Abstract:
The potential gained from "tapping" the knowledge of domain experts in war-gaming and decision-modeling has yet to be fully realized in the Intelligence Community. While there may be a number of reasons for this, perhaps chief among them is the cost of involving human experts, and the belief that intelligence based on expert opinion is poor quality and unreliable. To this end, we are working on a middleware platform, called Collaborative Panel Administrator (CPA). CPA is specifically designed to support virtual panel management and asynchronous data collection. It also allows for data aggregation and imputation to facilitate incremental validation of intelligence models. To maximize objectivity of panelist input, the CPA panel management facilitates blind collaboration, in which identities of panelists in the same panel can selectively be hidden from each other, even while working on the same intelligence model, thus reducing the effect of group think and adverse social dynamics. CPA mainly works as a data hub between intelligence modeling tools and corresponding model validation services. In this paper, we discuss in detail motivation and design of the CPA. Where appropriate, we also discuss how distributed agents would provide an effective means of realizing some of CPA functionalities.