Re: Draft datatypes message

I will admit up front I have done a poor job of following the datatypes 
discussion, but perhaps that will allow me to point out places where 
things are not clear.

On Friday, June 21, 2002, at 09:12  AM, Brian McBride wrote:
> It is not possible to have the answers to Test B, Test C and Test D all 
> be yes.  Either B and C can be yes or D can be yes.  We have to decide 
> which of these is the most important to have; (B and C) or D.

It seems to me that we would not want B and C to be yes. We would want B 
alone to be yes. It would be helpful to explicitly state that this is 
not possible.

A nit is that QNames in N-Triples should not have <>s around them, 
because then they look like new URI schemes. I think it would be clearer 
to say:
    <ageInYears> rdfs:range xsd:decimal .

Another question that comes to mind is whether we can have:

Test D2:
   <Jenny>      <ageInYearsDecimalNumeral> "10" .
   <ageInYearsDecimalNumeral > rdfs:range xsd:decimal .
   <ageInYearsDecimalNumeral>  rdfd:abstract <ageInYears> .
   xsd:decimal rdfd:concrete xsdr:decimal .

   <John>  <ageInYears>      _:a .
   _:a     <xsdr:decimal>   "10" .

Which seems to avoid the overloading of the previous. Then I could use 
simple inference rules to transform the former into the latter. This is 
how I approach the similar problem in something like authorName vs. 
author.

N3 Rule:
{?thing ?prop ?num . ?prop rdfs:range ?range . ?prop rdfd:abstact ?abs . 
?range rdfd:concrete ?valprop } => { ?thing ?abs _:a . _:a ?valprop 
?num } .

Otherwise this seems to be a good summary.
--
Aaron Swartz [http://www.aaronsw.com] 
4FAC4838B7D8D13FA6D92EDB4145521E79F0DF4B

Received on Friday, 21 June 2002 10:43:57 UTC