Confidence, pedigree, and security classification for improved data fusion

Automated data fusion can reduce the burden on intelligence analysts who can be overwhelmed by available data. Automation can also reduce delivery time of intelligence products to users. Users of fused intelligence only see finished products, and have little insight into the sources that contributed to those products. Such insight is generally not needed or desired, if the users trust the underlying process. To gain that trust, automated data fusion systems require knowledge of the quality (confidence) and pedigree (source and heritage of previous processing) of all data processed. This paper discusses the design and prototype of a standard representation of confidence, pedigree, and security classification information known as confidence encapsulated atomic data. Current research is focused on solving issues of data overhead and throughput. The paper describes the need for and benefits of the standard, reviews the prototype, discusses results of performance tests, and introduces concepts for future research.