The Planes System: Natural Language Access to a Large Data Base

, , , , , and

A prime obstacle for non-technical people who wish to use computers has been the need to either learn a special language for communicating with the machine or communicate via an intermediary. We feel that the time is ripe for computers to be equipped for natural language systems that can be used by persons who are not trained in any special computer language. In order for such systems to be of value to a casual user, the systems must tolerate simple errors, must embody a degree of "common sense," must have a relatively large and complete vocabulary for the subject matter to be treated, must accept a wide range of grammatical constructions, and of course, must be capable of providing the information and computations requested by the user. We are developing such a system called PLANES (for Programmed LANguage-Based Enquiry System) at the University of Illinois Coordinated Science Laboratory. PLANES includes an English language front end with the ability to understand and explicitly answer user requests and to carry on clarifying dialogues with him, as well as the ability to answer vague or poorly defined questions. We are also building a library of associated programs which includes functions for recognizing patterns within the database and for alerting a user when certain patterns of data occur which are of interest to him. This work is being carried out using a subset of the U.S. Navy 3-M database of aircraft maintenance and flight data, although the ideas can be directly applied to other record-based databases, both military and non-military.


  • 17205148 bytes

natural language processing

TechReport

University of Illinois, Urbana Champaign

Coordinated Science Laboratory

T-34

Downloads: 169 downloads

UMBC ebiquity