Re: Updated specification of Concise Bounded Descriptions

Good to see; it's a decent read. Nice to see I haven't got to change  
any code, too :D

I question the terminology in "Alternative Forms of Description":

--
and that concerns the use of symmetric properties. E.g. consider the  
properties dct:hasPart and dct:isPartOf. For any statement 'X  
dct:hasPart Y' we can infer 'Y dct:isPartOf X', and visa versa.
--

that's an inverse property pair, not symmetric. Symmetric, as regards  
properties, has very well defined meaning! A symmetric property is  
one such that

x p y => y p x.

You get exactly the same situation with inverse and symmetric  
properties in a CBD, but with a symmetric property you infer _the  
same property_, while with an inverse property you infer _its  
inverse_. (i.e. a symmetric property is its own inverse.)

I think you're trying to refer to properties for which a  
bidirectional pair of properties is inferable, but only one is  
present in the source graph. Symmetric isn't the right word for this  
situation: intuitively it works ("going both ways"), but it has  
unintended semantics in this domain.

-R

PS. I've deleted the rdfig cross-post for this reply.

On 4 Jun 2005, at 18:15, Patrick Stickler wrote:


>
> For those interested, there is an updated member submission
> from Nokia providing a revised and expanded specification of a
> Concise Bounded Description (CBD), with added discussion
> addressing questions and issues that have be noted since the
> original submission.
>
> C.f.  http://www.w3.org/Submission/CBD
>
> Cheers,
>
> Patrick
>
> --
>
> Partrick Stickler
> Nokia, Finland
> patrick.stickler@nokia.com
>
>
>
>
>
>
>

Received on Sunday, 5 June 2005 10:12:07 UTC