In this paper, our work begins with the classification of indefinite information presented in the first chapter and these classes include randomicity, fuzziness, non-partitionality, incomparability, incompleteness and unreliability. And the corresponding theories and methods of studying these information are simply described. Then the following chapter of this paper discusses the approximate processing ofrough boundary region- sort of problem of non- partitionality. Inthis part, inner and outer approximate boundaries are presented for the first time and their properties and processing idea are studied. And then we generalize the conceptions of inner and outer approximately definable sets. Next, chapter 3 puts forward a new data filtering method based on both binary information system and degree of rough membership. To some information systems, this technique to some degree avoids the complexity of deducing the rule reducts. It follows that some open questions are listed in the end of this paper.
|