This book contains a collection of essays on classification most of which are written by rather well-known contributors.
Overall the book is a disappointment and points to problems in library and information science (LIS) as a research field. Classification is often regarded as one of the core subdisciplines of the field and as one of the core qualifications of library and information professionals. Nevertheless, no classification researchers are today visible in bibliometric maps of LIS (e.g. White & McCain, 1998)!
One of the problems in this book is that it fails to define classification and to distinguish between different kinds of classification. By only considering systems like Dewey, LC and facetted classifications, it fail to consider, for example, bibliometric approaches in LIS as kinds of classifications and thus to consider the basic strength and weakness of different methods of classification. In computer science the term "ontologies" is very popular and can be considered a modern development in classification research. Vickery (1997) made a useful introduction to this research, but it is not considered in the present book.
If the electronic environment is to be fully considered, one need to compare the relative strength and weakness of all kinds of subject access points (cf., Hjørland & Kyllesbech Nielsen, 2001). One have to consider what utility-if any-classification codes can have in relation to all other kinds of access points.
In Chapter 3 Julian Warner actually do take a step toward considering inherent weaknesses in current approaches to Information Retrieval (IR), and this chapter was in my view the best one. I think he is right in making the point that the IR-tradition has built on the assumption that the system should provide a set of records that satisfy a query. What an IR system in his view should do is to enlarging the users' capacity for informed choice between the representation of objects in the given universe of discourse.
In recent years the methods of classification and more generally: Knowledge organization has been reconsidered. Hjørland & Albrechtsen (1999) claimed that the four basic methods are respectively empiristic, rationalistic, historicist and pragmatic. If one uses, for example bibliometric methods, one applies an empiricist method. The best representatives of the rationalist method are the facetted classifications. An example of the importance of historicist methods are given in Hjørland (2000) considering the classification of the social sciences. An unfolded comparison of all methods used in one domain is given in Hjørland (1998). In my view, the future of classification is connected to a combination of these four methods of classification and to the further clarification of strong and weak aspects of different methods and systems. Unfortunately, these issues are not addressed in the book, while it fail to answer the fundamental questions about the future of classification in LIS.

...