In investigating temporal aspects of data, general knowledge discovery and data mining methods should be refined differently according to the data representation, the method to be applied, and the application domain. In addition to the refined general methods, there are many special methods for mining temporal aspects of data. This book covers temporal data representation and exploration.
The first five chapters introduce the temporal data representation and exploration methods in a didactic and comprehensive way, appropriate for novices and graduate students. Data should contain temporal information explicitly. Temporal aspects of data representation and similarity computation, treatment, and mining--clustering, classification, pattern discovery, neural nets, and genetic algorithms--are described with examples. Two other important temporal data sources that are dealt with in detail are data warehouses and time series. Data mining is treated in a broad sense. Aspects of data preparation, analysis, and forecasting are also treated.
The chapters that follow loosely refer to the introductory chapters, and show how temporal data mining is applied to various domains: the domain and its specifics are introduced first, and then temporal data mining techniques specifically developed for it are shortly sketched. The domains treated are: medicine, bioinformatics, financial and business, Web usage, and spatiotemporal. These chapters contain up-to-date information about research results. Unlike the first five chapters, they do not explain in detail the methods applied by the referred papers, but give a structured overview of the state of the art for researchers and interested innovative implementers. The main value of these chapters is the structured overview of a diligently collected, selected list of interesting research results. Each chapter contains a rich list of references. The author did a great job of arranging many resources, to the benefit of the interested reader.
The author continued to collect papers on temporal data mining even after the book was published. The updated list is published monthly on her Web site.
The index of the book groups some expressions, but only according to the table of contents; it is sometimes redundant and it only gives one occurrence of the expression--either the defining one or the applied one. Some of the italicized expressions in the text lack any explanation, definition, or reference in the index. Also, some of the advanced chapters cover the specifics of the application area significantly more than the data mining methods used to solve problems related to that area.
The book has a theoretical viewpoint; with the exception of MATLAB, it does not refer to any of the commercial tools, but it refers to standards and research tools. I recommend this book to those interested in temporal aspects of data mining in various application domains.
