W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > July to September 2006

[TO ADVANCE] formsOfDistinct

From: Bijan Parsia <bparsia@cs.man.ac.uk>
Date: Wed, 6 Sep 2006 12:19:51 +0100
Message-Id: <8FF6804C-4B5A-494E-AADD-88897BF06B20@cs.man.ac.uk>
To: RDF Data Access Working Group <public-rdf-dawg@w3.org>

http://www.w3.org/2001/sw/DataAccess/issues#formsOfDistinct

I believe some members of the working group believe that this cannot  
be settled without advances on the core semantics. I think, however,  
we can make some progress and perhaps decide on somethings we *don't*  
want.

So, it is clear to everyone, I hope, that there are several possible  
sense of DISTINCT. For a simple set of examples, see:
	<http://www.cs.man.ac.uk/~bparsia/2006/row-tutorial/slide#35> (and  
the subsequent 2)

(Now the exact effect of these different senses will partially depend  
on the BNode scope:
	<http://www.cs.man.ac.uk/~bparsia/2006/row-tutorial/slide#27> and  
subsequent

For example, pairwise subsumption only makes sense, IMHO, if we have  
answer level scoping in answers. That is, bnodes in answers only  
corefer *within* an answer, not between answers. There it makes a  
*lot* of sense.)

So, the high level point is whether we want to support 2 forms of  
DISTINCT or only one. This has the effect of having three levels of  
redundancy, e.g.,:

	The normal case
	DISTINCT
	REALLYREALLYDISTINCT

(I hope these will nest. I *think* source lean answers are always a  
superset of answer lean answers, but I've not shown it.)

Since I don't believe the group will endorse answer scope, then I  
believe there are only two senses of distinct at play. Source lean  
and answer set lean. Pat and Andy have championed some form of Source  
leanness. I've championed answer set leanness.

I think the group has enough information right now to decide whether  
they have one or two forms of DISTINCT, and if one, which one. If we  
decide on two, we need to add a keyword or other way of indicating  
which sense of distinct is in play. If only one, we can defer which  
one for a while if the group would like to gather more information.

Oh, this only settles BNode redundancy. We still need to be clear on  
literal and blank redundancy.

I'm happy to provide more examples, clarifications, etc.

Cheers,
Bijan.
Received on Wednesday, 6 September 2006 11:20:08 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:27 GMT