Commander’s Learning Agent (CLearn)

Traditional decision aid software requires the manual input of commanders intent in the field, which is impractical as it requires an excessive amount of time to learn the various facets of the commanders job, and requires an unrealistic adaptation capability as the commanders mission changes dynamically in the field. Hence, there is a need to automatically capture the commanders current mission, augment with contextual knowledge, and assign priorities to resources supporting the commanders mission. In Phase I, we researched the requirements for learning agents that passively model individual and system behavior within the context of Air and Space Operation Centers (AOC), and shown the feasibility of building an  agent that sits in between the user and application User Interface, passively monitors user behavior in the background, applies machine learning to discover the patterns in the background, and offers to assist the user in the real-time inference mode. Here we propose to research and develop a full scale prototype of a Commanders Learning Agent for AOC environment leveraging Phase I results, and evaluate the prototype using AOC operations personnel, and knowledge engineers with experience in cognitive modeling of AOC user tasks, and investigate the feasibility of transitioning the technology to commercialization. BENEFIT: We anticipate two major results from this Phase II effort: (1) development of the Web-user Interaction Agent Framework (WIAF), and (2) development of a demonstration application of an Adaptive Interface and Event Monitor for Real-time Collaboration in the AOC (CLearn). Commercial applications include personalization and automation applications packages as Web services.