RE: Consistency Issues in multiple modelReferences

Hi Tomas,

my leaning would be that a single document should be kept or viewed as
consistent. I haven't yet seen use cases for allowing independent and
potentially inconsistent annotations. 

I think that an annotator tool should be able to provide at least basic
consistency checks, but verifying ontology consistency is in general a
hard problem, since there may always be meaning in an ontology that is
not expressed explicitly and is thus hidden from any validation.

Best regards,

Jacek

On Mon, 2006-05-29 at 18:29 +0100, Tomas Vitvar wrote:
> Hi,
> Could modelReferences be understood as perspectives of people who will
> create these annotations (assuming that more then one person can create
> annotations for one WSDL element)? Or should the person who annotates WSDL
> element by multiple modelReferences keep in mind that these should be
> consistent? 
> 
> In the first case, it should be clear which modelReferences will be used for
> the processing. For the second case, may be there could be a consistency
> check provided by an annotator tool already...
> 
> Tomas
> 
> > -----Original Message-----
> > From: public-ws-semann-request@w3.org [mailto:public-ws-semann-
> > request@w3.org] On Behalf Of Jacek Kopecky
> > Sent: 29 May 2006 17:55
> > To: verma@cs.uga.edu
> > Cc: public-ws-semann@w3.org
> > Subject: Re: Consistency Issues in multiple modelReferences
> > 
> > 
> > Hi Kunal, John,
> > 
> > since it would be hard for us to define inconsistencies completely
> > (without formalizing what we point to using some kind of logics), I'd
> > suggest that we can note that in case an inconsistency is discovered by
> > the processing agent, the WSDL document with semantic annotations should
> > be treated as an invalid SAWSDL document, i.e. no action should be based
> > on information in this document.
> > 
> > What do you think?
> > 
> > Jacek
> > 
> > On Tue, 2006-05-23 at 11:56 -0400, Kunal Verma wrote:
> > > Hi,
> > >
> > > If we support multiple modelReference for a WSDL element, do we plan
> > > to add rules about consistency. Specifically, should someone be
> > > allowed to annotate an element with two concepts (conceivably from
> > > different models/languages but translatable to a common
> > > model/language) that may contradict each other?
> > >
> > > These may become more of an issue in the following contexts:
> > >
> > > a) multiple operation based discovery.
> > >
> > > b) composition.
> > >
> > > c) use of protocols that use state information.
> > >
> > > Thanks,
> > > Kunal Verma and John A. Miller
> > >
> > >
> > 
>

Received on Monday, 5 June 2006 15:41:42 UTC