Testing Knowledge Based Systems: A Case Study and Implications

Abstract Knowledge engineering and risk analysis have become increasingly important for safety-related computer systems. The first offers methods to cope with vague, uncertain, and badly structured tasks, the second attempts to identify the extent and likelihood of the consequences associated with hazards. The methods and tools proposed in these approaches must themselves be reliable. Based on a case study in testing a transnational project - a knowledge based risk management system (KRM), the authors analyze the impact of the specific features of knowledge based systems on testing, propose a test design and identification scheme for prototyping development, investigate documentation, use of standards, and quality management in knowledge based system projects, and consider tasks necessary to improve the quality of development and use of risk analysis software.