ANSI and ISO activities
sowa <sowa@turing.pacss.binghamton.edu>
Date: Sun, 6 Feb 94 15:23:23 EST
From: sowa <sowa@turing.pacss.binghamton.edu>
Message-id: <9402062023.AA03547@turing.pacss.binghamton.edu>
To: cg@cs.umn.edu, kr-advisory@ISI.EDU, srkb@ISI.EDU
Subject: ANSI and ISO activities
Cc: sowa@turing.pacss.binghamton.edu
During the past year, the ANSI and ISO work on conceptual schema
modeling facilities has been progressing quite well. As a result of
new projects and various reorgs and disorgs, the tasks and committee
assignments have been shuffled around. Following is the current
line-up of ANSI committees that are particulary relevant to the
Knowledge Sharing Effort:
X3H4 on IRDS (Information Resource Dictionary Facilities) has changed
its focus as a result of a new ISO work item on PCTE (Portable Common
Tools Environment). They have voted to associate their work with the
ISO PCTE efforts and to design the new repository standards to fit with
the directions PCTE is taking. This new direction will eventually require
a conceptual schema, but their first priority is to finish the definition
of the standards for the current PCTE before designing the new version.
Therefore, X3H4 voted to transfer the conceptual schema work to
another committee X3T2, which also has a standards project on
conceptual schemas for data interchange. Consequently, X3T2 now
has responsibility for two closely related standards projects: one
for data interchange and one for repositories. We had all agreed
that these two projects should be done in close liaison, and putting
them both in the same committee makes a lot of sense.
Meanwhile, the ANSI technical report on the IRDS Conceptual Schema
has finally been finished, and the camera-ready copy has gone to
ANSI for publication. The printing process will probably take a
couple of months, but if anyone would like a preliminary copy (without
the ANSI cover and binding), the Ontek Corporation has agreed to
provide copies at the cost of reproduction and postage. For each copy,
please send a check for $20 (or for $25 outside the U.S.) to
Ms. Dale Dement
Ontek Corporation
22941 Mill Creek Drive
Laguna Hills, CA 92653
This report is not a standard, but it has been approved as a base
document for a standards project on conceptual schemas. This project
had been assigned to X3H4, but it is now moving to X3T2, where it will
be developed in conjunction with the X3T2 project on conceptual schemas
for data interchange. (Note: the complete report is 256 pages long
with lots of diagrams, and it would not be practical to make it available
by FTP.)
Following is a paragraph from page 6 of the report, which summarizes
the recommendations and proposed directions:
The recommendation is to develop one or more standards for the
conceptual schema of the IRDS, specifically for an IRDS Normative
Schema. The standard must describe the constructs necessary for
integrating heterogeneous modeling languages, modeling paradigms,
database languages and other forms of representation, based on the
unification or subsumption approach to integration. The IRDS
Normative Schema will be expressible syntactically using any one
of several equivalent logic-based Normative [Schema] Languages.
At least one Normative Language, based on a logic-based modeling
language called Conceptual Graphs, will be used for the initial
standards work.
In referring to this report in a bibliography, I have been using
the following reference:
Sandra Perez & Anthony Sarris, eds., _IRDS Conceptual Schema_,
Technical Report X3H4/93-196, American National Standards Institute,
New York.
The number X3H4/93-196 is likely to be replaced by some other
number when ANSI officially releases the report.
For the conceptual schema for data interchange, X3T2 has been using
the KIF report by Genesereth and Fikes as a base document. In Jan. 1993,
we had a meeting of the KIF, conceptual graphs, and SUMM developers, where
we agreed to adopt a common semantics based on the KIF model-theoretic
foundations. Since KIF and CGs are being developed with a common semantics,
it should be possible for both of them to be developed in parallel to meet
the X3H4 recommendations of "several equivalent logic-based Normative
Languages." The word "equivalent" is intended to mean that formally
specified mappings that preserve semantics can be defined in both directions.
As readers of these mailing lists are aware, the formal foundations of
KIF and CGs have been questioned by several correspondents, including
Fritz Lehmann and Matt Ginsburg. Although I believe that the foundations
are both sound and adequate to satisfy the requirements, I recognize
that the concerns are legitimate and that we must develop the foundations
in a thoroughly analyzed and documented form. Furthermore, we must also
show that the foundations are adequate to support the "heterogeneous
languages and paradigms" discussed in the IRDS report. This is an
intellectual IOU that must be repaid before a proposed standard could
be accepted.
Meanwhile, a new ANSI committee has been formed, X3J21 Formal Description
Techniques. This committee was set up in response to a new ISO working
group, which was chartered to standardize the Vienna Definition Methodolgy
(VDM) and the Z specification language (Z is pronounced /zed/). Both VDM
and Z have been widely used in Europe for the formal definition and
specification of programming languages and software. Both of them are
based on first-order logic with ontologies for sets, sequences, state-transition
diagrams, and other features needed for program and software specification.
However, Z uses a conventional 2-valued logic, but VDM uses a 3-valued
logic with an undefined truth value.
Ideally, the languages used for data interchange and repositories
should be compatible with the languages used for specifications.
At the founding meeting for X3J21, Roger Burkhart and I proposed a
subgroup to look at the common logical foundations for VDM and Z
in order to establish a common basis that can be shared with KIF, CGs,
and other conceptual schema languages. The problems of reconciling
the 2-valued vs. 3-valued semantics of Z and VDM are closely related
to many of the foundational issues concerning KIF and CGs as well.
Another committee that is involved with related issues is X3H7 on
Object Information Management. Several of the X3H7 members attended
the founding meeting of X3J21, and they are interested in using Z as
a tool for specifying object models. The object-oriented languages and
systems raise a number of issues concerning contexts and their semantics.
Those topics have also been discussed in these mailing lists, and the
common logical foundation must be adequate to support contexts.
John Sowa