Browse Books

Reviewer: Jose M. Ramirez

Databases are no longer the preferred source of information that they once were. Less structured sources, such as the World Wide Web with its enormous coverage and diversity, have become the primary sources of information. Researchers focus on these repositories, trying to find techniques capable of performing searches where exact queries, based on well-known structures, fail. This book presents a state-of-the-art report on the techniques, based on similarity or content, for retrieving information from unstructured repositories. The authors, using a very didactic style, present the theoretical background, practical techniques, experimental results, and research directions of the field. The book assumes some moderate knowledge on information retrieval, and is recommended as a starting point for researchers and practitioners that plan to work on similarity search. The book is organized into two parts. Part 1, "Metric Space in a Nutshell," describes the fundamentals of, and the techniques used to perform, partitioning, search, and transformation of search spaces (metric space, distance measures, similarity queries, and partitioning). Part 2, "Metric Searching in Large Collections of Data," focuses on very large repositories and describes advanced work in progress, such as approximate similarity search and distributed indexes (M-tree and hash-based). The book?s clean language, good illustrations and experimental results, and more than ten pages of references make it a valuable resource. That it also compiles, into a single volume, the theoretical foundations, strategies, techniques, and future directions of similarity search makes it an extremely valuable resource. Online Computing Reviews Service