On the use of a new discrepancy measure to correct incoherent assessments and to aggregate conflicting opinions based on imprecise conditional probabilities

We give a preliminary study of a new procedure to correct incoherent imprecise conditional probability assessments. The procedure is based on parametric optimization problems which have as objective function a new discrepancy measure. We show through simple examples how the procedure of correcting ncoherent assessments can be properly extended to aggregate conflicting opinions, and can be generalized to embed importance weights of each assessment.