Using an Ontology for Knowledge Acquisition

Abstract

We describe an approach to distributed knowledge acquisition using an ontology. The ontology is used to represent and reason about soldier performance. These methods are embedded in an immersive web based knowledge elicitation and analysis system. Background Advanced computer simulations are used for training (Haimson, 2005), mission rehearsal, planning, war-gaming, and research, to name a few. Recently, there has been great interest in using simulations to evaluate and analyze systems and technologies for the purpose of decreasing development and testing costs, evaluation of human-machine systems, and on improving system acquisition decisions. As new soldier technologies are being developed, the need for realistic behavior models has increased in order to fully assess the value of these new systems. However, soldier performance data during military operations in urban terrain (MOUT) are very limited. Techniques to obtain MOUT performance data is critically needed for modeling and analysis. Constructive computer models that simulate close combat in MOUT environments require representative human performance data to drive the development of autonomous agents. These agents simulate behavior within the model with the goal of mimicking human decision making activities and allow a better understanding of the relationship between system design, human cognition, and behavior. The inherent problem is significant gaps in data for generating models with enough fidelity for realistic human behavior representations (HBR). Approach To support ongoing efforts to improve these models, we sought to identify, develop, and validate techniques to collect and utilize soldier performance data. Simulations only take into account a fraction of the perceptual cues within an environment and a small number of behavior