Course-of-Action Forecasting

The Army Commander is currently overwhelmed with data and lacks techniques to turn this information into actionable intelligence. This "data rich, knowledge poor" environment requires the appropriate set of tools to parse the incoming data stream, identify concepts and their relationships, and turn it into knowledge - enabling the Commander to forecast the enemy's possible courses-of-action. The DAC Team proposes an Intelligent Agent solution to this problem. Our approach includes agents that perform natural language and ontology processing, spatio-temporal data mining, resource management and prioritization, and hypothesis management of possible courses-of-action through the use of Bayesian Networks. The approach is built on DAC's Agent Intelligence Platform (AIP), a distributed agent architecture that provides information fusion capabilities. This SBIR will extend AIP through the development of capabilities to support: spatio-temporal data mining, automated Bayesian Network management, and a "Value of Information" calculation providing the Commander with a measurement of the value of gathering additional information against a hypothesis versus acting on the current belief state.