Re: RDF semantics redux

Steve Harris wrote:
> On Tue, Nov 08, 2005 at 01:17:41AM -0600, Pat Hayes wrote:
> 
>>But, I have to say, all this seems like overkill. The same effect can 
>>be got by sticking to the local option, and requiring the server to 
>>supplying a special URIref as an answer binding to indicate an 
>>'anonymous' entity, along the lines suggested by Andy in
>>
>>http://lists.w3.org/Archives/Public/public-rdf-dawg/2005OctDec/0159.html.
> 
> 
> I like this.
> 
> It is what 3store has done for some years - originally it was a bug, but
> I met with spritited objections when I said it was going to be fixed, so I
> can report that users like it.
> 
> It apears to be a natural thing to do when machine processing result to
> assemble further queries, like the offspring example you gave in this
> mail. That may have been a consquence of the pre-SPARQL XML results format
> that 3store used, which is less clear on the distinction between
> labelled bnodes and URIs.
> 
> - Steve
> 

Yes - users seem to like it.

ARQ supports (A) via the syntax  <_:label>.  When used locally (via the 
programmatic API), ARQ returns the graph node itself (including nodes in 
not-so-virtual triples) so futher fine grain API calls or further SPARQL 
queries can be made.

We'd need to tweak the language of the XML Results format as it scoped bNodes 
labels to the document for network use.  Maybe a parameter in the protocol to 
ask for this mode or leave to service description.  That is, we need the first 
part of (C).


ARQ also supports (B) via parameterized queries which Leigh kindly wrote up:

   http://www.ldodds.com/blog/archives/000251.html

The parameterization is outside the syntax.

(this works because you pass in real graph nodes and the query parser made 
bNodes in the syntax into variables before query execution starts).

Both these separate told-bNodes from query variable bNodes.

	Andy

Received on Tuesday, 8 November 2005 12:19:46 UTC