Re: Note on PROV-AQ about LDP

Hi Stian,

On Thu, Apr 25, 2013 at 9:06 AM, Stian Soiland-Reyes <
soiland-reyes@cs.manchester.ac.uk> wrote:

> (Speaking as one of the coauthors of PROV-AQ)
>
> I think our expectation was that the issue of identifying graphs within
> RDF Datasets ("named graphs") would somehow be addressed by this WG as part
> of forming best practices for RESTful access to Linked Data.
>
> http://www.w3.org/TR/sparql11-http-rdf-update/#protocol-model touches on
> the problem; a dataset might contain a graph identified by an IRI the
> server is not authorative for.
>
> Its solution of ?graph=<graphuri> is elegant, but of course only works for
> compliant servers, typically graph stores with a sparql endpoint.
>
> But if I retrieve an RDF Dataset from <http://example.com/dataset2> as
> say TriG, and find it contains graph <http://example.org/graph2>, then
> unless I have out of band information that <http://example.com/dataset2>
> implements Sparql HTTP updates, I am not at liberty to just mint the URI <
> http://example.com/dataset2?graph= http://example.org/graph2>, and so I
> am not able to refer to the graph IRI in my Prov AQ link. (At best I can
> refer to <http://example.com/dataset2 >)
>
> Given the current LPD spec, I agree it would probably be out of scope
> there, as it does not mention RDF datasets. It might still be within the
> scope of the LPD WG to publish a Note that addresses the graph
> identification problem beyond SPARQL graph stores.
>
Actually this was discussed to some extent in the LDP F2F [1] and there is
an action [2] to look in to this though it is doubtful whether it will be
feasible to publish a note to address the graph identification problem
within the current charter of the WG.

Best Regards,
Nandana

[1] - http://www.w3.org/2012/ldp/meeting/2013-03-13#line1354
[2] - http://www.w3.org/2012/ldp/track/actions/46