Re: axiom URIs

On 5 Feb 2007, at 17:10, Matthew Pocock wrote:

> Sorry - my last post displayed a degree of pique. Didn't mean to  
> throw my
> teddy. Let's start again.

Sure.

> I would like a mechanism to identify an axiom, seperate from any  
> other axioms
> it is associated with.
>
> I would like that identity to be representable as a URI,  
> independent of the
> URI of any OWL document.
>
> I would like that identifier to be independent of the structural  
> identity of
> the axiom, and ideally independent of any structure of the axiom,  
> though the
> converse may not hold (i.e. structural identity may depend upon the
> identifier).
>
> I would like it to be possible to do this totally without going to a
> representation of the axiom in /any/ serializtion format.
>
> I would like any details about how that URI may be associated with  
> axoims with
> different structural identity or different semantic import at  
> different times
> to be up to each community to decide.

I don't particularly agree with these requirements. At least, not the  
full set. I'll leave it at that.

> As for how this gets serialized in XML or in the functional  
> version, I'd
> prefer it to go on the axiom XML as an attribute, but would settle  
> for it
> being a well-known and well-understood annotation defined in a core  
> OWL
> namespace.

If the former, there's no way for it not to be problematic with  
respect to other serializations, e.g., RDF/XML. I don't think  you'll  
get very much support for it, if recent discussions on this list are  
any indication. Hmm. Although, I suppose you could use some specific  
attribute in xml and have it get parsed in the functional syntax to  
an annotation (and thence, straightforwardly, to RDF). I think that's  
arguable and might satisfy everyone.

Cheers,
Bijan.

Received on Monday, 5 February 2007 17:29:05 UTC