Browse Books

Reviewer: Felipe Carino

Subrahmanian has written an excellent, timely book on diverse aspects of multimedia database systems. By “multimedia,” the author primarily means image, text/document, video, and audio data types; indexing; storage; networking; and usage issues with respect to database systems. Part 1, “Preliminaries,” describes the evolution of multimedia data types. The examples are interesting, and even people who are not experts in the field can relate to their use. This section provides terse overview chapters that describe relational and object-oriented databases. The overview of relational databases briefly describes relational algebra, the relational calculus, and SQL, using set-theoretic notation. The object-oriented database overview presents traditional object notions, such as classes, types, methods, and inheritance; describes the Object Database Management Group's standardization efforts for Object Definition Language and Object Query Language; and explains the emerging object-relational database management systems and products. The analysis is excellent, but some description or at least mention of SQL3 would have complemented it. Part 2 covers “Organizing Multimedia Content.” A chapter on n -dimensional data structures provides a clear, complete description and analysis of k D trees, point quadtrees (2D trees), MX-quadtrees, and R-trees. The analysis is completed by good illustrations and diagrams. The image database chapter introduces and defines raw image type representations, compression, and segmentation, which serve as a foundation for similarity-based retrieval analysis. The chapter on text and document databases covers material normally associated with information retrieval systems. Precision and recall definitions and formulas are described, as are concepts and constructs (stop lists, word stems, and frequency tables) used to match and retrieve documents. The remainder of this chapter describes latent semantic indexing, telescopic vector trees, and other retrieval techniques, including inverted indices. The video database chapter is interesting because it does more than describe video compression standards. The author specifies and develops definitions and formal notations for diverse video concepts. An SQL-like notation is used to explain how to match and retrieve video segments. The author describes two compact data structures for video data: frame segment trees and R-segment trees. The explanation is thorough and well illustrated and annotated. The final section covers video segmentation. The audio database chapter is similar to the video database chapter. It defines audio, provides examples, and presents indexing and compaction strategies, including the discrete cosine transform and discrete Fourier transform. The final chapter in Part 2 describes three multimedia database architectures for organizing and indexing data media types. Part 3, “Physical Storage and Retrieval,” is instructive and will help readers understand the complex issues associated with organizing and retrieving multimedia data. Chapters cover retrieving multimedia data from disks, CD-ROMs, and tapes. The disk chapter explains disk technology and Redundant Array of Inexpensive Disks (RAID). It describes the RAID-0, RAID-1, and RAID-5 architectures and their tradeoffs. It also has one of the best descriptions I have read of how to support VCR functions. The CD-ROM and tape chapters also describe basic technology, data placement, scan, seek, and retrieval issues. I found the discussion of tapes, tape striping, and tape libraries useful, since this material is not normally covered in database textbooks. Finally, Part 4, “Creating and Delivering Networked Multimedia Presentations,” covers creating multimedia presentations, media servers, and future directions (primarily the Web). The chapter on multimedia presentation covers authoring and related formalisms and the Bellman-Ford algorithm. The chapter on distributed media servers covers retrieval plans for continuous data types (video and audio); the author provides formal descriptions and retrieval plans for locating and delivering data from one or more sources. The book concludes with a description of retrieving information from the Web. This chapter includes discussions of push/pull in mobile systems and watermarking issues for intellectual property protection. This well-written book covers a wide range of multimedia database topics. From a relational or object/relational database system point of view, some discussion of concurrency and transactions would have been useful. The author provides URLs for commercial products and related research at the end of each chapter. The book will be useful to people who have a rudimentary knowledge of databases and would like to understand how database systems incorporate multimedia data types. I recommend it to practitioners interested in multimedia and as a textbook for advanced undergraduates and graduate students with some database knowledge. Practitioners and investigators who want to understand how complex multimedia data are stored, indexed, and retrieved should also read it.