Granular Computing Based Data Mining in the Views of Rough Set and Fuzzy Set

Granular Computing Based Data Mining in the Views of Rough Set and Fuzzy Set

Guoyin Wang (Chongqing University of Posts and Telecommunications, P.R. China), Jun Hu (Chongqing University of Posts and Telecommunications, P.R. China), Qinghua Zhang (Chongqing University of Posts and Telecommunications, P.R. China), Xianquan Liu (Chongqing University of Posts and Telecommunications, P.R. China) and Jiaqing Zhou (Southwest Jiaotong University, P.R. China)
DOI: 10.4018/978-1-60566-324-1.ch007

Abstract

Granular computing (GrC) is a label of theories, methodologies, techniques, and tools that make use of granules in the process of problem solving. The philosophy of granular computing has appeared in many fields, and it is likely playing a more and more important role in data mining. Rough set theory and fuzzy set theory, as two very important paradigms of granular computing, are often used to process vague information in data mining. In this chapter, based on the opinion of data is also a format for knowledge representation, a new understanding for data mining, domain-oriented data-driven data mining (3DM), is introduced at first. Its key idea is that data mining is a process of knowledge transformation. Then, the relationship of 3DM and GrC, especially from the view of rough set and fuzzy set, is discussed. Finally, some examples are used to illustrate how to solve real problems in data mining using granular computing. Combining rough set theory and fuzzy set theory, a flexible way for processing incomplete information systems is introduced firstly. Then, the uncertainty measure of covering based rough set is studied by converting a covering into a partition using an equivalence domain relation. Thirdly, a high efficient attribute reduction algorithm is developed by translating set operation of granules into logical operation of bit strings with bitmap technology. Finally, two rule generation algorithms are introduced, and experiment results show that the rule sets generated by these two algorithms are simpler than other similar algorithms.
Chapter Preview
Top

1. Introduction

Across a wide variety of fields, data are being collected and accumulated at a dramatic pace. It is necessary to acquire useful knowledge from large quantity of data. Traditionally, data mining is considered as the nontrivial extraction of implicit, previously unknown, and potentially useful information from data. That is to say, knowledge is generated from data. But in our opinion, knowledge is originally existed in the data, but just not understandable for human. In a data mining process, knowledge existed in a database is transformed from data format into another human understandable format like rule.

Granular computing is a label of theories, methodologies, techniques, and tools that make use of granules in the process of problem solving. The philosophy of granular computing has appeared in many fields, and it is likely playing a more and more important role in data mining. Rough set theory and fuzzy set theory are two very important paradigms in granular computing.

In this chapter, a new understanding for data mining, domain-oriented data-driven data mining (3DM), will be proposed. Moreover, we will introduce basic concepts of granular computing and analyze the relationship of granular computing and 3DM. We will also discuss the granular computing based data mining in the views of rough set and fuzzy set, and introduce some applications of granular computing in data mining. In the end, some conclusions are drawn.

Complete Chapter List

Search this Book:
Reset