An axiomatic foundation of Dempster-Shafer theory
Chun-Hung Tzeng · Proceedings of the 17th conference on ACM Annual Computer Science Conference · 1989
This abstract introduces an axiomatic study on Dempster-Shafer theory of handling uncertainty in AI systems. A Dempster-Shafer belief function, based on probabilities, is intended to measure the degree of support a piece of evidence provides for various propositions in its domain. Dempster's rule provides a method of combining evidence that may justify partial beliefs. Shafer argued that the Bayesian theory is a special case of this theory and that Bayes rule of conditioning is a special case of Dempster's rule of combination. However, the development of evidence models, including Dempster's rule, is based on intuition and lacks the axiomatic foundation of probability theory.