Re: Upcoming wave of quad/namedgraph implementations ,was: Reification - whats best practice?

At 12:52 10/09/04 +0100, Dave Beckett wrote:
>One issue I see is that Named Graphs as currently defined by only
>allow URIs whereas several of the context and provenance proposals
>use or allow blank nodes - several have been given in earlier
>messages.
>
>In our work for the W3C RDF Data Access Working Group (DAWG), it
>would be good to know whether URIs alone are sufficient, or if as we
>suspect, many people want graphs named with blank nodes such as for
>triples from inference.  I thought there were problems with the
>scoping of blank-node named graphs, somewhere in one of the named
>graph papers.

FWIW, my implementations both would allow sub-formulae to be associated 
with either bnodes or URIs.  Generally, I'd expect to be able to use a 
bnode wherever a URI can be used (and such would be valid entailment 
according to RDF semantics).

Of course, if you want to make reference to such things from *outside* a 
graph within which it occurs, it seems reasonable to insist on having a 
URI.  (Though in some of my work, I have found that I want to be able to 
create nested (graph-)scopes for bnodes.

#g


------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact

Received on Sunday, 12 September 2004 14:39:43 UTC