motivation for bNodes/existentials in RDF; note for parsers

Pat, Dave,

I just explained to another colleague of
mine, Massimo, that RDF formulas are not
just sets of ground facts...
  (and
    (p1 s1 o1)
    (p2 s2 "lit2")
    ...
   )

but that RDF can express existential quantification...

 (exists (?b1 ?b2 ?b3 ...)
  (and
    (p1 s1 ?b1)
    (p2 ?b2 "lit2")
    ...
   )

Have you added some explanation about that
to the model theory spec?

Dave, I wonder if the syntax spec should
say something about "in the past, RDF
parsers have parsed anonymous nodes
by generating arbitrary URIs; don't do that;
make sure the parser client can tell the
URI references from the anonymous nodes".


Something about how if you've got two
documents that both say:

	<Book>
	  <author>Fred</author>
	  <title>ABC</title>
	</Book>

i.e.
  (exists (?x)
   (and
    (type ?x Book)
    (author ?x "Fred")
    (title ?x "ABC")
   )
  )

then each entails the other, but
if you skolemize each of them and got

   (and
    (type sk234234 Book)
    (author sk234234 "Fred")
    (title sk234234 "ABC")
   )

for the one document and

   (and
    (type sk7675 Book)
    (author sk7675 "Fred")
    (title sk7675 "ABC")
   )


for the other, then the
skolemized formulas don't ential each other.

I think I asked for this a while ago...
after somebody... Lynn Stein, I think...
asked me why we wouldn't just specify
anonymous nodes as getting skolemized
at parse time.

Hmm... I see some text under
 http://www.w3.org/TR/rdf-mt/#skolem

"However, care is needed, since these 'arbitrary' names have the same
status as any other urirefs once published. Also, skolemization would
not be an appropriate operation when applied to anything other than the
antecendent of an entailment. A skolemization of a query would represent
a completely different query."

I wonder if that covers it; Lynn? Massimo?

Pat, maybe a conrete example like the above
would be worthwhile?

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

Received on Friday, 22 March 2002 21:59:45 UTC