On Tue, 2007-02-27 at 17:31 +0100, Axel Polleres wrote: > Hassan Aït-Kaci wrote: > >> > >> hmm... only one Atom in the 'then' part. That's > >> weaker than SPARQL's CONSTRUCT, if I'm reading this > >> correctly. That seems kinda odd. > > > > > > This is meant to represent only Horn clauses, Dan. > > BTW: if you want to model full SPARQL construct, you need a lot beyond > RIF Core, so SPARQL coverage is, I guess rather a phase 2 thing. > if you restrict yourself to basic graph patterns and *bnode-free* > CONSTRUCT parts, then you can well translate SPARQL's construct into > RIF, see no prob here, you can just split of multiple triples patterns > in the SPARQL CONSTRUCT part to multiple rules with the same body. There's a lot of SPARQL stuff that I'm not interested in... I was just thinking of rules a la graph-pattern => graph-pattern. But I can see how existentials (bnodes) in the conclusion are tricky. > Anyway, I think we should define a RIF dialect which covers SPARQL, > as pointed out in > > http://lists.w3.org/Archives/Public/public-rif-wg/2007Jan/0058 > > already, as I think it would be kinda nice to have a SPARQL like > exchange syntax for RIF rules on top of RDF data. Yes, interesting idea. > axel > -- Dan Connolly, W3C http://www.w3.org/People/Connolly/ D3C2 887B 0F92 6005 C541 0875 0F91 96DE 6E52 C29EReceived on Tuesday, 27 February 2007 22:04:22 GMT
This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:33:37 GMT