Re: Cannes entailment, a question:

On Tue, 25 Jun 2002, pat hayes wrote:

> >On Tue, 25 Jun 2002, pat hayes wrote:
> >
> >>
> >>  >in two parts.
> >>  >
> >>  >One, for symmettry, where a literal "really is" a string,
> >>
> >>  Well, that question is now moot, since a literal is *never* a string.
> >>  A literal is a 3-tuple. Thus:
> >>
> >>  >do we have an
> >>  >analagous situation to the Cannes entailment?
> >>  >
> >>  >	eg:book dc:title "the big book of RDF" .
> >>  >
> >>  >entails...
> >>  >
> >>  >	eg:book dc:title _:a .
> >>  >	_:a xsd:string "the big book of RDF" .
> >>
> >>  No, that cannot possibly be valid. The thing inside those quotes is
> >>  either not a string or not a literal. Im not yet quite sure which is
> >>  correct: maybe its neither of them.
> >
> >Hang on, you've spotted the get-out (that the thing in quotes is a
> >Literal, but not an xsd:string), but given that I fail to see why the
> >above entailment should not hold.
>
> Because the value of dc:title can always be chosen to not be an
> xsd:string and still satisfy the first triple. Those interpretations
> will block the entailment.
>
> >Unless the datatyping proposal gives a
> >uniform treatment to _all_ xsd datatypes, I can't see how we can vote in
> >favour of it.
>
> Well, I agree. If the literals in the graph really are 3-tuples, then
> ALL of the datatyping debate - all of which was under the false
> assumption that literals were items in datatype lexical spaces, ie
> strings - has been rendered moot. We simply don't have datatyping in
> RDF graphs, with these literals: it's meaningless. The 3-tuple
> <1,"10","french"> isn't in the lexical space of any XSL datatype.

Urk! I had read the datatyping proposal with the notion that

	jenny eg:age _:a .
	_:a xsd:int "10" .

used some f:Literal->XSD lexical space such that f(<1, "10", null>) =
"10" - ie, that trivial, unlanguage-tagged, unxmlised literals were
being used for their mapping to the XSD datatype lexical space.

Apologies for terminological confusion in that last sentence.

> >If we have to accept non-regularity in datatyping then
> >there are far simpler approaches that support the xsd types "natively"
> >and other types using a daml-like construct that seem much more natural.
>
> I'd like to know the details of them. We have considered the daml
> ideas before, at length, and they were resoundingly rejected by the
> WG.

This is an aside, so I'll follow up with it separately.


-- 
jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/
Tel +44(0)117 9287088 Fax +44 (0)117 9287112 RFC822 jan.grant@bris.ac.uk
I am now available for general use under a modified BSD licence.

Received on Tuesday, 25 June 2002 11:39:36 UTC