Perspectives on imperfect information processing

Expert system and decision support system technologies have many potential applications. It is our belief that many, if not most, relevant applications will require an appropriate blend of the features of each of these technologies. We denote such a system as a knowledge support system. One of the inherent characteristics of a knowledge support system is that the knowledge that is brought to bear on applications will come from multiple sources and perspectives and will be inherently uncertain, imprecise, incomplete, inconsistent, and otherwise imperfect. It will be especially important that knowledge support systems address information imperfection concerns in ways that are realistic. Several approaches to this problem have been proposed. They rely on a broad spectrum of mathematical models and algorithms and are widely scattered over a multidisciplinary array of publications. The underlying assumptions are often implicit. These factors make it relatively difficult for a system designer to select the “best” method(s) for a given problem. It is important that this be done, and done in a way that encourages human interaction with the resulting support system. Many trade-offs are inherent in this that involve the whole spectrum of analytic and behavioral concerns. Major approaches for the representation of imperfect knowledge and their use in approximate reasoning algorithms for knowledge support systems are described.