[Haralick and Shapiro, 1993b] represent these various measurements over the interval unit (see figure 43).
Let j and k be two elements of [0,1] such that . We are interested in the interval of belief describing event A:
As new information becomes added to what we knew till now, the part assigned to uncertainty decreases being distributed either towards Bel(A), i.e. the belief in event A, or to , i.e. the non-belief in A. The more information there is, the more uncertainty decreases.
This leads us to the measurement of plausibility Pl(A) of event A which is the sum of the belief Bel(A) in the occurrence of A and of uncertainty on A. This means that the maximum extension possible for belief in A lies in all the belief which does not disprove A.
In the same way, the doubt about the occurrence of A, corresponds to the part of the belief assigned to the negation of A plus the uncertainty related to the occurrence of A.
Belief functions and plausibility functions satisfy:
Union and intersection of subsets satisfy: