The issue is stated as:
“Should we create an RDF/OWL based meta model
view?”and the main arguments advanced that “This would
provide a better way to leverage
semantics to create new types of artifacts, etc.,
allow clients lacking full S-RAMP knowledge to get more
value out of these documents. It would also provide an
alternative description of the spec using RDF/OWL.”
I have a couple of concerns:
- I have no problem with people developing a
separate ‘profile’ or implementation using OWL or RDF but we
need to remain clear as to which format is the reference
format and ‘core standard’ – and the format chosen should
reflect our data model’s needs not the other way around (I
explain further below);
- I’m not convinced that the argument based on
consistency with the OpenGroup SOA Ontology holds water:
I’ve been through it superficially so can’t judge in detail
but I know that it did receive a mauling at the hands of
some people whose opinions I value in this area. I don’t
think that S-RAMP should remain orthogonal to the SAO
Ontology but it is too early to assess whether there is any
value in establishing a relationship;
- I’m not convinced either that a rendition in OWL
provides for a better way to leverage semantics – there are
too few real-world OWL tools, particularly in this domain,
for this to be true – while a very large portion of
SOA-based work can, for all its shortcomings, work with
schema;
- As regards the requirement to allow clients to
gain more knowledge (presumably, and more correctly,
‘information’) from these assets, a standard and protocol
exist already, SD-Share (‘Semantic Description Share’). It
has been approved by the European Standards Agency, CEN, and
is currently being considered by ISO (in ISO/IED JTC1-
SC34): http://www.egovpt.org/fg/CWA_Part_1b.
I would argue for re-use of existing tools
I won’t go into more philosophical reservations
about the merits of OWL as I think that debate is premature
and possibly redundant. The focal point of the TC is to
produce a common data model for SOA repositories and to a
large extent that can and should be done initially without
reference to any particular data modelling language: we need
first to establish that the model is fit for purpose and
then decide what representation best captures that, not the
other way around!
Best regards,
Peter
From:
Linda Terlouw [mailto:linda.terlouw@icris.nl]
Sent: Monday, 14 March, 2011 07:10
To: s-ramp@lists.oasis-open.org
Subject: [s-ramp] SRAMP-3 RDF/OWL vs XML Schema
Hi all,
Attached you can find a document about XSD versus OWL for
issue SRAMP-3. If you have any comments, please let me know.
@Vince, can we discuss this in the next call?
--
Met vriendelijke groet/Kind regards,
ICRIS B.V.
Consulting and Research
Linda Terlouw
Enterprise Architect
Mob: +31 (0) 6 24 380 962
E-mail: linda.terlouw@icris.nl