Ginsberg's email@example.com (Anthony K. Sarris)
Date: Tue, 16 Aug 94 12:14:11 PDT
Content-Type: text/plain; charset="us-ascii"
From: firstname.lastname@example.org (Anthony K. Sarris)
Subject: Ginsberg's note
Cc: email@example.com, firstname.lastname@example.org
I thought I'd take a stab at trying to clarify a few issues raised in Matt
Ginsberg's recent note to the srkb list:
>Date: Mon, 15 Aug 94 15:51:36 PDT
>From: Matthew L. Ginsberg <email@example.com>
>To:firstname.lastname@example.org>>Subject: standardization of semantics
>I just saw a posting about this that included the remark
>> The US is looking to Sowa, Genesereth and Fulton to prepare a unified
>> position on CLF based on Conceptual Graphs (CGs), Knowledge
>> Interchange Format (KIF) and the STEP Semantic Unification Meta-Model
>I'm not sure who in the US is "looking to" these guys. My opinion
>remains that the Sowa/Genesereth/Fulton gang is leading the effort by
>self-proclamation and not by virtue of technical merit. Sowa
>champions CGs, Genesereth KIF and Fulton presumably SUMM.
ANSI X3T2, the US Technical Advisory Group (TAG) to the ISO/IEC
JTC1/SC21/WG3 Conceptual Schema Modelling Facilities (CSMF) rapporteur
group is 'looking to' those guys for a particular purpose. They are
actively supporting (through their participation) the US effort to help
develop an international standard for conceptual schema modelling
facilities. Quite awhile ago the US proposed basing the syntax of
conceptual schema modelling languages on symbolic logic -- starting with
FOL and aspects of set theory -- and extending beyond that as necessary.
Logic was chosen for the syntax because it is neutral with respect to the
contents it expresses. Several existing and emerging logic-based languages
were looked at. X3H4, the US TAG for the ISO Information Resource
Dictionary System (IRDS) had been studying CGs, and to a lesser degree SUMM
and KIF. X3T2, the US TAG for ISO communications/data interchange
standards, was studying KIF and SUMM. Other standards development
organizations (e.g., the newly formed X3J21) are working with Z and VDM. As
the US joined in the new international project on CSMF, it seemed to make
sense to pool our expertise and resources. Genesereth, Sowa and Fulton had
been informally coordinating their efforts already, and decided to make
this coordination a bit more rigorous. Additionally, US conceptual schema
efforts have recently been combined under a single TAG, the newly
reconstituted X3T2, which is now known as Information Interchange and
Interpretation. ISO CSMF has also adopted the notion of a logic-based
syntax for the emerging international CSMF standard. In order to put
forward a US contribution in this area, X3T2 is drawing upon the work of
Sowa in CGs, Genesereth in KIF, and Fulton in SUMM, as well as others. The
X3J21 people currently have their platter full with standardization of Z
and VDM as formal descriptive techniques under various ISO SC22 Programming
Languages projects. However, X3T2 is coordinating with them, and staying
abreast of developments with Z and VDM.
It should be made clear though that standardization of CGs, KIF, SUMM, Z or
VDM is not exactly the same as that portion of the CSMF standard that deals
with CS syntax or language. X3T2 and ISO CSMF looked for a good 'standard'
source for this level of logic, and no single source was deemed sufficient.
So the idea was to draw upon CGs, Z, KIF, SUMM, etc. There may well be some
aspects of the CG, KIF, Z syntaxes that are simply not relevant to the CSMF
effort. That's perfectly fine. Each has its purpose, following or
community, etc. and they can continue to use their language formalism as
they see fit. X3T2 intends to take what is necessary and useful from those
languages and use it to formulate a candidate CS syntax. The Common Logic
Foundation streamlines the effort in that it gives us all a better idea of
what common syntax is at the core of all of these logic-based languages.
Once the standard is published, any formalism that conforms to it (at
whatever conformance levels are specified) will be equally 'standard'.
Various methods and products may extend their capabilities beyond those
specified in the standard. 'Those guys' above are not getting any special
treatment over anyone else who participates in the open standards process.
It seems to insult other participants to imply that we'd let ourselves be
railroaded by those three. Just ask them how easy it is to get the rest of
us to agree on anything. But yes, they are contributing good work that
they've all done -- some of which will get used and some of which won't. If
their participation puts them in a better position to develop and sell
standards-compliant products, then, hey, that is part of how the standards
process in supposed to work. That's part of the benefit, and makes up for
sharing things in an open environment rather than competing with a bunch of
closed, proprietary offerings (which none of the end users want to see
I should also be clear that the logic-based syntax is only one-piece of the
CSMF standards puzzle. There is an overall architecture of what a CSMF is.
And, most importantly (at least I think), there is a normative conceptual
schema that contains a set of semantic constructs sufficient for general
conceptual modelling, enterprise model integration, and in the longer run
full content model interchange and interpretation (what Fritz Lehman has
nicknamed Electronic Knowledge Interchange or EKI).
So let's not fight too much about syntax. We can have reasonable technical
arguments and agree to what makes sense for all or most of us. Let's adopt
a workable, neutral, extensible syntax and agree upon a base set of
semantic (ontological) constructs and get an initial CSMF standard
published in the next couple of years that: 1) vendors can implement to and
2) end users can gain some benefit from, through the application of
products based on it.
A few additional comments:
>I know of
>*no* fielded AI application that uses any of these methods, although
>it is obviously possible some small efforts have escaped my attention.
>I know of no significant efforts *at all* that use KIF, with the
>exception of a couple of single-scientist research acgtivities. Why
>is there this rush to standardize on things that aren't even in actual
The standards process has changed in the last few years, from one of
standards adoption to standards development. The old days when one or two
large vendors came in with a candidate standard based largely on an
existing product of theirs is over. If we wait now until a product already
has substantial market share to standardize on it, it is too late. The
standards development process now is trying to strike an admittedly
delicate balance between R&D (which we are not intending to do) and
existing commercial off-the-shelf products. That balance often involves
taking the workproduct of R&D organizations, consortia, etc. as sources and
trying to craft a standard based on them. Something that meets industry's
needs, has sufficient proof of technical viability, and which does get too
much in the way of vendor creativity and individuality. It is not easy. and
it is why standards development organizations need all the help they can
>In any event, I can probably make it to the September meeting in
> Matt Ginsberg
>As noted above, new 'experts' (the official standards-world name for
>contributors, participants or whatever) are always needed and welcomed.
>Differences of opinion are certainly allowed, but the intent is to reach
>consensus as much as possible (in the voting system, minority 'no' votes do
>not stop the process, but they receive considerable and careful attention).
>The important rule of thumb is if you don't like what you see, then it is up
>to you to propose an alternative and win the support of the committee through
>technical argumentation (and, yes, sometimes politics -- but not too much we
22941 Mill Creek Drive
Laguna Hills, CA, USA 92653