A double-clustering approach for interpretable granulation of data

摘要: This paper addresses the problem of forming information granules well-defined and clearly delineated semantics from numerical data. In particular, an approach is proposed to induce fuzzy a pattern space. The mainly based on “double-clustering” algorithm, which identifies, in first instance, clusters original space, then projections these provide most informative granulation induced are described terms sets that can be used form propositions, represent linguistic fashion. A well-known benchmark considered illustrate show its ability extracting meaningful granules.