Hi Jim --
You wrote
Could you post to the CUO-WG list your basis for
suggesting this [Executable Open Vocabulary English] is a candidate solution
for CDSI, so the group can explore and discuss it. Glad
to!
Herewith a note about this, inline, and also as a pdf
attachment.
I look forward to our further
discussions.
-- Adrian
Adrian Walker
Reengineering
Phone: USA 860 830
2085
--------------------------------------------------------------------------------------------------------------------------------------------------------------
Basis
for Executable Open Vocabulary English as a Candidate Solution for
CDSI
Note for the Cross Domain Semantic Interoperability Working
Group 20061115
Adrian
Walker
Reengineering
adriandwalker@xxxxxxxxxJim
has set up an excellent framework for our discussions. As I understand
it, the framework asks us to focus on the problem of interoperability, and to
come up with candidate solutions. The only limit he imposes on a
candidate is that, for interoperation of N stovepipe software systems, there
should be no more than 2N adapters. This rules out candidates that would
require N**2 adapters, one for each pair of stovepipes.
As covered in
our opening discussion on Nov 13th, most if not all, of the candidates so far
are ontologies in technical notations such as OWL. It is a feature of
such ontologies that they contain comments in free English that say what the
authors intended. However, since these are comments, they are thrown away by
any software that uses the ontologies. Thus, the semantic intent of the
authors is lost at run time, and cannot be easily communicated to the user
community.
As mentioned in our meeting on Nov 13th, there is some
emerging technology that supports a representation shift in the way that we
write down and use knowledge for 2N interoperability. The general idea
is that knowledge authors can work in a kind of Wiki environment, writing
their knowledge into their browsers in open vocabulary, executable English.
The technology is live, online [1,2]. Shared use is free.
To see
why it may be a good idea to shift to this more human level of representation,
consider the process of trying to make stovepipes interoperate using only
lower level, technical notations. Typically, a team will assemble around
a (virtual) whiteboard, and will discuss in English (or another human
language) the meanings of the respective stovepipe concepts, and how to map
them back and forth. A 2N consensus is arrived at, and is mapped into an
ontology in a technical notation, with English comments. The ontology is
then used as a basis for writing adapters for actual interoperation. When this
software reaches users, the only recourse they have in the face of a
counter-intuitive output is to look at the comments in the ontology and to try
to figure out whether the intent was correctly mapped into say, raw
OWL.
If, on the other hand, the team inputs its collective knowledge in
executable English into browsers, there are several advantages.
Knowledge can be input and tested by running it over sample data, in a tight
edit-run loop, without getting into details at the programming level.
Then, the supporting technology can carry through the semantic intent of the
authors all the way to the user level. In particular, any output of the
system can be explained, step-by-step, in English, at the level of a
non-technical business user.
For simple examples of this, please see
[3,4], and please also run the examples and look at the explanations.
(There's nothing to install, just point a browser to [2]).
While this
approach may seem radical, it actually complements [5,6,7] current work on
technical ontology notations such as OWL and RDF while integrating two extra
kinds of semantics [8,9]. It also appears promising for scalable
interoperation among SQL databases [10], and for interoperation between SQL
and RDF [5].
In summary, we can capture more interoperation semantics,
and deploy it immediately, using emerging technology to write down knowledge
for interoperation in open vocabulary, executable English. The candidate
technology for this is live online, and is free for shared use by anyone with
a browser.
Page 1 of 2
References
[1]
ww.reengineeringllc.com/A_Wiki_for_Business_Rules_in_Open_Vocabulary_Executable_English.pdf[2]
Internet Business Logic system. Online at
www.reengineeringllc.com
[3]
www.reengineeringllc.com/demo_agents/OntologyInterop2.agent[4]
www.reengineeringllc.com/demo_agents/SemanticResolution1.agent[5]
www.reengineeringllc.com/demo_agents/RDFQueryLangComparison1.agent[6]
www.reengineeringllc.com/demo_agents/OwlTest1.agent[7]
www.reengineeringllc.com/demo_agents/FeaReferenceModelOntology2.agent[8]
www.semantic-conference.com/program/sessions/S2.html[9]
Backchain Iteration: Towards a Practical Inference Method that is Simple
Enough to be Proved Terminating, Sound and Complete.
Journal of Automated Reasoning, 11:1-22
[10]
www.reengineeringllc.com/Oil_Industry_Supply_Chain_by_Kowalski_and_Walker.pdfPage
2 of 2