Summary of a Panel Discussion on Representation of Uncertainty and Imprecision in Decision Support Systems

ZIMMERMANN. The modelling of systems about which decisions are to be taken must be based on three key ingredients: a language in which the system can be described, a calculus by which the language can be manipulated to derive inferences and conclusions or induce causes and assumptions, and representation within the language of the specific system of interest. It is an empirical matter to choose amongst possible modelling methodologies, and it is by no means clear that a universal methodology will ever be at hand. However, each such formal methodology must involve both a consistent set of axiomatics and an (often considerable) approximation of the complexities of the system being modelled. Natural language is a powerful modelling and decision-making tool. The use of the fuzzy system theory methodology allows the modeller to take advantage of the power and deliberate vagueness of natural language in a consistent, replicable manner.