Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Enhancing Knowledge Base Security and Efficiency

IP.com Disclosure Number: IPCOM000104336D
Original Publication Date: 1993-Apr-01
Included in the Prior Art Database: 2005-Mar-19
Document File: 4 page(s) / 87K

Publishing Venue

IBM

Related People

Foster, NJ: AUTHOR

Abstract

Disclosed is a scheme that reduces the risk of an exposure or compromise of a knowledge database while improving the efficiency of knowledge processing.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Enhancing Knowledge Base Security and Efficiency

      Disclosed is a scheme that reduces the risk of an exposure or
compromise of a knowledge database while improving the efficiency of
knowledge processing.

      Data contained in a Knowledge-Based System (KBS) may be
sensitive, e.g., a patient's medical history or a pricing strategy.
For reasons of flexibility and maintenance, only the main inferencing
rules are held permanently within the system;  a large portion of the
knowledge base is stored in external databases.  The current trend is
towards deploying KBS on remote workstations where the risk of
information exposure or compromise is a serious possibility.  This
proposal reduces exposure in two ways:  tokens and encryption.

      TOKENS - The idea is to hold knowledge objects and
relationships once, as lexemes in some type of 'lexeme store', and
use 6-byte tokens to represent them thereafter.  This should both
save space and improve processing speed.  Because most of its
activity is concerned with searching (for matches, connections,
inferencing chains, etc), what these tokens actually represent is
irrelevant to the KBS.  Only when system users require details of
some inferencing conclusion, or the steps which led to a conclusion,
do the 'real' knowledge objects and relationships have to be
presented.  The format of the token is very simple and shown in Fig.
1.

For example, to represent a semantic network using triples in the
form:

     <object   relationship   object>
means handling varying length strings.  If the length of a knowledge
object/relationship can vary from, say, 1 to 100 bytes then each
triple can potentially be 300 bytes in length - and enough storage
must be set aside to...