BOOK AND SOFTWARE REVIEWS
Belew, Richard K. Finding out about: a cognitive perspective on search engine technology and the WWW. Cambridge: Cambridge University Press, 2008. xxvii, 356 p. ISBN 0521630282. £24.99
This monograph can be concisely described as a gentle textbook-style introduction to the modern trends and techniques of information retrieval. Alhough the research field is concealed in the subtitle as 'search engine technology' much of the material present in the book consists of the common building blocks of information retrieval theory found in most other textbooks in the field.
It appears that one of the author's purpose has been to present the theory from a more human and user-oriented perspective by letting the chapters proceed from the search situation to document representation, query matching, and evaluation – and not until this has been developed in some detail we encounter a more mathematical characterization of the sub-processes. In this fashion one could say that the text moves in the opposite direction compared to, for instance, the well-established textbook Modern information retrieval (1999) by Ricardo Baeza-Yates and Berthier Ribeiro-Neto, where the focus is distinctly more systems-oriented. In general I think that the structure found in Finding out about... is appealing from a pedagogical perspective and merits evaluation in a didactic situation.
A somewhat perplexing aspect of the book is that the location of some sections and the presentation of the more formal matters leave you pondering about the intended target audience. The disposition of the book would indicate that it is aimed for a quite diverse audience (for instance library students, information professionals and even information science researchers) and not only for readers comfortable with technical prose, which is more or less unavoidable for the subject. On the other hand, when the text eventually gets into the technical details (as in Chapter 5, Mathematical Foundations) it seems that the reader is expected to be quite fluent in reading mathematical notation. Perhaps this chapter is not intended to be as generally available as the previous chapters, but here the author could have at least attempted to apply the same gentleness as before by providing the reader with the intuitive notions behind the mathematical concepts presented and even some calculation examples. Also, the details of the theoretical derivation of for instance Zipf's law and the binary independence model could have been relocated to an appendix and the main text could have been reduced to a condensed treatment of the general ideas involved. It is a fair guess that only a small fraction of the book's readers are genuinely interested in the precise mathematical derivation of these theories.
The content of the book is quite up-to-date (though it was first published in 2000) and touches upon some more advanced angles, such as the use of machine learning and logical inference in information retrieval and question answering. There is also an inspiring chapter on future directions towards the end of the book. As stated above, the disposition of the book is logical and has a pedagogical freshness to it, but some sections look uncomfortably located in Chapter 5: they do not strictly belong to the mathematical fundamentals, but seem to have been judged by the author to be too technical to be located anywhere else. As a case in point, the latent semantic indexing model and the binary independence model would be more naturally placed in the vicinity of the vector space model for comparison and contrast. All-in-all the book gives a quite comprehensive overview of this complex and evolving field and should work well as a textbook for an introductory course. One is not left with the urgent sense that anything important has been neglected or under-emphasized. In addition, the type setting and the general format of the book is pleasant, with a balanced and judicious use of illustrations. A welcome addition to the collection of guides into the fascinating world of information retrieval.