The authors take a conceptual approach to selected ideas and problems in machine learning. They limit themselves to the learning of conceptual descriptions from the supervised presentation of examples. An introductory chapter discusses the notion of concept acquisition, the major issues of concept acquisition, and the literature pertaining to this topic. The system ML-SMART is introduced and an overview of the learning methods used by ML-SMART is given.

Chapter 2 considers issues in knowledge representation, especially those pertinent to machine learning. Structures are introduced to represent the examples of the concepts to be acquired. The form and extent of background knowledge and the structure of the learned concept descriptions are presented.

In chapter 3 the authors consider the memory management problems of machine learning and describe the use of ideas and structures from relational database theory to implement learning methods. The authors emphasize specialization and generalization operations.

Chapters 4 and 5 focus on inference techniques. Planning strategies and criteria are presented. The authors mention a connection with explanation-based learning and highlight the integration of inductive and deductive techniques.

Chapter 6 presents the experimental results obtained with the ML-SMART system in the domains of diagnostic knowledge, speech recognition, and the recognition of visual signals. The diagnostic knowledge generated by the learning system is contrasted with that of an expert system developed for the same type of task.

Chapter 7 describes the development of a classification strategy for the rules produced by the knowledge acquisition system. Although the knowledge acquisition process employed a classification strategy, the authors present approaches for modifying the existing strategy; these modifications can improve the performance of the underlying system.

Based on the first chapter, I infer that the purpose of the book is to present information on the ML-SMART system and the results of experiments performed on the system. This purpose is achieved. The best feature of the book is that it provides a comprehensive treatment of a machine learning initiative. Information of this nature is typically disseminated piecemeal. The book’s format enables the authors to pull numerous considerations together.

This book is difficult to read. Its language and style are often awkward and unusual. The book appears to have been developed from transcribed lecture notes. The type style of the text and the uneven style choices for figures and tables further impede the book’s readability.

The intended audience for this book ranges from fourth-year college students through researchers. It is best suited for advanced graduate students and researchers. While the authors attempt to be introductory and give references to the literature for additional background, an understanding of the contents of the book would require a fairly extensive background and prior understanding of machine learning and related artificial intelligence topics.

The book includes a fairly comprehensive reference list. An index is also provided; it does not appear to be complete. Numerous learning systems are mentioned in the introduction, only some of which are indexed.

Overall, this work meets a specific need for a select group. The book will be valuable to those seeking an extensive view of a machine learning system. It is not suitable for those seeking an introduction, overview, or broader exposure to machine learning.