Verification and Validation of Knowledge-Based Systems
Papers from the AAAI Workshop
Daniel O'Leary and Alun Preece, Cochairs
AAAI-98 marks the eleventh anniversary of the first AAAI V&V workshop. Since then, a considerable amount of progress has been made in defining effective verification & validation (V&V) techniques for knowledge-based systems (KBS), integrated within KBS development methodologies. AAAI V&V workshops have been instrumental in bringing academic researchers together with industrial practitioners and have been a major focus for the KBS V&V community's research efforts.
While many problems have been solved in the last decade, there remain many unresolved problems, and the evolving nature of KBS technology generates new V&V challenges. Many of the issues involved in V&V of rule-based systems are now well understood, but few modern KBS are purely rule-based. Current challenges lie in V&V techniques for complex, hybrid knowledge-representation and reasoning systems. Increasing adoption of distributed KBS technology (for example, multiple-agent systems, distributed enterprise knowledge bases, and web-based KBS) have raised new--and hard-- V&V requirements.