ARTIFICIAL INTELLIGENCE DIALECTS OF THE BAYESIAN BELIEF REVISION LANGUAGE

Abstract

Rule-based expert systems must deal with uncertain data, subjective expert opinions, and inaccurate decision rules. Computer scientists and psychologists have proposed and implemented a number of belief languages widely used in applied systems, and their normative validity is clearly an important question, both on practical as well on theoretical grounds. Several well-know belief languages are reviewed, and both previous work and new insights into their Bayesian interpretations are presented. In particular, the authors focus on three alternative belief-update models the certainty factors calculus, Dempster-Shafer simple support functions, and the descriptive contrast/inertia model. Important "dialects” of these languages are shown to be isomorphic to each other and to a special case of Bayesian inference. Parts of this analysis were carried out by other authors; these results were extended and consolidated using an analytic technique designed to study the kinship of belief languages in general.