Generation of fuzzy evidence numbers for the evaluation of uncertainty measures

Abstract

Uncertainty is an important dimension to consider to evaluate the quality of information. In real world, information tends, usually, to be uncertain, vague and imprecise leading to different types of uncertainty, such as randomness, ambiguity and imprecision. Methods to quantify uncertainty, will help to quantify information quality. This paper presents a general measure of uncertainty framed into the fuzzy evidence theory named GM, quantifying in an aggregate way the three basic types of uncertainty: non-specificity, fuzziness and discord considered within the framework of Generalized Information Theory (GIT). Monte-Carlo simulations are used to study the behavior of GM with respect to the up-cited uncertainty types. Results show that the total uncertainty GM behave properly as we increase and decrease the various types of uncertainty.