Browse Books

Reviewer: John A. Fulcher

Uncertainty in datasets can arise due to the data collection method (for example, the imprecise nature of input sensors), cumulative noise due to forecasting/prediction/interpolation techniques, or even by noise that is deliberately added (for example, as a precursor to privacy-preserving data mining). Whatever the cause, it poses significant challenges for database administrators, data analysts, data miners, and knowledge engineers alike. The three broad areas covered in Aggarwal's book are modeling and system design, management, and mining of uncertain data. A total of 27 authors have contributed to the 16 chapters contained in this edited volume. In chapter 1, Aggarwal provides brief synopses of the 15 chapters that follow. Chapters 2 and 3 cover uncertain data representation and modeling. Chapter 4 is devoted to probabilistic graph modeling, such as Bayesian and Markov networks. Two real-world systems are described in chapters 5 and 6: Trio and MayBMS. Data integration is the focus of chapter 7. The resolution of aggregate queries using probabilistic data stream techniques, such as "sketches," is the focus of chapter 8. Chapter 9 discusses the join operation that needs to be redefined for uncertain data, due to its inherently probabilistic nature. Indexing uncertain data is covered in chapters 10 and 11, with the latter focusing on spatiotemporal data. Probabilistic Extensible Markup Language (XML) is the topic of chapter 12. Mining of uncertain data is the focus of the next three chapters-more specifically, clustering in chapter 13, general transformations in chapter 14, and frequent pattern mining in chapter 15. The book concludes by describing how some of the techniques introduced in the preceding chapters can be applied to biomedical imaging, as a representative application domain. Aggarwal's book is a timely publication, in that it provides a good summary of the current state of the art in the area of uncertain data modeling, management, and mining. It should be of interest to researchers and graduate students involved in the area, as well as novices who wish to become acquainted with the topic. Online Computing Reviews Service