Using graphical models and multi-attribute utility theory for probabilistic uncertainty handling in large systems, with application to the nuclear emergency management

摘要: Although many decision-making problems involve uncertainty, uncertainty handling within large decision support systems (DSSs) is challenging. One domain where critical emergency response management, in particular nuclear response, making takes place an uncertain, dynamically changing environment. Assimilation and analysis of data can help to reduce these uncertainties, but it do this efficient defensible way. After briefly introducing the structure a typical DSS for emergencies, paper sets up theoretical that enables formal Bayesian be performed environments like architecture. In such probabilistic DSSs input conditional probability distributions are provided by different experts overseeing aspects emergency. These probabilities then used maker (DM) find her optimal decision. We demonstrate unless due care taken composite framework, coherence rationality may compromised sense made explicit below. The technology we describe here builds framework around which updating modular way, ensuring both efficiency, provides sufficient unambiguous information enable DM discover expected utility maximizing policy.