Information granulation for rough fuzzy hypergraphs

Abstract

Rough-granular computing is a basic model for applying the concept of both rough set theory and granular computing methodology into different knowledge discovery processes; to classify data sets. In the literature, authors have used techniques of rough-granular computing which involve the construction of certain generalizations of Pawlak’s rough set model by deploying sets with structures such as graphs or, more generally, hypergraphs. Indeed, a hypergraph model, with its fairly strong descriptive potential to analyze structures of complex information systems, has occupied a central position in granular computing. A specific concern for this paper is to provide a generalization of the formulation of approximations in rough set model of information granulation. Generalized approximation operators are constructed to account for various combinations of information resident in input granules, instead of mere local descriptions of specific parts of an information system at multiple levels of attributes. An algorithm which describes information granulation for hypergraphs and accounts for information provided by overlapping granules is constructed and illustrated. Finally, an extension of this generalized technique to fuzzy hypergraph models is indicated.