Re: Shapes, Individuals, and Classes - OSLC Motivations

On 11/12/2014 7:44, Arthur Ryman wrote:
> Holger,
>
> I am not following your suggestion. Please clarify at the telecon
> tomorrow.

I will.

> My brief response is:
>
> RDFS does not express constraints.
>
> owl:imports has semantics defined by the OWL spec, which does not express
> constraints.
>
> I don't see how named graphs apply. In OSLC, we follow Linked Data which
> means resources have RDF representations, i.e. sets of triples. Named
> graphs are part of RDF Datasets so you'd need to use Trig or N-Quads.

My brief response is that named graphs are useful to build union graphs 
consisting of a controlled subset of all graphs for a given task. 
owl:imports is the best established means to say that one graph uses 
another, regardless of the fact that it is part of the OWL spec. We have 
used named graphs for similar scenarios as you describe for many years. 
Trig or N-Quads are just serializations, while the actual implementation 
may be something like a Jena MultiUnion graph. Applications receiving 
RDF as a payload have any freedom to select the graphs that they need 
for execution of constraints. And no-one stated that RDFS expresses 
constraints - that's the gap that this WG is trying to fix (if we ever 
get any conclusion on anything, which I start to doubt).

Holger

Received on Tuesday, 11 November 2014 22:40:18 UTC