Re: Some thoughts on the RDFa DOM API document

On Tue, 27 Apr 2010 09:42:32 -0400
Manu Sporny <msporny@digitalbazaar.com> wrote:

> Good catch, Benjamin... I wonder why Blank Nodes aren't allowed as
> predicates?

It's a bit of an annoying restriction. A lot of N-ary relationship
problems could be solved by bnode predicates. e.g. 

	<#school>
		[
		  rdfs:subPropertyOf teaching:teacher ;
		  teaching:subject thesaurus:Maths ;
		  rdfs:label "maths teacher"@en
		]
			<#joe> .

N3 allows blank node predicates; the SPARQL query language does too
(though pretty much all SPARQL engines sit on top of RDF quad or triple
stores, so in practice they never get a chance to see any bnode
predicates). RDF doesn't though. Along with literal subjects, it's an
oft-mentioned wishlist item for a future version of RDF should there
ever be one.

TimBL noted that blank node predicates would be useful when he reviewed
the RDF Concepts & Abstract Data Model:
<http://www.w3.org/2002/07/29-rdfcadm-tbl.html#xtocid48022>.

So, while RDF doesn't allow bnode predicates, we should bear in mind
that a future version of RDF might. This means that while RDFa and the
DOM API shouldn't allow bnode predicates, we should try to avoid making
decisions that would make it difficult to add that feature in a future
version.

-- 
Toby A Inkster
<mailto:mail@tobyinkster.co.uk>
<http://tobyinkster.co.uk>

Received on Tuesday, 27 April 2010 15:48:56 UTC