Building Large Knowledge-Based Systems: Representation and Inference in the Cyc Project

Abstract

The Cyc project, started by Doug Lenat at MCC in 1984, is the most ambitious knowledge representation project ever undertaken. It embodies Lenat's current ideas for a system intended to encode all of commonsense knowledge. By the year 1999, he hopes that no one would even think of buying a computer that doesn't have Cyc running on it. The book by Lenat and Guha is a report on the project as it was in 1989. A review of that book must distinguish four different things: the book itself, the Cyc project as it was when the book was written, the Cyc project today, and the developments that the designers are planning for the future. Of these four, the last two are probably the most interesting. This review has been difficult for me to write, because my thoughts about Cyc have changed a great deal since I first read the book in the spring of 1990. Doug Skuce showed me a copy of his review (which also appears in this issue), and it is similar to what I had originally intended to write. I agree with his complaints about the confusing organization of the book and the lack of precise definitions. Despite our reservations, we both used