Assessing confidence in Situation Awareness

Situation Awareness enables the discovery of aggregations and the identification of interesting patterns in underlying data that can be leveraged to further the understanding of the battlespace. While there have been steady efforts within the information fusion community to increase the level of automated reasoning supporting Situation Awareness, there remain unresolved issues such as the development of standard metrics of trust. Aggregation methods applicable to one sort of analysis may not be parameterized in the same fashion as another. New methods may be introduced. Together, these belie the possibility of a universal a-priori understanding of the factors that may temper a method's reliability. To accommodate such variability, this paper adopts a non-parametric approach to the assignment of a confidence metric. It introduces a measure similar to Hubert's Γ but which incorporates a measure previously shown helpful in assessing the effectiveness of object refinement engines. Results illustrate its application.