Using a Matcher to Make an Expert Consultation System Behave Intelligently

This paper describes how even a simple matcher, if it can detect simple relationships between statements in the knowledge base, will support many features that will make a knowledgebased consultation system appear to behave more intelligently. We describe three features that are useful during the knowledge acquisition phase (involving the building and testing of the knowledge base), and four features that are of assistance during a consultation. Although these features are described in terms of the Prospector environment [2], it will be clear to the reader how these features can be transferred to other environments.