KARDIO: a study in deep and qualitative knowledge for expert systems by Ivan Bratko, Igor Mozetic and Nada Lavrac, The MIT Press, Cambridge, MA, 1989, pp 260

The first half of the book deals with the dynamic aspects of knowledge representation. It contains a comprehensive survey of knowledge representation formalisms, the architecture of various services of knowledge based systems, and their management. The emphasis is however on hybrid systems, including semantic networks, frame systems, structural inheritance networks, etc. The book demonstrates a deep understanding of these systems and of the underlying theories. The analysis is further constrained to a typical member of the family of hybrid representation systems based on KL-ONE. The second half of the book is devoted to the study of the problems associated with the revision of knowledge bases. It is often assumed, when discussing inferences, that knowledge bases grow monotonically. This excludes the possibility that a fact or definition has to be retracted or modified once entered. Here, instead of proposing a solution, the book takes a more general view on the problem of revision, which is then used as a yardstick for the effectiveness of previous solutions such as network editing approaches, knowledge base editing, deletion and additions of definitions systems, etc. The target audience of the book falls into three groups: teachers/students, researchers, and other practitioners in the field. Overall the book is a popularized version of the author's PhD thesis. As such it contains a considerable amount of theory and formalism which is of limited value to the first and last groups. It is however valuable reading for prospective PhD students and most serious researchers in the field. And it is still a good PhD thesis.