Granular Covering Selection Methods Dependent on the Granule Size

Abstract

In today’s increasingly computerized world, the amount of electronic information is rising at an incredible pace. The huge number of internet resources, in particular diverse data bases, forces users to use some methods of acceleration to get the awaited result in a reasonable time. There are a lot of problems which require acceleration, for instance, the speed of search engines, classification modules, statistical tools, marketing tools, and many more. There are a lot of approaches, which try to overcome problems through the speed of information processing. One of the main groups is based on usage of software methods as acceleration, optimization, and approximation tools. An exemplary paradigm which, among others, gives the tools for approximation of a huge amount of information, is the granular rough computing paradigm. This is a sub-paradigm of the rough set theory which was proposed by Professor Pawlak in 1982. It provides tools which are useful in widely understood classification problems, and in lowering the amount of information with the preservation of important knowledge.

In this paper we focus our attention on the granular methods developed recently by Professor Polkowski. An important element of these methods is the covering search method. We show and experimentally check three of our methods dependent on the number of objects inside granules.