A Review of the Knowledge Granulation Methods: Discrete vs. Continuous Algorithms

The paradigm of granular rough computing has risen quite recently — was initiated by Professor Lotfi Zadeh in 1979. This paradigm is strictly connected with the Rough Set Theory, which was proposed by Professor Zdzislaw Pawlak in 1982. Granular rough computing is a paradigm in which one deals with granules that are aggregates of objects connected together by some form of similarity. In the rough set theory granules are traditionally defined as indiscernibility classes, where as similarity relations we use rough inclusions. Granules have a really wide spectrum of application, starting from an approximation of decision systems and ending with an application to the classification process. In this article, approximation methods are shown in the framework of Rough Set Theory. In this chapter we introduce both discrete and continuous granular methods known in the literatureas well as our own modifications along with a practical description of the application of these methods. For described here granulation methods, we have chosen suitable methods of classification which can work properly with shown algorithms.