Coordinated Information Campaigns on Social Media: A Multifaceted Framework for Detection and Analysis

The prevalence of coordinated information campaigns in social media platforms has significant negative consequences across various domains, including social, political, and economic processes. This paper proposes a multifaceted framework for detecting and analysing coordinated message promotion on social media. By simultaneously considering features related to content, time, and network dimensions, our framework can capture the diverse nature of coordinated activity and identify anomalous user accounts who likely engaged in suspicious behaviour. Unlike existing solutions that rely on specific constraints, our approach is more flexible as it employs specialised components to extract the significant structures within a network and to detect the most unusual interactions. We demonstrate the effectiveness of our framework using two Twitter datasets, the Russian Internet Research Agency (IRA), and long-term discussions on Data Science topics. The results demonstrate our framework's ability to isolate unusual activity from expected normal behaviour and provide valuable insights for further qualitative investigation.