Research and Development of Granularity Clustering

Abstract

The rapid development of the Internet and information systems result in massive, high-dimensional, distributed, dynamic and complex data which often have characteristics of incompleteness, unreliability, inaccuracy, inconsistency and so on. Traditional cluster methods are very difficult to satisfy these data cluster demand. The granular computing has developed the cluster analysis research area, enhanced the use value further, and made theoretical significance closer to reality. The cluster may be carried on at different levels and from different angles through the granularity transformation. This will be beneficial to the solution of the problem. This article gives an overview of the essential relationship between granular computing and clustering, the advantages and scope of granule clustering, the research results about granule clustering based on rough sets, fuzzy sets and quotient space. Then the article discusses the necessity that three granular computing models integrate into each other to be used in granule clustering through comparative analysis, and make the analysis and the forecast to establish a unified granule clustering model at last.