Re: [regrep-semantic] PR2: Explicit support for OWL Ontology/Class inplace of ClassificationScheme / ClassificationNode

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

ewallace@cme.nist.gov wrote:

>Zachary Alexander wrote:
>
>  
>
>>I think that this violates ebXML version 1.06 requirements.  The ebXML
>>registry should be payload neutral.  I think that this should trigger a
>>change in the charter of this subcommittee.  I think the charter should
>>be changed to explicitly state that this subcommittee is dedicated to
>>creating an OWL based ebXML Registry.
>>    
>>
>
>I share Zachary's concerns about practical effects of basing the whole 
>registry on OWL.  
>
Where do you get the impression that PR2 implies basing the whoile 
registry on OWL?
It simply says that a Registry needs to allow using of OWL ontology / 
classes where
ClassificationScheme/ClassificationNode are being used currently.

Another PR could suggest the same for a different Ontology syntax.

>It was also not clear to me at the last SCM SC telecon, 
>what the extent of the proposed use of OWL really was.  All of these things 
>need more exploration before a commitment to a design decision like this is 
>made.
>  
>
We are not making any design decsions yet. We are simply brainstorming 
on what we
perceive as key requirements for our eventual spec to address.

>On the other hand, I don't see that allowing use of an OWL Ontology in
>place of ClassificationScheme and allowing reference to OWL Class where
>ClassificationNodes are currently used implies creating an OWL based ebXML
>Registry.  
>
Indeed it does not. Sorry if PR2 is misleading that way. Suggets verbage 
improvements
now that I offer an explaination.

>The most significant change to the RIM suggested by this 
>requirement is removal of the constraint on ClassificationScheme-like things 
>to have a tree structure.  The words "allow use" in PR2 suggests to me
>that ClassificationScheme and ClassificationNode will not be deprecated
>by supporting OWL Ontologies and classes in these places in the RIM.
>  
>
My current thinking is that we may leave ClassificationScheme and 
ClassificationNode
around for backward compatibility but deprecate their use. I could be 
convinced of
removing them if our final proposal makes them unnecessary (which I 
think it will).


-- 
Regards,
Farrukh