Re: issue: Identifying type of model

Hi Joel, 

your issue is now logged as issue 3 at
http://www.w3.org/2002/ws/sawsdl/issues/#x3

I feel there is a hidden sub-issue here:

What are we identifying in modelReference - a thing, a part of a model
(like a class in OWL or concept in WSML or something similar) or a whole
model (an OWL file, a WSML file etc.)? I believe the current draft leans
towards the former, i.e. a thing defined by the model. If that's the
case, then a thing can be described using multiple languages (like Eric
was pointing out yesterday with some kind of indirection) and so SAWSDL
could have a hard time actually saying what language is used for a
particular pointer.

Even if the URI identifies a part of the model, resolving it can still
lead to the description of the full file, for example
http://example.org/ontology#class identifies a class, but when resolving
this, the client will actually read http://example.org/ontology

Assuming here that modelReference does identify a thing (a part of a
model), I would go with the intent of your option 1 - our spec doesn't
say anything, the client will have to see it to know if it can
understand a particular modeling language; this will allow multiple
languages describing the same thing and the client choosing whatever it
understands best.

Please see more comments below.

Jacek

On Tue, 2006-04-18 at 15:26 -0400, Joel Farrell wrote:
> Hi All,
> 
> The modelReference attribute points to a concept in some semantic model and
> SAWSDL is independent of the language used to express that model.  The
> question is: Do we need to identify the language as part of the annotation?
> In other words, can a tool that is processing the WSDL file determine what
> kind of a model (OWL, WSML, ODM...) is at the URI pointed to by the
> modelReference?  I can think of three answers:
> 
> 1. Yes, it can read the document and see if it can recognize it.  This
> would require no change to our spec.

The web also gives us things like media types so the client can do an
HTTP HEAD on the URI and see what media type it has, and this should say
whether it's an OWL file, WSML file or something else.

> 2. Yes, but only if the URI includes a file extension like ".owl".   Is
> this a reasonable restriction? If so, our spec need not change.

I don't think the Web people would like this, URIs should be mostly
opaque to the users, and it's good practice not to include .owl (or
similar things) in URIs in fact, so that the URI needn't change when the
language choice changes.

> 3. No, it needs to be explicitly stated.  This could be done via a
> modelType attribute that pairs with the modelReference attribute or it
> could be specified once per interface or once per WSDL document
> (definitions).  If something like this is needed, it will have an
> implication if we decide that a modelReference can be a list of references.
> 
> The spec currently does not restrict a WSDL file from using a different
> type of model for each individual annotation. (It is silent on the issue.)
> Once this issue and the multiple models per modelReference issue is
> resolved, the spec will have to be explicit about this.
> 
> Regards,
> Joel Farrell
> 
>

Received on Wednesday, 19 April 2006 13:16:31 UTC