Browse Prior Art Database

Context Based Next Command Prediction

IP.com Disclosure Number: IPCOM000105422D
Original Publication Date: 1993-Jul-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 131K

Publishing Venue

IBM

Related People

Kevern, B: AUTHOR

Abstract

A method for using contextual and historical information to predict the next command a user will enter is disclosed. The underlying mechanism adapts to the user to best predict what will next be entered.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 49% of the total text.

Context Based Next Command Prediction

      A method for using contextual and historical information to
predict the next command a user will enter is disclosed.  The
underlying mechanism adapts to the user to best predict what will
next be entered.

      As a user enters commands on the system, one can see that the
commands entered tend to be related.  For example, if a user's task
was to write a C program, one would probably guess that the set of
commands most likely enter by the user would include those that run
the editor, compile the program and invoke the newly created program.
The set of commands most likely entered by a user performing a
particular task reflects the context within which the user will work.

      The assumption used in this implementation is that first,
rather than hard coding a prescribed set of contexts within which a
user may execute, it is possible to learn the different context by
monitoring a user's interaction with the system.  Second, it is
assumed that after the system has gathered enough contextual
information, that this information may be applied to predict the next
command the user will enter.

      Learning the different contexts a particular user might execute
within is achieved by monitoring the user's interaction with the
system.  As the user executes commands, the system attempts to
establish the user's CURRENT CONTEXT.  This is achieved by storing in
a list the last N commands the user has executed.  The size of the
list (or N) reflects the number of commands the system will use to
try and establish the user's current context.  As a new command is
entered by the user, it is saved in the list and the eldest command
in the list is discarded.  In this manner, the system continously
updates what it believes is the user's current context.

      The information in the current context is used to update a
structure called the CONTEXT DATABASE.  The context database reflects
all of the knowledge the system has gathered about the contexts the
user may work within.

For example, if the system detected the user entering the following
list of commands:

1.  STRSEU
2.  CRTCPGM
3.  CALL

      the current context would be equal to (STRSEU, CRTCPGM, CALL)
where STRSEU was entered first by the user, followed by CRTCPGM and
finally CALL.

      The system, upon receipt of this current context, would update
the context database by adding the following two lists to the
database:

  (CRTCPGM,CALL) and (STRSEU,CRTCPGM,CALL)
thereby capturing the fact that the CALL command was once used after
the CRTCPGM command the that the sequence of STRSEU, CRTCPGM and CALL
has been seen before.

      If the user later chose to enter the STRSEU command again, the
current contex would be equal to (CRTCPGM, CALL, STRSEU) and the
system would add the following two lists to the database:

  (CALL,STRSEU) and (CRTCPGM,CALL,STRSEU)

      As can be seen by this example, the system gathers con...