Knowledge modeling of program supervision task and its application to knowledge base verification

This paper presents a knowledge-level analysis of the program supervision t ask based on two different systems: PEGASE and PULSAR. A knowledge-level an alysis of a knowledge-based system reveals the organisation of the knowledg e it uses and how it uses this knowledge to solve the task. It is also the key to determine the properties that it assumes about domain knowledge. The se aspects of knowledge-level analysis have been successfully used as a fra mework to compare different systems, mostly for knowledge engineering purpo ses. This paper also describes how domain knowledge assumptions have been e xploited in the implementation of a verification module for program supervi sion knowledge bases.