Fundamentals of predictive text mining is a second edition that is designed as a textbook, with questions and exercises in each chapter. The authors mention that the book was extensively class tested and additional supplemental material is available for instructors. The book can be used with data mining software for hands-on experience for students. The software used is of a professional grade. The content is supposedly suitable for senior undergraduates as well as graduate students, but I found the information presented to be of an advanced nature that may not suit undergraduates. The book is well written and proceeds gradually from simple to complex methods in text mining. It is arranged in nine chapters.

The important new topics added in this edition include: deep learning, graph modeling, social media mining, and dependency parsing. The authors describe the basics of text mining and how textual data can be converted into numeric values in chapter 1. Chapter 2 shows in greater detail how automation is used to convert textual material into a machine processing format. The preferred format mentioned is Extensible Markup Language (XML), as most documents come in this format already. The authors then discuss some of the difficulties in dealing with punctuation marks embedded in texts during automatic processing in order to identify key phrases. Each chapter comes with useful algorithms that could be used in converting text to machine processing. Chapter 3 addresses the main topic of this book: prediction. It is shown that the prediction problem for text is text categorization. For machine processing, text material is categorized using a sparse matrix. The authors point out that predictive text mining needs samples of prior experience. Some of the methods discussed are from big data processing, for example, similarity and nearest-neighbor methods. This chapter heavily uses mathematical concepts.

Chapter 4 deals with information retrieval, which is often the goal in document search. For example, when users submit a query to a help desk, it becomes essential to see all of the matches from the query to stored documents and retrieve the one that closely matches the query in order to respond to the query. This chapter contains several useful techniques that a user could adopt for information retrieval. In chapters 5 and 6, the authors discuss adapting big data tools such as similarity and clustering to document handling for prediction. Examples are given to illustrate this. Chapter 7 is devoted to discussing the adaptation of predictive mining techniques for structured, unstructured, and hybrid data. Chapter 8 discusses several case studies related to text mining techniques. The book concludes with a list of references that help the reader to explore related topics from the published literature. There is also an index.

I was pleased with the level of rigor shown in the discussions. As mentioned at the outset, the material quickly gets quite mathematically involved, so undergraduates may find it very challenging. The book will be very useful for people planning to go into this field or to learn techniques that could be used in a big data environment.

More reviews about this item: Amazon