An ordered family of consistency measures of belief functions

Abstract

We propose a family of measures of consistency (and induced conflict) derived from a definition of consistent belief functions introduced previously. Besides satisfying the desired properties of monotonicity, boundedness, and extreme values, the novel family encompasses the existing probabilistic and logical consistency measures which are shown to correspond to two extremes of the family (lower sharp bound and upper asymptotic limit respectively). We illustrate the definitions and measures of consistency within an example of vessel destination estimation with inconsistent sources.