Browse Prior Art Database

Generic Knowledge Base Acquisition And Specification

IP.com Disclosure Number: IPCOM000121338D
Original Publication Date: 1991-Aug-01
Included in the Prior Art Database: 2005-Apr-03
Document File: 4 page(s) / 161K

Publishing Venue

IBM

Related People

Arnold, VD: AUTHOR [+5]

Abstract

Disclosed is an approach to gather information for a knowledge database in a concise and specific manner with clear relationships and meanings established for the data. This database contains both constant data and relationships, for the use of this data, in a set of tables that are then the specification of the required configuration. These tables are easily accessed by any programming language using its database query facilities. The relationships, in the tables, represent the necessary logic applied to the constant data to configure a processor system using a set of generic rules. The generic rules define the process and actions that will occur for all entries in the tables, independent of the specific device characteristics. Generic rules that are too complex for representation in tables are programmed directly.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 40% of the total text.

Generic Knowledge Base Acquisition And Specification

      Disclosed is an approach to gather information for a
knowledge database in a concise and specific manner with clear
relationships and meanings established for the data. This database
contains both constant data and relationships, for the use of this
data, in a set of tables that are then the specification of the
required configuration.    These tables are easily accessed by any
programming language using its database query facilities.  The
relationships, in the tables, represent the necessary logic applied
to the constant data to configure a processor system using a set of
generic rules.  The generic rules define the process and actions that
will occur for all entries in the tables, independent of the specific
device characteristics.  Generic rules that are too complex for
representation in tables are programmed directly.

      This articles to expert systems, such as the 9370 Hardware
Placement and Connection Services expert system configurator HP&CS,
as well as nonexpert system configurators, such as the newer 9221
Configurator.  The approach is not limited to configurator
implementation.

      The previous method of specifying the Knowledge database, as
described in [1,2,3] was by English rules. This method which is
typically described in Expert System literature for knowledge
acquisition is more of an art than a science.  Product and system
developers and engineering specifications were queried to develop the
English statements of what the process and product represented.  The
combination of these English rules then represented the system
knowledge base.  The English rules developed in this manner tended to
leave relationships undefined, vague or only related in the mind of
the expert.  Each device had different but similar English rules.
Terminology used in the English rules was inconsistent and unclear.
Different words were often used for the same actions and their
incompleteness left a lot to be interpreted.  As new devices were
added to the knowledge base, the specific rules grew and the amount
of programming and testing for them also grew.

      This article solves these problems by:
1.   Templates always request the information in the same way, thus
providing completeness of information and consistent meaning to the
information.
2.   Tables for the knowledge and specification are the same tables
used by the configurator for its data and logic.
3.   Fewer generic rules than specific device rules, thus less
programming.
4.   Generic rules are consistent and do not need to be retested as
new devices are added to the tables.
5.   Improved development cycle time to support a new device in the
configurator.
6.   Easier maintenance of the knowledge and code.
7.   Allows configurators with different goals to use the same data
and relationships to apply to their goals and achieve the same
results.

      Configuration parameters and a...