Re: ISSUE-44 (duplicate quads): Duplicate quads generated by TriplesMaps [R2RML]

David,

On 31 May 2011, at 16:54, RDB2RDF Working Group Issue Tracker wrote:
> ISSUE-44 (duplicate quads): Duplicate quads generated by TriplesMaps [R2RML]
> 
> http://www.w3.org/2001/sw/rdb2rdf/track/issues/44
> 
> Raised by: David McNeil
> On product: R2RML
> 
> As far as I can see the R2RML spec does not describe the proper behavior in the case where TriplesMaps produce identical quads. I think a couple of sentences should be added to the spec to specifically say what should happen in this case.

Good point. I added at the end of Section 1.4 [1]:

[[
RDF graphs cannot contain duplicate triples. Placing multiple
equal triples into the same graph has the same effect as placing
it into the graph only once.
]]

Is this sufficient clarification?

Thanks,
Richard


[1] http://www.w3.org/2001/sw/rdb2rdf/r2rml/#mapping-overview


> 
> My understanding of the correct behavior is:
> * R2RML maps are to produce SPARQL RDF datasets.
> * SPARQL RDF datasets contain a collection of RDF graphs (many named and one default).
> * RDF graphs are sets of triples.
> * As "sets", RDF graphs do not contain duplicate triples.
> * Therefore SPARQL RDF datasets do not contain duplicate quads.
> * Therefore R2RML maps should not produce duplicate quads.
> 
> In particular looking forward to SPARQL 1.1 aggregates I believe this is important to specify.
> 
> Note, SPARQL queries run against the quads produced by an R2RML map can produce duplicate tuples.
> 
> 
> 

Received on Tuesday, 31 May 2011 18:25:04 UTC