In this paper we propose an extension of a multi-criteria majority-rule sorting model that allows the handling of problems where the decision alternatives contain imprecise or even missing evaluations. Due to the imprecise nature of the evaluations we offer the possibility of assigning an alternative to one or more neighboring categories, both as input for inferring the model parameters as well as the output of the classification. Our contribution also contains an algorithmic approach for extracting the parameters of this model during an elicitation process, which is validated across a wide range of generated datasets.