Use of Domain Knowledge Models to Recognize Cooperative Force Activities

This paper provides results of experiments on the use of domain knowledge models to recognize military groups, units, and activities from input track data. The group, unit and activity assessments are provided in the form of a situation estimate which is evaluated against ground truth. Our experimental architecture allows us to evaluate the quality of the resulting situation estimate as a function of the quality of the input track data. A situation assessment integrates and fuses low-level sensor reports to produce hypotheses at a level of aggregation of direct interest to a military commander. The elements of a situation assessment include 1) ̊hypotheses about entities of interest and their attributes, 2) ̊association of reports and/or lower level elements with entities of interest, and 3) ̊inferences about the activities of a set of entities of interest. In estimating the quality of a situation assessment, these elements may be scored at any level of granularity from a single vehicle to the entire situation estimate. Scoring involves associating situation hypotheses with the ground truth elements that gave rise to them. We have previously presented a method for scoring the quality of the situation estimate. In this paper we extend the previous results to the ability to score estimates of hierarchical groups / units and their activities. We discuss how our approach is instantiated in algorithms and describe results of experiments performed for DARPAÕs Dynamic Database program.