Computer verification of the completeness of a simulation problem description by natural language interaction

Abstract

A research project in natural language man-machine communication is currently being conducted at the Naval Postgraduate School. The system being developed, called NLPQ, is an application of a more general system, known as NLP, which consists of a rule language and the programs to compile and execute those rules. NLPQ currently consists of particular sets of NLP rules which allow a user at a time- sharing terminal to input an English text description of a queuing problem, have the computer construct an internal problem representation, and then have it produce an English text description of the problem and a GPSS simulation program to solve the problem. The research described in this thesis produced the INTERROGATOR, a set of NLP rules for inspecting the internal problem representation to insure that it is ready to produce a GPSS program. The INTERROGATOR produces questions about missing or erroneous information, which the user may then respond to in English. It may also be used in a question-answer mode to input the entire problem.