Sowa, John F., Knowledge Representation: Logical, Philosophical, and Computational Foundations.

(Pacific Grove, CA: Brooks/Cole, 1999, pp. 512).

This is a splendid book for those academic accountants who want to expand their computer science and information systems horizons. For most readers of The Accounting Review, the topic of knowledge representation exists only at the periphery

of their teaching and research interests, especially as defined by Sowa (xii) as "the application of logic and ontology to the task of constructing computable models for some domain." However, Sowa's mindset portends the way in which the principles of accounting information systems (AIS) will be taught in the five- to ten-year future, so if you are thinking of branching out to that part of accounting, you should start here. Moreover, the impending arrival of new business technologies like strongly typed enterprise systems (with business object components), e-commerce and e-reporting ontologies, and the semantic web makes some expertise in knowledge representation appealing for all accountants in the near term. For an overview of all of these trends, this book is an unparalleled source.

As I (and other semantic modeling instructors in AIS) rediscover every term in interactions with students, the most fundamental principles of knowledge representation come from the field of philosophy, followed in turn by those emanating from the fields of linguistics and behavioral science. Computer science and its semantic subsets (database theory, object-oriented analysis and programming, and artificial intelligence) are the Johnny-come-latelies of the field, and Sowa makes this intellectual progression and heritage obvious in every chapter. His dedication sets the tone: "To the spirits of the great knowledge engineers, Aristotle, Leibnitz, Kant, Peirce, and Whitehead."

I have been using Sowa's ideas since the publication of his first book, Conceptual Structures: Information Processing in Mind and Machine, in 1984, and I would guess that many accounting readers will find his present book a difficult first read. My advice is to skim it on the first time through--a strategy that the author seemingly encourages when he recommends that any chapter can be surveyed by reading just the first section and the first paragraph of each remaining section. In my opinion, it is Chapter 1 (Logic), Chapter 2 (Ontology), and Chapter 7 (Knowledge Acquisition and Sharing) that constitute the strongest message for accounting scholars. The intervening chapters--3 (Knowledge Representation), 4 (Processes), 4 (Purposes, Contexts, and Agents), and 6 (Knowledge Soup)--add more computational, logical, and notational detail. Although Sowa might not recommend it fully, I am convinced that a noncomputer-scientist can glean many valuable lessons here without incurring the additional overhead of learning his twin notations of predicate calculus and conceptual graphs. I have only a passable knowledge of both. yet I find the reference material here eminently usable.

As a liberal arts undergraduate in economics at Boston College, I was required to take four years of philosophy. (By contrast, I was only allowed to take two courses in accounting, and I had to sleep in a campus corridor overnight to get those precious slots.) In retrospect however, my logical, epistemological, and ethical training is a debt I will owe forever to the Jesuits there. I became a semantic modeler long before people like Ted Codd, Peter Chen, John and Diane Smith, and the Unified Modeling Language triplets (Rumbaugh, Jacobson, and Booch) invented ways tar me to use that modeling. I often tell my students that my modeling mentor was my demanding metaphysics teacher: Father "D-minus" Murphy, S.J. It was he, following Plato, who first ingrained in me that the concept of "a horse" (which can describe an occurrence like Traveler or Secretariat) was vividly different from the concept of "horseness" (which can describe classes like nonexistent unicorns). This is a knowledge representation lesson I remember every time I have a student denote the difference between a "course" and a "course offering," or between "the actual time it took to install an Oldsmobile windshield" and "the engineered time it was supposed to take."

I mention my personal history in this review because I think Sowa's book offers a similar opportunity for those prospective conceptual modelers who missed people like Father Murphy while they were students. Sowa is one of the world's acknowledged experts in logic, ontology, and modeling. His book takes some effort, but the exertion is well worthwhile. I remembered how worthwhile just a few short weeks ago, when I had a protracted debate in class with an Executive M.B.A. student who insisted that a piece of inventory that had undergone an expensive restoration process must be given a new "item#" for database purposes. She said that her large automobile company handled its inventory that way, so it had to be correct. This is an "identity" problem that is impossible to solve out of context, but she was unwilling to accept that solution. I finally resolved matters when I able to expound philosophically on the "ship of Theseus" problem from the Greeks (Sowa, 270). The fact that philosophers had been debating a similar paradox for centuries made it easier for her (and the class) to accept an imprecise solution. The person who put that example at the tip of my tongue was John Sowa, and maybe he can do the same for you in your work.

WILLIAM E. McCARTHY
Michigan State University