Re: datatypes and MT

Pat Hayes wrote:
[...]
> >>  >If we want to say "my shoe size is some
> >>  >integer whose decimal representation is '10'",
> >>
> >>  But I don't want to say that. I want to say that my shoe size is 10.
> >
> >Yes, but we're not here to design a syntax; we're here to
> >clarify the existing one. And the existing one doesn't have
> >that expressive power, I don't believe.
> 
> Is
> 
> Pat shoeSize "10" .
> 
> illegal RDF, in the current syntax? (Assuming that 'Pat' and
> 'shoeSize' were urirefs, of course.)? Because I read that as saying
> that my shoe size is ten.

That is legal RDF/xml syntax (or rather: it corresponds
to RDF/xml syntax in a way that the WG has agreed on),
but the way you read it is exactly
the issue we're debating. I read it as saying that something
denoted by the uriref Pat is related by a relation
denoted by the uriref shoeSize to a two character string
'1' '0'. i.e. it corresponds to
	(shoeSize Pat "10")
in KIF.

The shoeSize relation might be defined in such a way
that allows you to conlude that your shoesize is ten.

[...]
> >The usual interpretation of "10" -- e.g. in KIF -- is
> >a string of two characters, no? I don't see what's unusal
> >about what I'm suggesting.
> 
> OK, Ive been assuming that the quote marks around literal labels are
> only a syntactic device for marking them as literals, not intended to
> be interpreted as actual quotation marks. They are an inheritance
> from XML, right?

Not in any particular way; n-triples syntax is independent
from XML.

> And XML is completely sloppy about use and mention,
> and uses quotation syntax to mean all kinds of things.  If we are
> obliged to interpret those as genuine quote marks,

we are not obliged

> then I give up;
> there is nothing to debate: all literals are character strings, end
> of story. But then if I write
> 
> Pat shoeSize "10"
> 
> then I am saying that my shoe size is a character string, right?

That's my suggestion.


>  From here on I will omit the double quotes to avoid confusion.

er... huh? You'll have to tell me how your new syntax
relates to RDF/xml syntax, then.

[...]

> >I don't see the simplicity anywhere.
> 
> Well, I do. The motivating example for me was the use of range
> information to fix a datatype, as in
> 
> aaa shoeSize 10 .
> shoeSize rdfs:Range xsd:integer .
> 
> which seems to me to be eminently simple.

That would be nifty if RDF/xml had such a syntax.
I believe/suggest that it does not.

So that example is irrelevant, from my perspective.

[...]
> >Yes, as I said, the mapping is unambiguous. This doesn't seem
> >all that awkward to me.
> 
> But if leading zeros are optional, the lexical-to-value mapping is
> not invertible, so what is its inverse supposed to be?

(a) as a relation, it *is* invertible; its inverse is
not a function, but it's a well-defined relation. (b) if
you need an inverse that's a function, the XML Schema
spec does give a canonical lexical representation for
each value.



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

Received on Wednesday, 7 November 2001 11:35:28 UTC