Re: RDFTM Disposition of Comments (RDFTM-DC001)

* Steve Pepper <pepper@ontopia.net> [2005-03-24 11:17+0100]
> 
> * Steve Pepper
> |
> | We are talking about the *base* RDF model in which a resource can
> | only have a single URI as an identifier.
>  
> * Dan Connolly
> |
> | Er... no, that's just not the case. A resource can have any
> | number of URIs as identifiers.
> 
> I don't think we disagree (I hope :) but there is clearly an
> issue of terminology that needs to be sorted out here.
> 
> I may have this completely wrong and, if so, correct me, but
> my understanding is as follows:
> 
> 1) Resources in RDF are represented by nodes
> 2) A node may have at most one URI reference (some have none)
> 3) The URI reference of a node is an identifier for the resource
>    represented by the node
> 4) In the absence of a vocabulary beyond that defined by RDF
>    itself, a resource can thus only have a single identifier.

I believe we prefer to say that, from an RDF perspective, resources 
can have many URIs; nodes in an RDF 1.0 graph however can be labelled
with up to one URI. RDF's graph-syntax constraint doesn't change the
world: in the world, things have multiple names. Hence OWL identity
reasoning etc to help systems figure out when two nodes denote the same
thing.

Dan
 
> Clearly an RDF graph may contain multiple nodes (with different
> URIrefs) that are *intended* to represent the same resource, but
> in the absence of a property such as owl:sameAs, there is no way
> for an application to know that they do in fact represent the
> same resource.

But RDF has no problem acknowledging that fact of the matter: things have 
multiple URIs. Which is why it was important to build OWL etc on top of 
RDF...

Received on Thursday, 24 March 2005 11:07:11 UTC