A Multi-Agent Multi-Tiered Approach To Information Fusion
Joseph Barker, Robert S. Woodley, Gil Vincent, Eric R. Lindahl · 2007
The amount of information available to decision makers today is astounding. To manage this, decision-aid software is needed that can organize and clarify large amounts of data. In this paper, we propose a multi-level, multi-agent architecture that accomplishes data fusion through layers encompassing initial data input and classification, secondary classification and grouping, and finally human-agent interaction. Small, limited-scope agents cooperate to complete the analysis/fusion process. This cooperation is mediated by corresponding opinions in the mode of belief calculus and subjective logic utilizing our evidential reasoning network (ERN). Thus, the user gains the ability to control and filter fusion activities based on the quality and pedigree of the underlying data. We present an initial set of agents along with a preliminary testing scenario and discuss the results of running the system through this scenario