The objective of Veracitas is to investigate the applicability of hybrid Belief Revision and Truth Maintenance techniques to increasing information trustworthiness in Situation Awareness databases. The specific problem for the DoD is that current Situation Awareness and Data Fusion applications are implemented in systems having no capability for data base veracity. These databases are populated by sources that often provide conflicting and illogical information as well as missing information. Once an operator attempts to fuse and understand data base information with conflicts and missing data, this can easily misdirect Situation Assessment reasoning processes to completely incorrect results within a Situation Awareness system, without the operator ever knowing what is happening. Conventional Truth Maintenance and Belief revision techniques have shown great promise in this area, but have not been widely applied because of computational complexity, the difficulty of developing accurate domain models, and brittleness. Veracitas proposes a novel hybrid approach and architecture for a practical solution to this problem.Veracitas will increase information trustworthiness in Situation Awareness databases by eliminating conflicts and missing data. This will eliminate the misdirection of Situation Assessment reasoning processes to completely incorrect results and possible newsworthy events. This technology can be applied to IT systems management to aid in Ballistic Missile Defense.
Keywords: Neural Net, Model Based Reasoning, Belief Revision, Truth Maintenance, Data Fusion, Situation Awareness, Database Veracity
