Constraints and Goals under the Conceptual Graph Formalism: One Way to Solve the SCG-1 Problem

Abstract

The Sisyphus-I problem is a room allocation problem used as a common testbed to compare different knowledge acquisition and problem solving methodologies. This paper shows how it can be represented and solved using conceptual graphs (CGs). Since CGs offer a graphical representation of knowledge for, among other things, facts, constraints and goals, this paper shows how the subsumption relation defined on CGs help reformulate the problem in terms of a classification problem. It also shows how a graphical representation of this classification structure can be helpful in the generation of explanations pertaining to the behavior of the system, or as support to a knowledge engineer who must assist the system in its task. Thus we claim that such a classification structure, which can be visualized, may contribute: 1) to solving the problem, 2) to keeping track of the behavior of the system, and 3) to interpreting possible solutions.