Guest Editor’s Introduction KBS Validation: Fronr lools to Methodology

K OWLEDGE-BASED SYSTEMS have outgrown academic applications, and their widespread use in industry has boosted work on the methodology required to design, implement, and validate them. Although knowledge acquisition was a prominent focus of research and practice from the beginning of this industrial push, only recently have the issues of the KBS life cycle and verification and validation been put at the same level of discussion. During the first years of KBS production, researchers thought of the KBS life cycle as rapid prototyping, while V&V was handled on an as-needed basis. Soon methods (and their support tools) were developed to detect problems in rule-based systems, such as redundant, subsumed, or missing rules. However, the field lacked a comprehensive view of KBS validation and the role of V&V in the KBS life cycle.’ European developers sensed this need and decided to address it at a workshop at the 1990 European Conference on AI, and through ESPRIT’s various KBS projects. It was in this context that the European Workshop on Verification and Validation of Knowledge-Based Systems (EuroVAV ‘91) took place at Jesus College in Cambridge, England, in July 1991. About half of the 58 delegates were from industry, and THE FZELD OFMVOWLEDGE-BASED SYSTEMSHASLACKED A COMPREHENS~VZEWOF~SVALZDATZONANDTHE ROLE 0~ V&J% THEKBSLZFE CYCLE. BYPRESENT~NG KEYWORKFROM EuRoVAV’~~, THZS SPECZAL SERIES HZGHLZGHTSANDEXAMZNES SZGNZFZCANTTXENDSZN THZS ZMPORTANTAREA.