Detecting Anomalies In Constraint-Based Systems

This paper addresses the problem of the automated verification of constraint-based systems, This issue has not been adequately covered in the research literature, as current expert-system verification methodologies and tools have largely focused on rule-based systems. As a result, these tools do not incorporate the methodologies necessary to verify constraint-based reasoning systems. However, there is great similarity as well as commonality of terms between rule-based systems and constraint-based systems, This suggests that verification techniques for rule-based expert systems can be successfully modified for application to constraint-based systems, The paper presents the authors\u27 experience in verifying the internal consistency of a constraint-based reasoning system knowledge base. The approach is based on automatically identifying several anomalies in a constraint-based system knowledge base without having to execute the system. Many of these types of anomalies are derived directly or indirectly from those that are also found in traditional rule-based expert systems, but new ones specific to constraint-based systems are identified. Detection methods and possible repair strategies are discussed, and a test example is described and evaluated. (C) 1997 Elsevier Science Ltd. All rights reserved