[oslc-cm] repetition pattern of related properties (was: OSLC Change Management 2.0 Specification is in Convergence and Ready for further Review)

[Olivier Berger]

Hi.

Le dimanche 23 mai 2010 à 11:24 -0400, Steve K Speicher a écrit :
> > 
> > I was wondering why not define a single relationship property:
> relatedObject
> > of (tuple) type: (Resource, Type of Resource)
> > That way we can relate to any object as long as it is defined as a 
> > type of resource by the service provider. Similarly for the 
> > relationships other than "related".
> > 
> > We could even define the relationship into the relationship
> property: 
> > relatedObject of (triple) type: (Resource, Type of Resource, Type of
> > Relationship)
> 
> We had discussed these alternatives within the workgroup when defining
> these properties.  Mainly around the "oslc:manages" property, where it
> was discussed that is could be extended to other resource types:
> change sets, assets, requirements, etc.  Though since we are building
> on an extensible data model, we felt we didn't need to build
> extensibility into the relationship properties.  It had more value to
> define more semantics of the property by adding resource type
> information for a number of scenarios like query and reporting. 
> 
> One this we could consider, perhaps at a more general OSLC Common
> Properties, is the usage of Dublin Core's dc:related property.  Though
> by the nature of building off an RDF data model, we can already infer
> when resources are "related".   We haven't seen a strong scenario need
> for this general purpose dc:related property.
> 

In a Semantic Web approach, it's much better I think that relations
convey their domain-related semantics, if possible, right from the start
(provided there's concensus on what these mean), even though, in
implementation terms, these could all be inheriting from a more generic
kind of relation type.

Implementation wise, it's probably better to know if one particular
application needs to load other documents to get a larger view of a
problem, if the links are already descriptive enough, as it would be
some waste of processing, to access some document just to know their
type and finally decide to discard them ;)

Just my 2 cents a bit late ;)

Best regards,
-- 
Olivier BERGER <olivier.berger at it-sudparis.eu>
http://www-public.it-sudparis.eu/~berger_o/ - OpenPGP-Id: 2048R/5819D7E8
Ingénieur Recherche - Dept INF
Institut TELECOM, SudParis (http://www.it-sudparis.eu/), Evry (France)