596

Journal of the Operational Research Society Vol. 47, No. 4

REFERENCES 1. A. W. HoLT, H. R. RAMSEY and J. D . GRIMES (1983) Co-ordination system technology as the basis for a programming environment, Electrical Comm. 57, 307-314. 2. S. GREENSPAN (1985) Requirements modelling: a knowledge representation approach to software requirements definition, Technical Report CSRG-15, Computer Systems Research Group, University of Toronto.

Critical Heuristics of Social Planning: A New Approach to Practical Philosophy WERNER ULRICH John Wiley and Sons, Chichester, 1994, 504 pp. £19.95 ISBN 0-471-95345-8 Werner Ulrich's book published originally in 1983 was, and remains today, one of the foundations of the critical systems approach. It has in fact already been reviewed for The Journal of Operational Research Society by Mike Jackson 1 and I believe the conclusions of his review still stand: 'a massive debt is owed to Ulrich for his painstaking analysis of the foundations of the systems approach and for his own contribution A Critical Systems Perspective'. The book has now been issued for the first time as a paperback and this should hopefully make it available for a wider range of readers. My only regret is that this reissue has not taken the opportunity of providing a second edition and allowing the author to reflect upon the significant developments in the area since the original publication. Warwick Business School

JOHN MINGERS

REFERENCE 1. M. JACKSON (1985) The itinerary of a critical approach . . . critical heuristics of social planning: A new approach to practical philosophy. J. Opt. Res. Soc. 36, 878-881.

Neural Networks J. G. TAYLOR (Editor) Alfred Waller in association with UNICOM, Henley-on-Thames, 1995. xx ISBN 1-872474-27-6

+ 379 pp. £45.00

This book is derived from papers on neural networks presented to a conference on Adaptive Computing and Information Processing at Brunei University in January 1994. But it is more than simply a record of papers presented there. The papers have been updated and further ones added with the aim of making the book a survey of neural networks and their applications as at mid1994. In this it largely succeeds. Its viewpoint is, however, an academic one and contributions from industry are restricted to the research groups of the defence industry and its contractors. The financial services industry, which has driven much of the wider interest in neural networks, is represented only by a paper by A. N. Refenes of the London Business School arguing strongly for the use of proper statistical measures of performance in assessing the achievements of neural networks in finance. This lack of weight is inevitable, given the genesis of the book in a conference, but it limits the book's attraction to non-academic audiences. The book starts with a useful overview by J. G. Taylor explaining the various types of neural network and their biological progenitors. It then discusses the areas in which neural networks are applied: speech recognition; pattern recognition; control; and the analysis of time series. Finally it suggests that neural networks are here to stay but will increasingly be used in conjunction with other techniques. Of the conference papers, the one I found most impressive was by a group at British Aerospace's Sowerby Research Centre. This was concerned with machine vision and in particular the process of 'recognising' the components of a scene. Eighty full-colour images of outdoor scenes, both urban and rural, were scanned and digitised. They were then divided manually into 'seg-

Operational Research Society is collaborating with JSTOR to digitize, preserve, and extend access to Journal of the Operational Research Society. ® www.jstor.org