Belief maintenance: an integrated approach to uncertainty management
Kathryn Blackmond Laskey, Paul J. Lehner · National Conference on Artificial Intelligence · 1988
Belief maintenance represents a unified approach to assumption-based and numerical uncertainty management. A formal equivalence is demonstrated between Shafer-Dempster belief theory and assumption-based truth maintenance extended to incorporate a probability calculus on assumptions. Belief propagation through truth maintenance automatically and correctly accounts for non-independencies among propositions due to shared antecedents. Belief maintenance also incorporates an ability to represent and reason with defaults. The result is a framework for non-monotonic reasoning about the application of a quantitative uncertainty calculus.