Kiyoshi Niwa presents his experiences with building a knowledge-based system for use as a cooperative consultant in engineering risk management. Apparently the contents are mostly based on his doctoral dissertation work. The book has two main themes: a general description of AI techniques that can be used in the construction of expert systems, and a detailed discussion of the specific design, implementation, and use of a knowledge-based system for construction project risk management. Niwa presents these two themes together over the course of nine chapters.
The overview and description of general techniques and methods for building knowledge-based systems may interest someone totally new to the area (e.g., a professional in engineering risk management), but the material is not adequate by itself to be of value as a general reference source for expert systems methodologies. Many of the explanations have philosophical overtones that make them hard to read because they are not always clear.
As a case study on the building of a system for engineering project risk management, however, the book scores high marks. The author takes the reader through all the necessary steps of building a knowledge-based system in this domain. He gives reasons and support for the various decisions he makes along the way about knowledge representation structures, inference techniques, knowledge acquisition methodology, and the problem-solving paradigm. Reading the book from this perspective is definitely valuable and useful, especially if the reader has an interest in computer-based tools that support engineering risk management.
The concluding chapter, which contains lessons for successful implementation, is useful reading for those embarking upon an expert systems development project for the first time. In closing, the author strongly emphasizes the notion of human-computer cooperative systems. One gets the feeling that he expects the reader to finish with a strong conviction that these systems are the wave of the future in AI and expert systems. Unfortunately, the author’s arguments are not very convincing or clear that this paradigm is something dramatic or new.