This proposal describes a specific problem within the Intelligence community. Because of a lack of GEOINT tools for capturing storing and delivering expert knowledge from experienced analysts, new analysts have difficulty in quickly gaining the level of skill that those expert analysts possess. Recent work in cognitive science has shown that engineers presented with knowledge from domain experts in certain areas of expertise were able to perform tasks better after studying concept maps. This proposal suggests that an even more robust method would be to use conceptual graphs as a means to capture store and deliver domain expert knowledge for rapid training of new analysts. We show a design approach based upon the CORE toolset, an award winning set of Government owned cognitive tools for manipulating conceptual graphs. We also present a preliminary design as a baseline for the work to be accomplished in Phase I and describe our success over the years in the SBIR program.
Benefit: If successful, the Phase I program will lay the groundwork for full-scale development of an expert domain knowledge capture storage and delivery system for the Intelligence community. The Phase I program should be able to identify those major knowledge-handling functions necessary to perform these tasks and to demonstrate some of their functionality. The Phase II program will fully develop and test the individual components and resolve key technical issues before going on to Phase III in which institutional funding will be sought to build a formal software architecture for corporate training.
Keywords: Cognitive Reasoning, Neural Networks, Domain Knowledge, Conceptual Graphs, Semantic Networks, Concept Maps
