Developing Graph Theoretic Techniques to Identify Amplification and Coordination Activities of Influential Sets of Users

Mustafa Alassad, Muhammad Nihal Hussain, Nitin Agarwal

Abstract: Social media, with its accessibility and anonymity, has helped malicious actors to thrive and coordinate several campaigns. Such users successfully utilize social media to coordinate different kinds of movements that could influence political aspects, damage the crucial infrastructure and affect the economy of several countries around the world. Malicious users could coordinate to cripple the transportation system by closing the main highways and bridges in big cities or spreading false security information that causes panic and hysteria in large societies. Since the traditional community detection methods fall short in finding these users, our research proposes an integrated model to find, analyze, and suspend these coordinated malicious sets of users in online complex networks. The Focal Structures Analysis model is a two-level analysis to study individual-level features using closeness centrality and group-level features by implementing the spectral modularity method. The model decomposes the interactions between both individual-level and group-level to find key sets of users that are responsible for propagating behavior through online social media platforms. The proposed model is applied to a fake news YouTube co-commenter network. The outcomes were validated via modularity methods and depth-first search to measure each set’s influence at individual-level and at the entire network-level.