Sugeno integrals, as multiple criteria aggregation functionsthat take into account a form of synergy between criteria, are a an importantfamily of tools for modeling preferences, which are qualitative,defined on ordinal scales. This paper addresses the problem of the elicitationof Sugeno integrals from a set of data that associates a globalevaluation assessment to situations described by multiple criteria values.The way a family of Sugeno integrals that are compatible with datashould be updated at the arrival of a new piece of data is shown tobe analogous to the way a set of hypotheses evolve in the version spacelearning setting when new data are considered. In the fact, the elicitationof the family of Sugeno integrals is very similar to such an updating processin a graded extension of the version space setting, recently proposedin the framework of bipolar possibility theory.