Processing of domain experts' knowledge for development of a KBS for risk analysis

Knowledge-based systems (KBSs) are being widely used for risk analysis of structures. The KBS makes use of domain experts' knowledge, which is an essential element of input for risk analysis. Acquired pieces of knowledge must be processed and improved with the help of domain experts before use for risk analysis. Utilization of the acquired knowledge without processing can reduce the quality of risk analysis. Processing of the knowledge elicited from experts is essential to identify the conflicting pieces of knowledge, gaps, and redundancies. Further, it is necessary to eliminate conflicts and represent the knowledge in a consistent and complete manner suitable for carrying out risk analysis. Processing of the knowledge elicited from different domain experts is a difficult task. A new approach using graph theoretic technique is proposed in this paper for the processing of knowledge and its use in the creation of a knowledge base. This new approach helps in developing an effective KBS for risk analysis of structures. To demonstrate the approach, an example problem of risk analysis of roof structure against damage due to cyclonic winds is considered. © 2001 John Wiley & Sons, Inc.