature. These methods could all exploit temporal elements of coordinated activity. We describe preliminary research regarding coordination network semantics, coordination network construction, relevant observations in three political Twitter datasets and the role of cheerleaders in revealing social bots.