Transaction models; ontological "tools"

fritz@rodin.wustl.edu (Fritz Lehmann)
Date: Tue, 1 Aug 95 00:06:02 CDT
From: fritz@rodin.wustl.edu (Fritz Lehmann)
Message-id: <9508010506.AA07418@rodin.wustl.edu>
To: bsr@premenos.com, fritz@rodin.wustl.edu
Subject: Transaction models; ontological "tools"
Cc: cg@cs.umn.edu, edi-new@tegsun.harvard.edu, srkb@cs.umbc.edu
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk


     Nik Mitschkovetz wote in the edi-new list:
----begin quote---
Thanks Fritz:
The reasons you mentioned are exactly why I subscribed to this list.
What type of tools would be reqired to implement an Electronic
Commerce model as defined by Dan Schutzer in some of the Internet XIWT
documentation? The FSTC "Electronic Commerce Project Workshop" held in
Oct, 94 presented this idea of modeling the economic transactions
involving the exchange of  goods and services between two or more
parties (buyers, sellers, and agents). The model developed the notion
that commerce could be represented by the following activities:
        Diccovery
        Negotiation
        Ordering
        Billing
        Payment
        Distributon
        Accounting
        Ancillary Services
Not all the functions are required and their order may change
depending on the transaction. What is needed is a framework to held
define the interaction between the transactions.
Does this look familiar to anyone? Reasonable? Bad Idea? Good Idea?
----end quote----

     It looks very reasonable to me, although I suspect that the
Devil is in the Details, and I'm wondering how far this project
went with this kind of work; did they elaborate sub-levels of the
list phases?   I have not seen any of this work --- could you please
post to edi-new some way to get the report or proceedings of this
effort?

     I've been concentrating more on the semantics of data elements and
value-codes (which in EDI mostly are about WHERE, WHEN, and details
about how "STUFF" -- the product or money -- is moved from one location
to another).  However, this has to be done while keeping in mind the
role played by the data elements in just the activity-phases you list
above.  It seems that these are ways to create an itch (a legal obligation)
and ways to scratch the itch (move some goods, services, records or
money in satisfaction of the obligation).   The "tool" for modelling this,
in my view, would be a combination of: A. a logic-based representation-language
which can define any meaning of a particular data element or value code (some
candidates are Conceptual Graphs, KIF, HOL or Cyc-EL), and B: a universally
intelligible set of basic semantic "building-blocks" from which to
build up the definitions.  The latter depend on the conceptual structure of
the real world -- what is a CONTAINER, a VESSEL/SHIP, a PAYMENT-DATE,
a POST-CODE, an ACCOUNT, etc.  Defining and deciding upon the underlying set
of building blocks is the hard part -- you have to be rather philosophical
about it, and people have a hard time agreeing on such things.  This
kind of effort has already been done in the Artificial Intelligence
field with examples like the Japanese EDR, the Pangloss project, the
Summary Schemas Model (Roget-based) and the Cyc project.

    Examining the existing EDI standards EDIFACT and X12, I find lots
of rich treasure troves of useful concepts, but without any formal
definitions (hence no way to integrate automatically with similarly
annotated application programs).  I'm hoping that the BSR (Basic Semantic
Repository) project has come up with the needed building blocks, but
I have yet to see the fruits of their latest effort.  I hope it will be
disclosed soon.

     [On this list I expressed a few worries about ISO 11179 and the BSR.
These worries were dismissed as "mis-statements", "stupid" and "idle gossip"
(though later privately confirmed to me as basically well-founded, by three
insider sources), so I will just wait and  see whether the BSR is
suitable for principled, composite semantics for EDI elements and code
values.  I'm hopeful.]

     The short answer, though, is that the "tools" you ask about don't
exist yet as far as I know.  The EDI community in general is not too
receptive to such an "excessively theoretical" approach as I've described,
so the solution may come from elsewhere.  They are businesspeople with
more immediate business needs to worry about, and they think old-EDI
(traditional EDIFACT and X12) is really high tech stuff already.

     Again, I'm crossposting this to the relevant email lists, and
any replies should also be CC:'d to those lists.   For those who asked
privately, there is another site for the ISO 11179 standard for data
semantics, including the draft ontological "object taxonomy" (which was not
adopted as far as I know): FTP from speckle.ncsl.nist.gov.  In the version
I saw, the effort to force the taxonomy into a tree made it seem tediously
repetitive, which would be cured by using a "product of posets" approach.

                          Yours truly,   Fritz Lehmann
GRANDAI Software, 4282 Sandburg Way, Irvine, CA 92715, U.S.A.
Tel:(714)-85-0671               email: fritz@rodin.wustl.edu
=============================================================