RE: Namespace of ODRL

> -----Original Message-----
> From: Mo McRoberts [mailto:Mo.McRoberts@bbc.co.uk]
> Sent: Wednesday, July 17, 2013 12:16 AM
> To: Michael Steidl (IPTC)
> Cc: Renato Iannella; <public-odrl@w3.org>
> Subject: Re: Namespace of ODRL
> 
> 
> On 16 Jul 2013, at 11:49, Michael Steidl (IPTC) <mdirector@iptc.org>
wrote:
> 
> > Renato, I think it is an agreement that "2" is used as the major version
> > number.
> >
> > All:
> > Coming back to only one or more namespaces: a user of terms from this
> > namespace would like to know what a specific term is for - as Ray
> expressed
> > this by the pan and ingredients distinction. If ODRL has a machine
readable
> > definition of all these terms then it must be considered how to express
> such
> > a distinction.
> > Even in the current Vocabulary is no qualifier if a term should be used
with
> > Policy Type, Actions, Constraints, Party and Role, or Asset and
Relation,
> > such a distinction is currently only made by the tables in the human
> > readable HTML presentation.
> 
> So I'm inclined to agree, and certainly RDF has the means to express that.
> 
> As an alternative to the 'one namespace or two' question, here's an
> alternative proposal:
> 
> Split the vocabulary into (preferred prefix in parens):
> 
> - A namespace for the model (http://www.w3.org/ns/odrl/2/...)
> 
> - A namespace for ODRL-defined actions
> (http://www.w3.org/ns/odrl/2/actions/...)
> 
> - A namespace for ODRL-defined constraints
> (http://www.w3.org/ns/odrl/2/constraints/...)
> 
> - A namespace for ODRL-defined functions
> (http://www.w3.org/ns/odrl/2/functions/...)
> 
> - A namespace for ODRL-defined policy types
> (http://www.w3.org/ns/odrl/2/policies/...)
> 
> - A namespace for ODRL-defined relation types
> (http://www.w3.org/ns/odrl/2/relations/...)
> 
> - A namespace for ODRL-defined scopes
> (http://www.w3.org/ns/odrl/2/scopes/...)
> 
> The remainder - which includes the "base" classes such as v:Scope, as well
> as the operators, conflict terms and undefined terms - would be moved
> into the model (because re-defining those as an extensibility mechanism
isn't
> particularly useful).
> 
> While this is certainly a little more complex, it does mean that there's a
very
> clear split between things which constitute the *mechanics* of ODRL versus
> the various instances/subclasses/subproperties which make up the
> vocabularies, with each controlled vocabulary inhabiting its own namespace
> to make the distinction clear.
> 
> This would mean that, for example, v:Action would become odrl:Action
> <http://www.w3.org/ns/odrl/2/Action>, while v:acceptTracking would
> become act:acceptTracking
> <http://www.w3.org/ns/odrl/2/actions/acceptTracking>.
> 
> Each of the schema documents at
> {actions,constraints,functions,policies,relations,scopes} would reference
the
> model, but the reverse would not be true (i.e., the model is completely
> agnostic to the actual terms used, provided they are correctly-formulated,
> not only conceptually, but implementation-wise too).
> 
> How does this sound to people?

I fully agree, this split up is very close to what IPTC has done for its
news exchange formats: a namespace for the basic structure and for each
value vocabulary a specific namespace. Also the split up of the ODRL
vocabulary is ok, moving the operators to the basic structure namespace
makes a lot of sense.

Michael

Received on Wednesday, 17 July 2013 12:56:13 UTC