Detecting, Characterizing, Tracking and Forecasting Rare Events in Multi-Sourced Networks: An Application in Analyzing Isolated Malicious Activities

This project targets the challenging problem of combating diverse Isolated, Malicious Activities (IMAs), or "lone wolf" attacks, such as homegrown violent extremist attacks. Although IMAs seem completely isolated from one another, without outside direction or communication, IMAs share some common traits when examined under the proper lens (e.g., from the right/relevant data sources, in the subspace spanned by the right/relevant activities and/or at the right/relevant time steps). Thus, IMAs form a specific type of group, i.e., so-called "familiar strangers," where people with similar intentions, who do not have direct connections, may share common traits. Some known common traits include sympathetic discourse toward known terror organizations on public social networks or frequent visits to places with ties to known radical elements.