W3C home > Mailing lists > Public > public-prov-wg@w3.org > April 2012

Re: PROV-ISSUE-319 (dgarijo): Domain of hasAnnotation [Ontology]

From: Timothy Lebo <lebot@rpi.edu>
Date: Fri, 13 Apr 2012 19:23:37 -0400
Cc: Daniel Garijo <dgarijo@delicias.dia.fi.upm.es>, Provenance Working Group <public-prov-wg@w3.org>
Message-Id: <C57A3B12-8C51-40CC-A30B-5D15D5380075@rpi.edu>
To: "Miles, Simon" <simon.miles@kcl.ac.uk>
Simon,


On Apr 13, 2012, at 3:44 AM, Miles, Simon wrote:

> Hi Daniel, Tim,
>  
> OK, I understand this is another RL-related issue and we do the best we can to meet all constraints. You are welcome to close the issue.
>  
> It is probably too late to consider at this stage, but would it be solvable by having a class X that is a super-type of Entity and Activity? I've raised the possibility of such a class with individuals for a variety of reasons, e.g. X would be what one asks the provenance of (rather than just entities).

we had prov:Element in prov-o for a while, borrowing the term from DM.

Naming it explicitly and having the two subclasses entity and activity is one "work around" for not having unions, though it's not logically equivalent.

It faded away in favor of simplicity. Also, applying this "work around pattern" in other areas of the ontology led to some awkwardly-named classes hanging around that distracted from the fundamentals (e.g. "Roled" class had subclasses Association, etc. that were allowed to have property prov:hadRole, when a minCard 1 from DL would suffice. Even asserting the minCard 1 on those classes that CAN have hadRole does not PREVENT anything else from having itů. OWA axioms just shouldn't be used to enforce CWA expectations.).

What we're left with is an ontology that can't be used to guide how to write its instance data. We have to rely upon a pile of documentation and examples. 

-Tim

Tracker, tuck this under ISSUE-265


>  
> Thanks,
> Simon
>  
> Dr Simon Miles
> Senior Lecturer, Department of Informatics
> Kings College London, WC2R 2LS, UK
> +44 (0)20 7848 1166
>  
> Mapping Dublin Core to the Open Provenance Model:
> http://eprints.dcs.kcl.ac.uk/1386/
>  
> From: dgarijov@gmail.com [dgarijov@gmail.com] On Behalf Of Daniel Garijo [dgarijo@delicias.dia.fi.upm.es]
> Sent: 08 April 2012 20:10
> To: Timothy Lebo
> Cc: Provenance Working Group
> Subject: Re: PROV-ISSUE-319 (dgarijo): Domain of hasAnnotation [Ontology]
> 
> Hi Simon,
> this issue is now pending review.
> Are you ok with Tim's answers and proposals? Can we close it?
> (it seems more related to DM than PROV-O).
> 
> Thanks,
> Daniel
> 
> 2012/3/15 Timothy Lebo <lebot@rpi.edu>
> Simon,
> 
> On Mar 14, 2012, at 8:16 PM, Provenance Working Group Issue Tracker wrote:
> 
> > PROV-ISSUE-319 (dgarijo): Domain of hasAnnotation [Ontology]
> >
> > http://www.w3.org/2011/prov/track/issues/319
> >
> > Raised by: Simon Miles
> > On product: Ontology
> >
> > What does it mean that hasAnnotation does not have a specified domain
> > (my ignorance of RDFS)? If it means that it applies to anything,
> 
> yes, any rdfs:Resource can be described using prov:hasAnnotation.
> 
> prov:Entity and prov:Activity are two subtypes of rdfs:Resource that can be described using prov:hasAnnotation
> 
> The largest reason for not having the domain defined is that we can't have an owl:unionOf in OWL-RL.
> 
> 
> > then
> > what is the distinction between using hasAnnotation and just giving an
> > arbitrary non-prov RDF statement?
> 
> Do you have an example of non-prov RDF statement that brings you concern?
> 
> 
> 
> > What is its connection to
> > provenance?
> 
> I think this is a concern on DM, not the ontology.
> I'd suggest:
> 
> 1) clearing up your rdfs:domain concerns and reassigning this ISSUE to DM
> or
> 2) start a new ISSUE on DM about it's connection to provenance.
> 
> Regards,
> Tim
Received on Friday, 13 April 2012 23:32:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 13:07:03 GMT