Re: datatyping: local and global

At 09:00 28/01/2003 -0500, Art.Barstow@nokia.com wrote:

>In Mike Dean's formal objection to the Datatyping Solution proposed
>in the Jan 23 LC WD docs [1], one of the alternate proposals he
>suggests is:
>
>[[
>an RDF datatyping approach that includes global
>datatyping (as in Proposal F from [4]) possibly with
>optional local datatyping
>]]
>
>Did the RDF Core WG consider such an option (to support both local
>and global datatyping)?  If so, why was it rejected?

By global datatyping, we mean here that we can say:

   a prop "10" .
   prop rdfs:range xsd:integer .

and an RDF processor will conclude:

   a prop "10"^^xsd:integer .

Yes it did, at considerable length.  This was the issue that caused the WG 
the greatest difficulty.

The short answer is that there is a choice.  If you have global datatyping 
as above, you cannot also have:

   jenny age "10" .
   johny age "10" .

and conclude that jenny and johny have the same age (you need a range 
constraint to draw this conclusion.)  The WG decided that it preferred to 
have the latter over the former, major concerns including the effect on 
current implementations and existing data.

Brian

Received on Wednesday, 29 January 2003 06:38:56 UTC