D-SCRlPT: A Computational Theory of Descriptions

This paper describes D-SCRIPT, a language for representing knowledge in artificial Intelligence programs. D-SCRIPT contains a powerful formalism for descriptions, which permits the representation of statements that are problematical for other systems. Particular attention is paid to problems of opaque contexts, time contexts, and knowledge about knowledge. The design of a deductive system for this language is also considered. 1. Int roduction 1.1 Ways of Representing; Knowledge Methods advocated for representing knowledge in artificiall intelligence programs have Included logical statements (McCarthy, Sandewall), semantic networks (Quillian, Schank), and procedures (Hewitt, Sussman and McDermott), All these approaches shpre one fundamental concept, the notion of predication. That Is, the basic data structure In each system Is some representation of a predicate applied to objects. In this respect, the various systems are more or less equivalent. But this basic idea must be extended to handle problems of quantification and knowledge about knowledge. Mere the systems do differ. We will argue, though, that these differences result from the descriptive apparatus used in the particular systems being compared, rather than from an inherent advantage of, say, procedures over declaratives or vice versa. Advocates of PLANNER (e.g. Winograd, p. 2153 have argued that the predicate calculus cannot represent how a piece of knowledge should be used. But this is true only of the first-order predicate calculus. In a higher-order or non-ordered declarative language, statements could be made which would tell a theorem prover how other statements are to be used. PLANNER, on the other hand, has no way of directly stating an existential quantification, but this does not mean that procedural languages are necessarily Incapable of handling that problem. Our belief, then. Is that the type of system used to represent knowledge Is unimportant, so long as It has sufficient expressive power. This paper presents an attempt at such a system, the language D-SCRIPT. As the name implies, the most Interesting feature of D-SCRIPT is Its powerful formalism for descriptions, which enables it to represent statements that are problematical in other systems. No position will be taken as to what kind of language D-SCRIPT is. Since It is intended to answer questions by making deductions from a data base, it can be thought of as a theorem prover. Since it operates by comparing expressions like the database languages of PLANNER and CONNIVER, It can be thought of as a pattern-matching language. And since It Is Turing universal …