Interval logic for expert systems

Present models of handling uncertainty in expert systems are based mainly on probability theory (Bayesian inference) and Zadeh's fuzzy approach. This paper adds a new model for inexact knowledge representation and reasoning under uncertainty. The proposed logic system is based on the assumption that the logical scale is not discrete (as in multi-valued logic systems) and the degree of truth/falsity could be presented in the form of an interval value. In spite of the basic definitions of the logical notions and operations, several examples are given of formalizing verbal expressions to interval clauses and further reasoning with them. However, this paper does not cover some particular models for rule-based expert systems and inference algorithms developed on the basis of interval logic. These items, together with the results of their implementation, could be found in related works mentioned in the reference section.