Knowledge Acquisition for Level II/III Fusion

"We propose to build on IET's existing product line of probabilistic and relational knowledge representation and reasoning tools to develop a knowledge elicitation environment that can be used by subject matter experts (SMEs) to generate knowledge bases ofprobabilistic relations for tactical battlespace applications such as Level II/III information fusion systems. An agent-behavior based "tactical model language" will be developed to guide SMEs in self-elicitation of battlespace entities and relations.Biased by the causal directionality of the elicitation, weak prior probabilities for uncertain relations can be estimated automatically. These can be strengthened by automated data mining if data is available; a facility to elicit qualitative probabilityestimates directly from SMEs will also be provided. There is potential for adoption of a probabilistic relational elicitation tool in multiple multi-billion dollar marketplaces, including those concerned with customer relationship management, naturalresource exploration, machine fault diagnosis and pharmaceutical discovery. This is correlated with the increasing awareness, on the part of both management and engineers concerned with large-scale systems development, of the need to engineer systems thatcan handle uncertain and incomplete inputs, support the acquisition of probabilistic relational knowledge from subject matter experts, and behave gracefully in the face of unexpected situati