Measures of divergence on credal sets

When there is uncertainty about the values of a variable, it is possible to have different pieces of information about it. This can be due to the fact that they represent the points of view of different agents and these agents have followed different criteria when codify the pieces of information by means of numbers. The aim of this paper is to provide tools for the comparison of these representations. This is done in the framework of imprecise probabilities, more concretely the credal set theory is used. The reason is that credal sets are very general, including as particular cases possibility theory, probability theory and belief functions. So it can serve as basis to compare the representations of information obtained by following these formalisms. The measures used for comparison are: inconsistency measure, inclusion index and informative distance. © 2005 Elsevier B.V. All rights reserved.