Introduction to Modern Information Retrieval (2nd edition)

The book is aimed at students of library and information science programmes, both postgraduate and undergraduate, and at an international audience. The book covers practically all areas of information retrieval, some parts in more depth than others, such as searching, cataloguing and classification, evaluation and research issues.

Material has been added and updated to most chapters, but the most significant change from the first edition is splitting the areas of cataloguing and classification into two separate chapters and providing more detail on each area.

This could potentially be useful to computing students as well, as the detail of how information retrieval systems are implemented is described at reasonable length in the book. However, such readers would probably want to skip over some of the introductory material which may be somewhat basic for their interests. Issues such as information seeking are discussed, and important work is identified and used.

The book is for the most part up‐to‐date with trends in information retrieval (IR). I do have some quibbles, however. There is a lack of discussion of machine learning in IR, which through models such as support vector machines (SVMs) are something of a hot topic – no references to current work in the area are given in the relevant section, for example the work of Thorsten Joachims (2001). There is a lack of appreciation of the impact of the probabilistic model of IR (e.g. there is an unsupported statement on page 175 that this model does not yield sufficiently better results than the Boolean or vector space models). The probabilistic model has become very influential recently through the BM25 weighting function, and most participants of the TREC evaluation conference use variations of the function.

However, my biggest quibble is a reference to work published in 1990 which claimed that the performance of serial scanning of text was nearly comparable with searching an inverted list. Stone (1987) showed that this was not the case, and also showed that a parallel inverted file method is hard to beat in terms of reducing run times for searches.

Having said all that, there are some very useful chapters for both my teaching and research. There is a lot of useful material on online searching, for systems such as Dialog, which I used on my courses. There are two chapters on users and information retrieval, which is of great use for anybody (like me) who wants to move into that area. The two chapters on natural language processing are also very interesting and useful. Important trends such as music information retrieval are also covered, and non‐text retrieval issues are reasonably addressed.

The writing for the most part is clear and the layout makes it easy to read. My only quibble here is that there are some typos in the book that could have been erased with some care in the editorial process. An index to the book is included at the end, and references to quoted material are given at the end of each chapter. I would recommend this book to any general reader who is interested in IR issues. It is particularly useful to students undertaking a course on IR in a library and information science department, and I use it as an essential text on the courses I teach at City University.