Decision tables as a knowledge representation formalism

Abstract

A knowledge engineering tool called GESS (General Expert System Shell) is presented which is used for building different types of expert systems. Knowledge representation is based on direct extensions of the idea of decision tables. It was found that this formalism reduces the knowledge acquisition bottleneck by assisting the knowledge engineer in the structuring of knowledge because it enforces completeness and allows for automated checking of contradictions and redundancies. It was also found that this formalism is quite flexible in expressing knowledge and implementing different reasoning paradigms as is discussed in this paper. Complex systems have been built with Gess using its capability of creating networks of decision tables and the accessibilty of the external computer environment from within Gess.