A Situation Development Architecture and its Relation to a New Fusion Process Model

Information fusion is critical to the effective analysis of battlefield intelligence, but because it remains mostly un-automated it places high demands on the limited resource of skilled human analysts. We propose a principled approach to information fusion in the context of Situation Development that is grounded in formal models and methods. Situation development involves fusing information from battlefield sources in an attempt to understand what the enemy forces are doing or intending to do. What makes this problem challenging is its dependence on deep domain knowledge concerning battlefield characteristics, enemy unit composition/disposition, patterns of enemy activities, etc. along with an ability to abductively reason from partial information to plausible hypotheses. Phase I focused on demonstrating the feasibility of an ontologically-based automated reasoning approach to producing hypotheses of enemy activities; the fusion capabilities embodied in this approach were related to a formal process model representing an extension of the JDL model. Phase II will involve 1) the development of a functional prototype of a situation development assistant that will generate ordered lists of plausible hypotheses of enemy activity, 2) investigation of tools for developing domain knowledge and working with hypotheses, and 3) an extended process model for level 2 fusion. The transition path for this technology is through the Future Based Knowledge for the Future Force Program with its close associations with FCS and DCGS-A.