I was just looking through the RDF Primer (what used to be the Model and Syntax document). It's pretty good, actually. It doesn't use the convoluted XML syntax - instead it uses N-Triples. Whoever said the W3C weren't evolving?!

Anyway, I had a thought arising out of yesterday's work on adding OpenIDs to FOAF documents. Wouldn't it be cool if RDF/XML had the 'class' attribute from HTML? It wouldn't contain any semantic value - but would simply be used to make manual parsing easier. If you are using a language framework which doesn't support the full set of SPARQL syntax (which is, to be frank, quite a lot of them), it'd be cool to be able to address chunks of the document through classes, like:

rdf.class['profile'][0].predicate

How would this be useful? Well, in combination with rdfs:seeAlso, you could use it to selectively pull in other graphs. It’d almost become like the ‘rel’ attribute - specifying the relationship of the current document to another document, so you can more narrowly specify rdfs:seeAlso parsing. Save bandwidth and all that.

Just an idea - probably no more than a passing fancy. If there was a way of boiling it down to 140 characters, I wouldn’t have even bothered blogging it… 
