Developing a strategy for expert system verification and validation

Abstract

Research and practice has produced numerous methods for expert system verification and validation (V&V) that augment traditional software and systems approaches to V&V. This paper shows how to develop a strategy for expert system V&V. A strategy for V&V has three components: 1) the criteria by which an expert system will be judged as valid, 2) a life cycle model that specifies what V&V can be done when, and 3) the constraints (and opportunities) imposed by the characteristics of the system being developed. Starting with a development methodology, we shown how to map V&V methods onto the software life cycle and then match V&V methods to system characteristics. We give an example of our approach to developing a V&V strategy in the context of an expert system development project.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>