Improving Information Fusion Capabilities through the Explicit and Pervasive Consideration of Uncertainty

A variety of forces have led to decreases in the average experience of intelligence analysts. This shift is an incredibly important development since a lack of experience has been strongly linked to: failures in locating key information sources; theincorporation of faulty information in reports; and the premature closing of investigations. To help mitigate these impacts and generally improve intelligence analysis capabilities we propose an approach to information fusion that employs an integratedprobabilistic framework in order to explicitly account for disparate indicators of uncertainty and thereby improve human decision making. The use of this integrated framework, in which information extraction and fusion decisions are made with a commonpool of evidence and inference procedures, will also allow much richer forms of inference than possible with the current state of the art technologies. To further extend the reach of the proposed system, Assure, to higher level fusion tasks, we willemploy a mixed-initiative interaction paradigm that will enable a form of cooperative information fusion and uncertainty analysis. Phase I research and development of a limited prototype will provide a solid foundation for the complete implementation ofAssure in Phase II and its eventual commercialization.