Comparison of Fuzzy-based Value of Information to Bayesian Inference in a Military Domain

Determining the relative importance of vast amounts of individual pieces of information presents a significant challenge in the military intelligence domain. Using Subject Matter Expert (SME) knowledge to create decision support tools requires aggregation of the various experts' knowledge. The goal of this research endeavor is to investigate a Bayesian inference model for aggregating opinions of military intelligence analysts with respect to the Value of Information (VoI) problem. This paper discusses ongoing VoI research and presents results from an experiment using the Bayesian Thurstonian ranking aggregation model. The Thurstonian rankings are compared to the “ground truth” as generated by the current fuzzy-based VoI prototype system. Results demonstrate the usefulness of the Thurstonian method in aggregating SME opinions and clearly demonstrate the well-know “wisdom of the crowd” effect. Implications related to ongoing VoI research are discussed along with future research plans.