A MULTICRITERIA EVALUATION METHOD BASED ON A PROBABILISTIC INTERPRETATION In this talk, we present a new way of representing and aggregating preferences using a stochastic interpretation. We consider a set of alternatives A and r (binary) preference relations on A. Our objective is to aggregate this information into a single preference relation. We consider a Markov chain where each state corresponds to an alternative, and compute the transition probabilities using the r preference relations. We then rank the alternatives according to the asymptotic probability distribution given by the Markov chain. In light of Arrow's theorem, the validity of such a ranking is questionable. We therefore extend the method to introduce incomparability and present a graphical scheme to represent alternatives. We demonstrate various properties of this method, especially in the important case where the considered binary relations are semiorders.