Re: bnodes

=?UTF-8?B?UmV0byBCYWNobWFubi1HbcO8cg==?= <rbg@talis.com> writes:
> Sandro Hawke wrote:
> > ...
> >
> > It often seems to me that bnodes should have been left out of RDF.
> > They're useful, but also painful.   (Or maybe, all things considered,
> > RDF should have been left out of RDF.  :-)
> >  =20
> If we want to keep RDF but drop something, there's only named nodes we
> can drop without loosing expressiveness. We would have to introduce an
> inverse functional property pointing to a URI-Literal. This would also
> make the owl:sameAs statement obsolete (it would just be one b-nodes
> with multiple name-properties). Another advantage would be that there
> are less arbitrary choices to take when describing a world with things
> with multiple names, currently you can arbitrarily decide which
> properties to associate with which of the same resources (the concept of
> having multiple things being the same, seems counter-intuitive to me
> anyway).

Yes, indeed.    I proposed this about six years ago [1], but never got
any traction with it.

RDF has a clunky design.  Most elegent, I think, would be bnodes +
strings + a uname property.  Another approach would be to get rid of
bnodes (and, yes, sacrifice some functionality.)  But, anyway, this is
water under the bridge.  (Still, it's useful to talk about sometimes, to
at least acknowledge the issues, even if we can't fix them.  Otherwise
people have a hard time learning it.)

      -- Sandro

[1] http://www.w3.org/2001/12/uname/

Received on Tuesday, 2 October 2007 17:09:16 UTC