Proceedings of the 8th International Conference on Information Fusion, Philadelphia, PA (July 2005)
A Network-based Truth Maintenance System (NTMS) is presented that leverages Bayesian Belief Network (BN) technology for its truth maintenance mechanism. NTMS applies probabilistic versions of integrity constraints in the form of small BNs to the knowledge base, providing an efficient and intuitive mechanism for detecting inconsistencies. The approach is applicable to any relational database by converting database entries into evidence for the probabilistic constraints. NTMS can optionally integrate BNs for Level Two fusion (Situation Assessment, or SA) with the probabilistic constraints; the instantiation for a particular problem consists of a SA BN that models the world state and facilitates SA on the knowledge base. The SA BN is augmented with the probabilistic constraint BNs, which provide metrics on the consistency of the SA BN. Inconsistencies in the SA BN are resolved by a combination of techniques including sensitivity analysis, default reasoning, consistency “forcing,” or seeking new evidence.
For More Information
(Please include your name, address, organization, and the paper reference. Requests without this information will not be honored.)