Re: Denotation of datatype values

At 09:27 AM 4/16/02 +0300, Patrick Stickler wrote:
> > I think that this way of phrasing it might be potentially misleading,
> > since it suggests that the pairings are actually in the MT.
>
>Perhaps they should be.

[GK wanders over to the stake in the ground, and gives it a hard kick, to 
make sure it's still firmly planted...]

>But we have been asked, to a certain extent, to provide an answer
>that extends to the point of obtaining a datatype value unambiguously
>and reliably. The present MT does not bring users to that point.
>A datatyped literal pairing does -- insofar as it identifies a single
>value which is obtainable by an application which groks the datatype
>in question.

Having an rdfd:range [[or substitute current vocab]] associated with a 
property tells you nothing about the denotation of an object of that 
property.  I.e. it does nothing to help "obtaining a datatype value 
unambiguously and reliably".  (I found that trying to make it do so leads 
to contradictions.)

All it does is limit the allowable literals at the property's sharp end.

#g



-------------------
Graham Klyne
<GK@NineByNine.org>

Received on Tuesday, 16 April 2002 08:35:44 UTC