Evidence measures based on fuzzy information

An organized body of results pertaining to the analysis of evidence, when the available knowledge is pervaded with imprecision, is provided. The evidence theory of Dempster and Shafer is extended to the case of fuzzy observations and fuzzy events. Upper and lower possibilities of such events are derived by iterating the generation process of upper and lower probabilities, as done by Dempster. Both probabilistic and ''possibilistic'' models are developed in parallel. These evidence measures are used for decision evaluation when the available knowledge is poor. The classical model of decision-making under uncertainty is thus extended to the case when the consequences of a decision are only roughly described and their probabilities of occurrence modeled by intervals or fuzzy numbers.