Re: better support for bnodes

Mark Birbeck wrote:
> 
> Hi Ivan,
> 
>> > We've now got a bnode that isn't attached to anything, with a
>> > predicate of bibtex:name.
>>
>> Hm. Is there? My understanding (but that may be wrong!) is that in your
>> outlined solution above the blank node is created because no object has
>> been assigned to bibtex:journal. If there is an object to it, then there
>> is no reason to define a blank node; the meta would then be attached to
>> the _:DJDKWBDADIH05 under the rules of finding a subject. Isn't that
>> what would happen? (It is not necessarily much nicer, though, but those
>> are the rules...)
> 
> 
> No, the bnode is still created, by the rules that apply when 'meta'
> and 'link' are used as children of an element that isn't another
> 'meta'. Of course, if the latter is the case then reification kicks
> in, but without it, we're just creating properties on a bnode.
> 

I stand corrected:-)

> So the rule I'm proposing to add is simply that if there is no object,
> the object should be the bnode identifier for the current
> element...which rather conveniently makes it match up with the bnode
> identifier created when we discover predicates defined by 'meta' and
> 'link' child elements. :) So we've achieved the 'chaining' that we
> used to have in early drafts, and that you get in RDF/XML as a result
> of the striping.
> 
> However, if you were to later add an @href, you would get the
> 'orphaned' bnode that I referred to, since although the bnode is still
> there, nothing is referring to it. But as I said, you could argue that
> this is 'correct'--you don't sound like you mind that consequence too
> much, Ivan. Is that fair to say?


Yep!

Cheers

Ivan

> 
> Regards,
> 
> Mark
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
URL: http://www.w3.org/People/Ivan/
PGP Key: http://www.cwi.nl/%7Eivan/AboutMe/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf