Computer Science Department
School of Computer Science, Carnegie Mellon University
DYPAR-1: Tutorial and Reference Manual
Mark Boggs, Jaime Carbonell, Marion Kee, Ira Monarch
DYPAR-I is a rule-based natural language interpreter, adaptable to many limited-domain applications such as data-base query, command interpretation for software systems, and simple expert system command and query tasks. DYPAR-I is not designed to parse full, unrestricted natural language text: such systems do not yet exist. Rather, its primary objective is to serve as a high level programming tool making it possible for anyone to write a grammar for a specific application. The tutorial portion of this manual introduces the basic grammar writing skills, which are subsequently illustrated in a fully worked-out example grammar that serves as a natural language query and update interface to a simple semantic-network data base. Appendix B, a reference section, includes all the basic and advanced operators and features of DYPAR-I.
DYPAR-I is implemented in FRANZ LISP and runs under VAX UNIX or VMS/EUNICE. A COMMON LISP implementation has just been completed. DYPAR-I is a proper subset of other, more powerful parsers of the DYPAR family, culminating in the as-yet-unreleased DYPAR-IV, under development at Carnegie Mellon University. DYPAR-I with a fair-sized grammar interprets sentences in real time on a lightly loaded VAX-11-780.
This document is a combination tutorial introduction to DYPAR-I and reference manual.