[
Lists Home |
Date Index |
Thread Index
]
[Jonathan Borden
> > Aside from this special case, and I suppose there are a few others
lurking
> > around, do you see other situations where it would be important to use
an
> > actual retrievable URI as an RDF node identifier?
> >
>
> Well actually that is the subject of TimBL's proposal to the TAG regarding
> the "meaning of rdf:Property URIs". One very reasonable way to define this
> would be to dereference the property URI and use what is returned to find
a
> definition for the property. Suppose the property URI -minus- fragid
> identifies a namespace. The namespace document might contain human
readable
> definitions of the properties defined 'within' the namespace. An OWL
> ontology obtained from the namespace might similarly define the
properties.
>
Hmm, and since all GETS to a hashed URL either go to the bare URL or are
located by a browser (for instance) within the document that has already
been loaded, you could argue that the hashed URLs are in fact real working
HTTP URLs (even if there are no anchors in the document on the server to
match the property URIs).
You could get the same effect by embedding into the rdf document one triple
that used the namespace URI as its subject and said what working URL to go
to for the property definitions. Then no one would have to get confused
about whether a property URI represented the concept of the bare URL.
Just another convention, but one that seems cleaner to me.
Cheers,
Tom P
|
