- From: Bijan Parsia <bparsia@cs.man.ac.uk>
- Date: Tue, 19 Sep 2006 08:35:31 +0100
- To: Eric Prud'hommeaux <eric@w3.org>
- Cc: Kendall Clark <kendall@monkeyfist.com>, Ivan Herman <ivan@w3.org>, "Ralph R. Swick" <swick@w3.org>, Dan Connolly <connolly@w3.org>, RDF Data Access Working Group <public-rdf-dawg@w3.org>, Sandro Hawke <sandro@w3.org>
On Sep 19, 2006, at 8:06 AM, Eric Prud'hommeaux wrote: > After going through the wordings with Ivan, Ralph, DanC and Sandro, > we came up with some new wording that are hopefully precise without > requiring too much context. I cc your coauthors so they can respond directly. > On Sun, Sep 17, 2006 at 06:53:05PM -0400, Kendall Clark wrote: >> >> >> On Sep 12, 2006, at 7:26 AM, Bijan Parsia wrote: >> >>> >>> http://www.w3.org/2001/sw/DataAccess/rq23/rq24.html >>> >>> I would prefer that: >>> Should DISTINCT be based on lean graphs? >>> Be phrased as >>> What is the definition of DISTINCT? >> >> I agree with this suggestion to the editors. > > "Should the keyword DISTINCT recognize logically equivilant graphs?" All the forms of distinct, except term-distinct, "recognize" (give the same answers to up to isomorphism) logically equivalent graphs. Now, it is true that I underemphasized term-distinctness (though I believe Andy's post mentioned it), but the choice isn't between term- distinct and (some other form), but between term-distinct, and source- lean distinct, and answer-lean distinct. So I think it's imprecise and actutally requires far more context than the simple "What is the definition" variants. >>> Should SPARQL care about graphs that are inconsistent by D- >>> entailment? >>> >>> as: >>> What are the answers of a query of a D-inconsistent graph? >> >> And with this one too. > > We preferred Kendall's wording > "What are the answers to a contradictory KB?" Yes that's fine. > though I am not sure the world will comprehend the interaction between > entailment and contradictory KBs... They don't need to. >>> Finally, I would prefer a different phrasing for: >>> """Many of these issues reduce to "Is SPARQL a graph query >>> language or a higher level query language?" """" >> >> I *really* dislike this current phrasing; it's far too tendentious to >> be useful. What is a "higher level query language" anyway? >> >>> But without specifying which issues do so and how, I think it's >>> more confusing than helpful. >> >> Agreed. > > We put a lot of thought into this one and arrived at > [[ > Many of these issues revolve around, "Should SPARQL be sensitive to > only the graph structure (per the 1st last call semantics) I object to the parenthetical. For example, DISTINCT is independent to a large degree to the underlying BGP semantics. Furthermore, DISTINCT was *unspecified*, that is, it didn't have a settled semantics in the 1st last call. So, I'm not sure that "many" of these issues revolve around that. I would raise it as a distinct issue thought. But it's not like settling "Query RDF syntax or semantics" will *settle* everything. It will be suggestive and directive, of course. [snip] > Given this, I removed this issue as well, yielding > [[ > Pending issues: > > * Should SELECTed variables be separated by commas? > > * Should the keyword DISTINCT recognize logically equivilant > graphs? As above, I object. > * Should SPARQL care about graphs that are inconsistent by > D-entailment? "Care about?" Obviously, on any approach, it "*cares*" about them. How about: What are the answers of a query against an inconsistent KB? (A la Kendall?) > * Should blank nodes be treated differently than variables in the > query pattern? > > Many of these issues revolve around, "Should SPARQL be sensitive to > only the graph structure (per the 1st last call semantics) or the > semantics of RDF graphs as well." The working group could use guidance > from the community on this point. See above. Kill the parens, and I would prefer "graph structure" be replaced with the more transparent "syntax". Cheers, Bijan.
Received on Tuesday, 19 September 2006 07:42:22 UTC