Résumé
When solving complex and critical decision prob-lems involving several aspects, collecting decision-maker (DM)preferences remains a delicate issue. Indirect parameter iden-tification procedures are more adequate in such situationsthan procedures consisting to ask directly the DM to providepreferences. In the case of methods based on additive modelssuch as MACBETH, disaggregation procedures are implementedto handle this problem. In case of sophisticated non-additivemodel, e.g. Choquet integral based fuzzy measure, most of theexisting indirect procedures assume a 2-additive fuzzy measureand the DM is asked to change her/his preferences if they donot match this model. This paper proposes an approach todetermine preferences model that achieves a trade-off betweenmodel sophistication and the fastidious MACBETH questioningprocedure when considering interactions between more than twoattributes. First, it proposes a practical approach to solve areal case of multiple criteria decision problem while consideringsophistical model of preferences. On one hand, particular binaryalternatives are selected for a first questioning procedure to de-termine the fuzzy measure parameters. In order to make the laterprocedure as simple as possible to handle, some simplificationsare introduced to reduce both the amount of information andthe cognitive effort requested from the DM. Our propositionconsists in fixing several non-linear optimization problems untilreaching the first k for which the k-additive fuzzy measure matchthe DM preferences. On the other hand, starting from a decisionmatrix and the answers to a second questioning procedure, valuefunctions are determined. The second contribution is a theoreticalone and consists in proving the existence of a k-additive fuzzymeasure matching the DM preferences. Finally, the approach isapplied to a real world case that concerns the comparison of someelectrical and thermal vehicles regarding their environmentalimpacts.