Browse Prior Art Database

Method for Storing And Mapping Expert Information for Knowledge Based System

IP.com Disclosure Number: IPCOM000120775D
Original Publication Date: 1991-Jun-01
Included in the Prior Art Database: 2005-Apr-02
Document File: 2 page(s) / 54K

Publishing Venue

IBM

Related People

Moon, CS: AUTHOR [+2]

Abstract

Disclosed is a method for storing expert knowledge in a matrix format that would facilitate fast and efficient access to the knowledge base. This method is useful when implementing a large expert system application in a small hardware platform. It is currently being used in the "Quality and Reliability Expert System (QRES)" which was initially designed for the IBM PC/XT*.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 93% of the total text.

Method for Storing And Mapping Expert Information for Knowledge Based
System

      Disclosed is a method for storing expert knowledge in a
matrix format that would facilitate fast and efficient access to the
knowledge base.  This method is useful when implementing a large
expert system application in a small hardware platform.  It is
currently being used in the "Quality and Reliability Expert System
(QRES)" which was initially designed for the IBM PC/XT*.

      Although all expert information regarding a manufacturing
process may be stored in matrices, a defect correlation
multidimensional matrix is the one that links defects from one sector
to the next guiding the knowledge propagation.

      The figure represents a typical manufacturing process by
sectors and demonstrates the correlation matrix transforming a set of
information produced by the sector (k) into information that may be
useful by the sector (k+1).  In other words, Entry X(k)ij correlates
the contribution of information regarding item (i) in sector (k) to
the information regarding (j) in sector (k+1).

      This method is used in the SPQL prediction algorithm of QRES to
correlate the defect distribution and contribution from one process
sector to the next.  The table above shows a hypothetical example of
a 4 x 5 defect correlation matrix for the sector (R).  In this case,
defect type 3 in sector $ indicated as DEF (3, R) contributes with
0.1 certainty to defect type 1 in sector R+1 (DEF...