of each concept are described and compared, providing a rich basis for further application-specific development in this area. The evolution of tools and frameworks for blackboard system development has proceeded in parallel with theoretical and application developments. This is to be expected in a research area focused on system architectures. The section on Development Environments contains four papers which highlight some of the current generic systems (shells) available for blackboard system construction. The goal of the present generation of development environments is to provide greater control and representational flexibility within an efficient run-time framework and to allow the integration of multiple reasoning representations within a single system. The final section of the book consists of selected papers on applications oftbe blackboard architecture. Given the strong orientation towards applications throughout the history of blackboard research (in fact, most developments have been driven by appli- cations), it is not surprising that a large and diverse body of literature has evolved on applications. Even in the few papers contained here, we find a variety of classic AI problems addressed: diagnosis, information fusion, planning, and design. This is only a small sampling, however. Individuals desiring additional material in this area might consult Blackboard Systems, which contains several additional papers on recent and historical applications. In summary, I believe BlackboardArchitectures and Applications is a valuable reference for individuals who are designing or developing complex knowledge-based applications. The book is not intended as a tutorial, however. Readers who desire a brief introduction to the subject might try a pair of articles in AI Magazine [1, 2] which provide an excellent overview and history of blackboard systems. 1. Nii, H.P., Blackboard Systems (Part 1),AIMagazine 7(2), 38-53. 2. Nii, H.P., Blackboard Systems (Part 2), A/Magazine 7(3), 82-106. Developing and Managing Expert Systems David S. Prerau Addison Wesley Reviewed by: Philip N o r t o n LOGICON 4561-C Interlaken CCE Alexanderia V A 22312 tel: (202) 646-5686 This book provides a step by step approach to developing and managing expert systems. The text thoroughly covers all of the proven techniques used in each step of the development process and provides very pragmatic advice covering typical problems faced with building expert systems. One common weakness of many business oriented expert systems texts is that they provide a general listing of expert system examples and usually end up focusing on a system that the typical business person cannot relate to. A case in point is Waterman'sA Guide to Expert Systems and James Martin's Expert Systems: A 7ktodal. Both texts are adequate beginner texts, but Waterman devotes too much attention to Prospector, and expert system for geologists, while Martin provides a lengthy overview of different systems that are available. These exhaustive lists, with descriptions of various expert systems, in a multitude of fields, usually overwhelm the reader with triviality. This information is quickly outdated and rarely provides the reader with any lasting knowledge. David Prerau provides a book format with very detailed information on developing expert systems which can be easily digested by the novice developer but can also provide practical insight for the business person who is dabbling in expert systems and wants to get his or her feet wet. As a complement to the text, Prerau dissects every aspect of system development and applies the concepts from the text to the "real world." This feature is what sets Developing and Managing Expert Systems apart from the rest. This book gives a very detailed discussion of expert system development, with each section of the book consisting of two parts. The first part of each section contains a very thorough discussion of the conceptual aspects of development. This discussion isone of the most readable that I have come across. The second part of each section details how the author applied these concepts to COMPASS, a large industrial expert system developed by the author at GTE laboratories. There is a definite learning advantage to reading the text and seeing the knowledge applied. In addition, the COMPASS descriptions are in smaller print italics and separated from the general reading so it does not interfere with the flow of the book. Readers can ignore the COMPASSdescription easily if they choose to, but it is my opinion they would be the losers. The COMPASS descriptions are in a journal type format that presents the development process from beginning to end. One very helpful aspect of the text is the author's use of checklists for each aspect of the development process. These lists provide valuable reinforcement for the reader and good summaries of the development phases. I found myself referring back to these lists many times, and I probably will again during my next expert systems development task. While the book is well written and very easy reading, it is not recommended for a highly technical person since many of the concepts are very basic. A computer scientist may want to read Expert Systemsfor Erpertsby K. Parsaye for a more computer science oriented look at the subject of expert systems. However, Developingand ManagingExpert Systems is highly recommended for project managers who need a crash course in expert systems. From the managerial standpoint, this book is one of the best I have read. It is also recommended as an introductory text for any graduate level business course in expert systems. S I G A R T Bulletin, Vol. 1, No. 3

/lp/association-for-computing-machinery/book-review-developing-and-managing-expert-systems-by-david-s-prerau-gjiBIs0gDJ