From SPARQL Working Group
Marcel Otto said:
> I'm having some difficulty understanding the intention of the SPARQL > Service Description property sd:availableGraphs. > > The current Working Draft says sd:availableGraphs "relates an instance of > sd:Service to a description of the graphs that may be used in the > construction of a dataset either via the SPARQL Protocol, with FROM/FROM > NAMED clauses in a query, or with USING/USING NAMED in an update request." > > Aren't the graphs, which one can refer to with FROM/FROM NAMED clauses in a > query, the sd:names of the sd:namedGraphs of a sd:Dataset? Is > sd:availableGraphs only an alternative way to describe the same fact?
The graphs referenced as part of a sd:Dataset are a part of the *default* dataset. That is, they are the graphs available for use with the GRAPH keyword in a query when no dataset is specified using FROM/FROM NAMED or in using the protocol.
The FROM/FROM NAMED keywords used in a query, or the default-graph-uri and named-graph-uri parameters used in the protocol, specify a dataset for the query to use that overrides the default dataset. In some systems, the graphs that may be specified in this way to construct a dataset is limited to a pre-defined set. The sd:availableGraphs property is available to publish this pre-defined set of graphs to the client.
To help clarify this issue, the definition of sd:availableGraphs has been updated in the Editor's Draft of the Service Description document to say:
""" Relates an instance of sd:Service to a description of the graphs which are allowed in the construction of a dataset via the SPARQL Protocol, with FROM/FROM NAMED clauses in a query, or with USING/USING NAMED in an update request, if the service limits the scope of dataset construction. """
We would be grateful if you would acknowledge that your comments have been answered by sending a reply to this mailing list.
Regards, Gregory Williams, on behalf of the SPARQL WG.