The literature on Multiple Criteria Decision Analysis (MCDA) proposes severalmethods in order to sort alternatives evaluated on several attributes intoordered classes. Non Compensatory Sorting models (NCS) assign alternatives toclasses based on the way they compare to multicriteria profiles separating theconsecutive classes. Previous works have proposed approaches to learn theparameters of a NCS model based on a learning set. Exact approaches based onmixed integer linear programming ensures that the learning set is bestrestored, but can only handle datasets of limited size. Heuristic approachescan handle large learning sets, but do not provide any guarantee about theinferred model. In this paper, we propose an alternative formulation to learn aNCS model. This formulation, based on a SAT problem, guarantees to find a modelfully consistent with the learning set (whenever it exists), and iscomputationally much more efficient than existing exact MIP approaches.
展开▼