Browse Prior Art Database

INTERACTIVE CONSULTING VIA NATURAL LANGUAGE

IP.com Disclosure Number: IPCOM000148895D
Original Publication Date: 1974-Jun-30
Included in the Prior Art Database: 2007-Mar-30
Document File: 34 page(s) / 2M

Publishing Venue

Software Patent Institute

Related People

Shapiro, Stuart C.: AUTHOR [+3]

Abstract

Stuart C. Shapiro Stanley C. Kwasny Computer Science Department Indiana Univerasity Appeared (without appendices) in Communications of the ACM 18, 8 (August, 1975), 459-462- INTERACTIVE CONSULTING via Natural Language Stuart C. Shapiro Stanley C. KwasnyComputer Science Department Indiana University Bloornington , Indiana Abstract Interactive Programming systems often contain help commands t o give the programmer on-line instruction regarding the use of the var- ious systems commands. We argue that it would be relatively easy t o make these help commands significantly more helpful by having them accept requests i n natural language. A s a demonstration, we have pro- vided Weizenbaumls ELIZA program with a script that turns it into a natural language system consultant. Key Words and Phrases: interactive programming, time-sharing sys- tems, natural language processing, computer assisted instruction. CR Categories: 4.49, 3.79, 3- 32, 3.42. Introduction Many interactive systems include a mechanism for automatic dis- semination of information regarding the use of its commands. Typi- cally, the user gets this lnforniation by entering a basic "help" com- mand and providing the name of the command he wants inforniatiorl i?bout. For example, on the DECsystem-10 3 , the user may type HELP, and get information on the HELP commands; HELP*, and get the names of docu- mented features; or HELP, and get information on the feature . Figure 1 shows the results of typing HELP and HELP* on the system available at Indiana University.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 7% of the total text.

Page 1 of 34

INTERACTIVE CONSULTING via Natural Language

Stuart C. Shapiro

    Stanley C. Kwasny
Computer Science Department Indiana Univerasity

Appeared (without appendices) in Communications of the ACM 18, 8
(August, 1975), 459-462-

[This page contains 1 picture or other non-text object]

Page 2 of 34

[This page contains 1 picture or other non-text object]

Page 3 of 34

INTERACTIVE CONSULTING via Natural Language

Stuart C. Shapiro

    Stanley C. Kwasny
Computer Science Department
Indiana University Bloornington , Indiana

Abstract

    Interactive Programming systems often contain help commands t o give the programmer on-line instruction regarding the use of the var- ious systems commands. We argue that it would be relatively easy t o make these help commands significantly more helpful by having them accept requests i n natural language. A s a demonstration, we have pro- vided Weizenbaumls ELIZA program with a script that turns it into a natural language system consultant.

    Key Words and Phrases: interactive programming, time-sharing sys- tems, natural language processing, computer assisted instruction.

CR Categories: 4.49, 3.79, 3- 32, 3.42.

[This page contains 1 picture or other non-text object]

Page 4 of 34

Introduction

    Many interactive systems include a mechanism for automatic dis- semination of information regarding the use of its commands. Typi- cally, the user gets this lnforniation by entering a basic "help" com- mand and providing the name of the command he wants inforniatiorl i?bout. For example, on the DECsystem-10 [ 3 ] , the user may type HELP, and get information on the HELP commands; HELP*, and get the names of docu- mented features; or HELP

, and get information on the feature

. Figure 1 shows the results of typing HELP and HELP* on the system available at Indiana University.

    The problem with such help commands is that the user must know which command he wants information about. If, instead, he only knows what he wants t o do and wants t o find out the proper command t o use, he is reduced t o a sequence of guessing command names. Help commands should be more user-oriented, allowing the user t o describe i n his own terms what he wants t o do. The system would interpret. the request and provide information on how t o accomplish the desired task.

Interactive systems consultants (help commands) are excellent

\t "

applications for natural language understanding programs. Since the context which the systems consultant must deal with is limited, even unsophisticated natural language programs are capable of dealing with it. The ease with which such consultants may be programmed and their usefulness argue that large interactive systems be provided with natural language consultants.

A Natural Language Tonsultant

Lest the reader fear that we are proposing an extensive research

project rather than a program w e l l wlthin the state of t...