×

Data mining methods for knowledge discovery. (English) Zbl 0912.68199

The Kluwer International Series in Engineering and Computer Science. 458. Dordrecht: Kluwer Academic Publishers. xxi, 495 p. (1998).
This excellent book is devoted to a very important area of the research and practice – data mining methods (DM) for knowledge discovery (KD) in databases. Recently, data mining is used in connection with database architectures such as data warehouses, or with applications, such as OLAP. Consequently, the book is highly actual and useful for a bright community of readers.
Chapter 1 introduces to DM and KD. In addition to basic definitions, the reader is informed about knowledge representation and methods, including production rules, regression, decision trees, correlation analysis etc. Methodological aspects are also mentioned and representative examples of KD systems are presented. The next four chapters provide a mathematical background on which it is possible to build particular methods. Chapters 2 and 3 discuss theory of sets, fuzzy sets, and rough sets. The material is presented in detail with a rich technical content. Also differences between computing with fuzzy sets and probabilities are highlighted there. In Chapter 4, the authors describe fundamental methods of Bayesian processing and classification. The chapter concludes by discussion of the probabilistic neural network. Chapter 5 concerns with population-oriented, evolution-like type of optimization. Various models of evolution mechanisms are developed. Genetic algorithms are presented as the most representative example of evolutionary computing.
Last four chapters cover main areas belonging to DM and KD. In the chapter 6, the authors concentrate on main issues in machine learning, particular on supervised inductive machine learning. Two approaches are discussed in detail: rule and decision tree algorithms. A family of hybrid algorithms completes the chapter. Chapter 7 describes neural network algorithms. Applications of the theory are focused on an analysis of large amount of numerical data or images. A fundamental technique of clustering is presented in Chapter 8. The authors modify generic methods for data mining and their use by the end user. Two methods based on fuzzy sets are developed and enhanced. The final Chapter 9 discusses basics of the data preprocessing. The Principal Component Analysis for pattern projection, feature extraction and reduction, and Fisher’s linear discriminant and linear transformation are described in detail.
The book is described on high scientific level. It provides a sound background for the new area of DM and KD. The reader can explore not only basics of DM methods but he/she also obtains a guide how to use them for discovering new knowledge. Since the chapters are accompanied by exercises, the material can be used as a textbook for advanced courses DM and KD.
Reviewer: J.Pokorny (Praha)

MSC:

68T35 Theory of languages and software systems (knowledge-based systems, expert systems, etc.) for artificial intelligence
68-01 Introductory exposition (textbooks, tutorial papers, etc.) pertaining to computer science
68T30 Knowledge representation
68P15 Database theory
PDFBibTeX XMLCite