Handbook of Computational Approaches to Counterterrorism

pp 99-127


SOMA: Stochastic Opponent Modeling Agents for Forecasting Violent Behavior

  • Amy SlivaAffiliated withNortheastern University Email author 
  • , Gerardo SimariAffiliated withDepartment of Computer Science, University of Oxford
  • , Vanina MartinezAffiliated withDepartment of Computer Science, University of Oxford
  • , V. S. SubrahmanianAffiliated withUniversity of Maryland College Park

* Final gross prices may vary according to local VAT.

Get Access


The modern global political environment is growing increasingly complex, characterized by webs of interdependency, interaction, and conflict that are difficult to untangle. Technological expansion has led to an explosion in the information available, as well as the need for more sophisticated analysis methods. In this security and information environment, behaviors in the domain of counterterrorism and conflict can be understood as the confluence of many dynamic factors—cultural, economic, social, political, and historical—in an extremely complex system. Behavioral models and forecasts can be leveraged to manage the analytic complexity of these situations, providing intelligence analysts and policy-makers with decision support for developing security strategies. In this chapter, we develop the Stochastic Opponent Modeling Agents (SOMA) framework as a stochastic model of terror group behavior, presenting several scalable forecasting algorithms and a methodology for creating behavioral models from relational data.