Re: [Fwd: Comments on SPARQL] (entailment, soundness, completeness)

Bijan Parsia wrote:
> On Sep 20, 2005, at 7:04 AM, Dan Connolly wrote:
> [snip]
> 
>>Enrico, elsewhere in your message about "Adoption of entailment in 
>>SPARQL"
>>of September 19, 2005 11:55:09 PM GMT+01:00, you wrote "here we don't
>>argue whether this is useful and how this is going to be used." Note 
>>that I
>>pretty much stopped reading at that point.
> 
> 
> I think you were mislead by Enrico's words there. There are plenty of 
> places in that note that he appeals to existing, documented SPARQL use 
> cases to motivate his technical points, e.g.,
> 
> """ON REDUNDANCY OF TOLD BNODES IN ANSWERS
> [issue <http://www.w3.org/2001/sw/DataAccess/issues#rdfSemantics>]
> 
> """
> Should queries of non-lean and lean graphs that entail each other
> give the same answers?
> """
> 
> The answer to this question should be *yes*. See use case 1,
> "Publishing on the Web", in
> <http://lists.w3.org/Archives/Public/public-rdf-dawg/2005JulSep/0430>).
> This is also relevant, as noted by PFPS, to enable interoperability
> between different interoperating implementations of RDF."""

The quoted email has two use cases - the same query is used on the same data 
in two separate situations.  The desired results are then different.

I can't tell if the proposed formulation reflects this or not - at the moment, 
I don't see any place where this is acknowledged.  Could someone kindly point 
such a place out to me, please?

	Andy

> 
> 
>>I'll be more motivated to study the technical
>>details when I know which user requirements, use cases, and 
>>applications
>>a proposal is intended to address and which it's not intended to 
>>address.\
> 
> [snip]
> Since the two major alternative approaches as I understand them 
> "virtual graphs" vs. entailment bases are, at least in some forms, 
> *equivalent* (I'm not clear pat means them to be anymore!) then the 
> whole first section of that email is motivate by spec clarity issues. 
> So, the use case *from my perspective* is producing interoperable 
> implementations for a variety of semantics imposed on the source 
> documents. For example, I may want to query the syntax of an RDF 
> document (i.e., with full asserted redundancy) for an editor 
> applicaiton, or I may want just the non-redundant information in the 
> graph (e.g., I don't want to have to post query filter out that someone 
> loving someone since mary loving john was already in there).
> 
> Cheers,
> Bijan.
> 
> 

Received on Tuesday, 20 September 2005 13:32:01 UTC