RE: Definition: Graph Pattern Matching

-------- Original Message --------
> From: Dan Connolly <>
> Date: 24 August 2004 17:58
> 
> I see some nice work on the core terms in the spec in
> sections 2.2 and 2.3
> http://www.w3.org/2001/sw/DataAccess/rq23/#GraphPatterns
> 
> I think getting those terms and definitions established
> will speed things up considerably.

And any help with defintions will be gratefully received.


> 
> Some reaction to $Revision: 1.39 $ of $Date: 2004/08/24 13:35:32 $
> 
> > let V be the set of variables
> > let B be the set of bNodes
> 
> I consider bNodes to be variables, so I'd rather:
> 
>  let V be the set of universal variables
>  let B be the set of existential variables
> 
> since RDF mt "effectively treats all blank nodes as having the same
> meaning as existentially quantified variables in the RDF graph in which
> they occur"

Good point.  bNodes are variables over individuals in the domain of
discourse.  query variables are variables over graph labels/terms.  I avoid
writing this into the doc as it is confusing.  May be it has to go in.

It also means, from the query point of view, bNodes are values/constants
just like everything else in the RDF graph (constant to the query - not the
RDF).

> 
> 
> > let A be U union L union V union B
> 
> those are traditionally called terms
>  http://www.earlham.edu/~peters/courses/logsys/glossary.htm#t
> so I'd use T there.

Some word to cover that would be good - I didn't find one in RDF Concepts
but I may have missed it.  Term is fine - using "T" is a bit confusing as
"triple" comes up a lot so even just swapping T to be term and something
else for Triple Pattern isn't going to be clear.   There isn't much use of
'term' after this definition so can we use a different-to-tradition symbol
here?

> 
> 
> Now we have a conflict with tradition here...
> 
> > let L be the set of all literals
> > let T be the set of triple patterns := A x A x A
> 
> Traditionally, a literal is something like P(x)...
> "These formulas are basically sets of clauses each of which is a
> disjunction of literals."
>  -- http://mathworld.wolfram.com/ResolutionPrinciple.html
> 
> Our "triple patterns" function much like literals. Hmm...

Sorry - you lost me a bit.  It is supposed to be literal as takne from RDF
concepts/MT. Would making it "RDF Literals" help?  It maybe the levels thing
again - variables over different spaces.

(I notice cofusion between the T as set of triple patterns the text saying
"T is a ground triple").

> 
> 
> I don't think I can parse the "Definition: Binding" section.

I'll try to rework it.  Any suggested text?  Either the formal part or
explanatory notes.

> 
> I'm inclinded to break out my larch tools to review this
> stuff.
> http://www.w3.org/XML/9711theory/
> http://www.w3.org/XML/9711theory/RDFAbSyn.lsl

That would be great.

Warning: the definitions become less solid as you go thought the doc - I
have tried to outline the direction (in text) when I haven't tried to give
formal, solid definitions.

	Andy

> 
> 
> 
> --
> Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Tuesday, 24 August 2004 17:43:38 UTC