Re: datatypes and MT

Pat Hayes wrote:

>>  > I think we need to nail down the handling of simpler cases before 
>> getting
>>
>>>  too involved in arcane syntax options.  By simpler cases, I mean how 
>>> are we
>>>  to interpret simple RDF like this:
>>>
>>>     <rdf:Description rdf:about="#me">
>>>       <ex:shoeSize>10</ex:shoeSize>
>>>     </rdf:Description>
>>
>>
>> dirt simple:
>>
>>     <...#me> <...#shoeSize> "10".
> 
> 
> No no, wait a minute. You are having your cake and eating it. That 
> triple doesn't conform to your proposed idiom;



Indeed, that triple doesn't conform to my proposed idiom, but
that's because the RDF from which it's transcribed doesn't conform.
I don't want folks to write that sort of RDF. I'm not
trying to have my cake and eat it too. I want folks to write

<rdf:Description rdf:about="#me">
   <ex:shoeSize dt:decimal="10">
</rdf:Description>

or

<rdf:Description rdf:about="#me">
   <ex:shoeSizeNumeral>10</ex:shoeSizeNumeral>
</rdf:Description>


 >

 it uses the literal label 
> in object position on an arc not labelled with a datatype mapping. If 
> you allow this kind of triple and also your bNode idioms, you need to 
> somehow connect them.



Sorta.

 >

 Can we infer one from the other? Both ways?


no.

> If so, 
> the simple idiom (as you have here) is equivalent to the extended bnode 
> form, so why do we need the latter? If not, how do we manage to 
> establish any kind of inferential connection between them?


We don't.


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

Received on Thursday, 8 November 2001 14:40:08 UTC