Modeling Risk from Large-Scale Disasters by Integrating Fragmented Knowledge: Decision Tool

Abstract

Risk problems are treated in different ways in different disciplines and by different experts. Field experts’ knowledge has often been used in solving complex problems of risk from large-scale disasters. A storehouse of expert knowledge and data is available for windstorm-induced damage to roof structures, and in this paper we use this knowledge to illustrate risk analysis using Bayesian networks. In building Bayesian networks, analysts and decision makers can take into consideration both information contained in fragmented expert knowledge and the many parameters involved in complex problems. We propose the use of a graph theoretical technique for processing knowledge and building Bayesian networks in developing decision support systems.