TYPE ORIENTED APPROACH TO KNOWLEDGE-BASED SYSTEMS.

A type-oriented programming (TOP) approach to building knowledge-based problem solving systems is developed. A problem solving system is considered as a system of knowledge elements. Each knowledge element class consists of a set of knowledge items, possibly inherited along class structures. Both elements and their items are given knowledge types. In TOP, signals are processed by interpretations which are signal and receiver type-specific functions. Types can be active (operator), passive (value) or context creating (knowledge organizing) in nature. Using the TOP approach, a multilevel context architecture for modelling and solving problems of high knowledge intensity is proposed. TOP is compared with a number of well-known knowledge-based programming paradigms and tools. For instance, the if-then rules are considered as a special case of representing active knowledge.