This study aims to develop an interactive probabilistic approach that asks small number of questions to the decision maker for multi-criteria sorting problems. Alternatives are assigned to categories by the decision maker through iterations. The assignment probabilities of alternatives are calculated by collecting information from the mathematical models that are solved to determine the possible categories of alternatives. Relative entropy is used to measure the assignment ambiguities of alternatives and the one with the highest uncertainty is asked the decision maker. Alternatives with a certain level of assignment uncertainty are probabilistically assigned to categories. The aim here is to minimize misclassifications by not allowing probabilistic assignments without getting sufficient assignment information from the decision maker. Although several interactive approaches are suggested in literature, there is only one study that classifies the alternatives according to their assignment probabilities. Our approach is tested against it on the energy trilemma index data published annually by the World Energy Council to evaluate the energy performance of countries. Experiments on the problem, in which 128 countries are evaluated on four criteria and assigned to four categories, show that the proposed approach is successful in reducing the cognitive burden of the decision maker.