Re: [regrep-semantic] ebXML Registry Profile for OWL

[Farrukh Najmi <Farrukh.Najmi@Sun.COM>]

ewallace@cme.nist.gov wrote:

Farrukh wrote:

  

Dear Colleagues,

Recall the proposal for creating an "ebXML Registry Profile for OWL" 
some time back as a normative spec deliverable from the SCM SC:

http://lists.oasis-open.org/archives/regrep-semantic/200506/msg00002.html

Asuman, Carl and I discussed this proposal and feel that the time is 
right for us to start this
profile specification  in earnest. Asuman, has kindly agreed to be the 
primary editor for the specification.
Thanks Asuman! Please count on me on all issues and help in any way that 
you need.

The tactics we propose is to leverage past work from the DAPD paper and 
recast the ideas into an ebXML Registry Profile template as exemplified 
by the draft
"ebXML Registry Profile for Web Services"
    

...

I have looked through the references given and still don't completely
understand what is being proposed.  First question is: 
  What is an ebXML Registry Profile?
  

From Wikipedia: http://en.wikipedia.org/wiki/Profile

"

In standardization, a profile consists of an agreed-upon subset and interpretation of a specification. Many complex technical specifications have many optional features, such that two conforming implementations may not inter-operate due to choosing different sets of optional features to support. Even when no formal optional features exist within a standard, vendors will often fail to implement (or fail to implement correctly) functionality from the standard which they view as unimportant. In particular, implementations of standards on mobile devices often have significant limitations compared to their traditional desktop implementations, even if the standard which governs both permits such limitations. Also, some writers of standards sometimes produce vague or ambiguous specifications, often unintentionally, but sometimes by intention. The use of profiles can enforce one possible interpretation.

Users can utilize profiles to ensure interoperability, and in procurement.

In some cases, profiles themselves can become standardised: for example, US-GOSIP, UK-GOSIP and the ISO ISP (International Standard Profile) series in the context of OSI networking, and the various mobile profiles adopted by the W3C for web standards.

"

My definition is that a profile is: "a normative specification that defines a standard way to extend or restrict a base specification that is the target of the profile."

From the references, it appears to be something similar to
a UML profile which specializes the ebRIM.  However, it uses bindings 
(mappings) instead of stereotypes.  From our previous round in SCM,
I think there is a pretty natural correspondence between some language
elements of OWL and language elements in the ebRIM.  However, there 
were some mistmatches, particularly when reusing Classification and 
ClassificationScheme.  Is one allowed to loosen cardinality constraints
on RIM elements as part of these bindings?  That would seem to be 
necessary as OWL ontologies are rarely taxonomies.
  

Not sure what a UML profile is but the propose \profile would simply define how the registry extension features are used to support
OWL ontologies.

See the following example:

ebXML Registry Profile for Web Service (draft):
<http://www.oasis-open.org/committees/download.php/14756/regrep-ws-profile-1.0-draft3.pdf> (pdf file)
<http://ebxmlrr.sourceforge.net/tmp/regrep-ws-profile-1.0.odt> (Open Office 2.0 source file)

See for example how standard WSDL discovery queries are specified in Discovery Profile chapter. Now imagine standard OWL Ontology discovery query in a similar chapter of
our profile.

Now you can apply tha same concept to other points of extensibility in ebXML Registry.

Restrictions may be less relevant to our profile spec.

The other question I have is about the UML that describes the language
supported by the profile (at least such a thing was in the Web Services 
profile provided as an example).  

What are you refering to here. Can you be more specific please.

This is essentially creating a metamodel 
of the target language depicted in UML.  The OMG Ontology Definition 
Metamodel (ODM) specification is already defining such a metamodel as a 
normative part of the spec. I would not want another standard to define a 
different one.  We had a similar situation with ISO Common Logic (ODM 
contains metamodels for a number of languages) which was resolved by 
incorporating the diagrams from ODM into the Committee Draft for CL.
  

I am lost on above and need some specific concrete example of what you are refering to here. Thanks.

-- 
Regards,
Farrukh

begin:vcard
fn:Farrukh Najmi
n:Najmi;Farrukh
email;internet:farrukh.najmi@sun.com
tel;work:781-442-9017
url:http://ebxmlrr.sourceforge.net/tmp/farrukhRacePointIcon.jpg
version:2.1
end:vcard