An overview of the HORNE logic programming system

Abstract

MIT continues to be a premier AI research center, and the book makes available some of the most important and influential results in a handy form. Since the original publication, much of the work has also been published in journals and other forms, and in some cases superceded by other work at MIT and elsewhere. However, the book chapters are much more tutorial, and hence easier to read, than corresponding journal papers and contain pointers for more detailed reading. The collection is still highely valuable, particulary for that sub-field of AI that the reader does not follow closely.

Parsing with restricted quantification: an initial demonstration1

“…Notice that this representation contains complete information about the taxonomy of voice features. My more general research (Frisch 1985) addresses some of the issues that arise when using taxonomic representations that do not contain complete information. This paper ignores such issues; let us simply assume that there is some representation of the taxonomic information shown in the above semantic network (call it TR 3) and that we have a procedure that can decide whether the representation entails certain relationships.…”

“…Though Cohn's system is unimplemented, he has argued (Cohn 1985) that it should solve the steamroller even more effectively than Walther's system. The H O N E logic-programming system (Allen er Frisch et al 1983) is based on an RQT Horn-clause logic. Its implementation incorporates a number of effective methods for dealing with a taxonomic representation.…”