Browse Prior Art Database

Translation of Term Subsumption Language into Conventional Rules

IP.com Disclosure Number: IPCOM000109487D
Original Publication Date: 1992-Sep-01
Included in the Prior Art Database: 2005-Mar-24
Document File: 3 page(s) / 131K

Publishing Venue

IBM

Related People

French, SW: AUTHOR [+3]

Abstract

Term subsumption languages are widely advocated as a better way to solve many types of expert system problems. They can be used to build parts of an expert system in a more systematic and verifiable way than in rules. This invention allows term subsumption to be used with existing commercial rule-based shells by translating a term subsumption representation directly into conventional rules and allowing subsumption definitions to be encoded in a "rule-like way".

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

Translation of Term Subsumption Language into Conventional Rules

       Term subsumption languages are widely advocated as a
better way to solve many types of expert system problems.  They can
be used to build parts of an expert system in a more systematic and
verifiable way than in rules.  This invention allows term subsumption
to be used with existing commercial rule-based shells by translating
a term subsumption representation directly into conventional rules
and allowing subsumption definitions to be encoded in a "rule-like
way".

      The language allows "terms" to be specified.  Each term, which
is either a concept (an atomic item) or a relation (list related
concepts), may specialize another term; that is, the language is a
type of term subsumption language.  Each term may have a definition
which is a constraint that each instance of that term must satisfy.
Automatic translation of this language into conventional rules is
made possible by requiring definitions to be written in a form that
is equivalent to the left-hand side of a conventional rule (this form
will vary from one specific rule language to another).

      For each term, rules are generated which:
           1. Infer an instance of term whose definition is satisfied
by that instance.
           2. Infer an instance of a parent term for each instance of
a child (a child is a term subsumed by the parent).
           3. Retract any instance whose definition is no longer
satisfied.

      The features of this approach are:
           o Integration of term subsumption-based and rule-based
programming.
           o Translation of term-subsumption representation into
conventional rules.
           o Automatic fact consistency (truth maintenance).

      A summary of the language syntax is as follows:
      TERM ::=(concept/relation) ID
                (specializes ID)
                {if relation then the list of related concepts is
given}
               (and is denoted by "<any string of text>")
                (and is defined by DEFINITION)
           end (concept/relation)
      DEFINITION ::= INSTANCE_NAME such that CONDITION
      INSTANCE_NAME ::=(ID : CONCEPT_NAME/(ID,..,ID) : RELATION_NAME)
      CONDITION ::={an expression following the syntax for a
left-hand side condition in CLIPS, where the variables are
INSTANCE_NAME}

      The "is denoted by" clause allows an English language
translation to be associated with each formal de...