Coalition of logic-based standards email@example.com
Date: Fri, 15 Nov 91 20:20:34 EST
To: SRKB@isi.edu, INTERLINGUA@isi.edu, KR-ADVISORY@isi.edu
Subject: Coalition of logic-based standards efforts
Several different organizations are developing standards for information
interchange, each for a different purpose, but with similar requirements.
These include the following:
a) For database and CASE tools, the ANSI X3H4 Committee on IRDS
(Information Resource Dictionary Systems) approved a standard
in 1988 that was based on Entity-Relationship Diagrams. The
corresponding ISO IRDS committee never approved the ANSI standard,
because they considered it too limited. The ANSI X3H4.6 Task Group
is now working towards a new logic-based standard using predicate
calculus and conceptual graphs.
b) For manufacturing data, the PDES committee (Product Data Exchange
using STEP) is also working towards a logic-based specification
language called SUMM (Semantic Unification Meta-Model).
c) For knowledge based systems, DARPA is sponsoring research leading
towards standards for Shared Reusable Knowledge Bases. The proposed
Knowledge Interchange Format (KIF) is a version of predicate calculus
with a LISP-like notation.
These groups are not completely independent, since some people are
members of both X3H4 and PDES, and I have been serving as a liaison
between X3H4 and SRKB. In order to establish closer cooperation among
these groups, Tom Gruber of the SRKB group attended the PDES meeting in
October. On November 12, I visited Richard Fikes and Tom Gruber at the
Stanford Knowledge Systems Lab to discuss collaboration between ANSI
and SRKB. In the afternoon, we were joined by several others, including
Mark Tuttle from UMLS, Bill Mark from Lockheed, Udo Pletat from the
LILOG project at IBM Stuttgart, and Thomas Sharp from the TIRS project
at IBM Santa Teresa. As a result of the discussion, we reached a
consensus on the following points:
1. Symbolic logic, in its various forms, is the best developed, best
understood, and most precisely formalized knowledge representation
language available. We should work towards a common logic-based
standard for the IRDS, PDES, and SRKB projects.
2. Most of the work to be done for these standards efforts is in
developing the ontologies and conventions for applying logic to
all of the content that must be represented in information systems,
manufacturing systems, databases, and knowledge bases.
3. KIF or whatever it evolves into looks like a good basis for
the predicate calculus version of the standard. If IRDS and
PDES have requirements that KIF does not yet support, then we
should work to revise or extend KIF rather than propose a rival
4. Conceptual graphs or whatever they evolve into look like a good
basis for the graphic version of the standard. If there are
features or requirements that CGs do not yet represent, then we
should work to revise or extend them to make them semantically
equivalent to whatever KIF evolves into.
5. Besides conceptual graphs, there are more specialized graphic
notations such as type hierarchies, entity-relationship diagrams,
NIAM diagrams, dataflow diagrams, Petri nets, pert charts, and other
forms that have proven to be useful for many kinds of applications.
Whenever possible, we should support these notations by ensuring
that they can be translated into the normative versions of KIF
and conceptual graphs.
6. We should work with the existing standards organizations, such
as ANSI, PDES, and ISO, in order to gain as wide an acceptance as
possible for the logic-based standards.
7. But these standards should be presented as a basis for enabling
knowledge sharing rather than a restriction that would prohibit
the development of new theories, languages, and technologies that
may be needed for special purposes and applications.
To close the loop with PDES, I sent a copy of these seven points to
Jim Fulton from Boeing, who is the principal author of the PDES/SUMM
report. He agreed that they provide a good working basis for future
collaboration. Richard Fikes said that he would make plans to attend
the next ANSI X3H4 meeting in January in order to discuss KIF and the
DARPA-sponsored knowledge sharing effort and to establish a basis for