Dynamic, Graph-Based Risk Assessments for the Detection of Violent Extremist Radicalization Trajectories Using Large Scale Social and Behavioral Data

This project aims to produce an empirically-tested dynamic risk assessment protocol to anticipate the imminent risk of violence, and a computationally efficient tool based on this new protocol that enables law enforcement to mine, monitor, and screen for the occurrence of risk indicators in large law enforcement databases. The application builds on and expands a previously NIJ-funded database of overt behavioral indicators associated with jihadist inspired terrorists.

The proposed study will enhance the existing database by adding a wider range of overt behavioral indicators and new cases since 2015. Machine learning will be applied to the enhanced database of behavioral indicators to develop algorithms for the dynamic risk assessment tool. This tool will be designed to integrate with existing law enforcement databases and sources to monitor and screen for in near real-time those individuals who pose significant risk for violent terrorism. This approach is considered machine learning because the probabilities and patterns are learned and refined using baseline data sets from validated terrorism cases. This set of learned probabilities/patterns is then used to on new law enforcement data sets to produce probability scores that indicate a potential violent terrorist.

This project contains a research and/or development component, as defined in applicable law.

CA/NCF