Hello,
Mark Birbeck wrote:
Hi all,
As Shane says, I think we've lost sight of the issue here. :)
I know, sorry I get distracted :)
On Fri, Jul 31, 2009 at 2:11 AM, Martin McEvoy<mar...@weborganics.co.uk> wrote:
[...]
If it were possible, it would allow a microformats like markup, which in my
view would be a whole lot more elegant and readable than the crufty,
slightly over the top approach RDFa has now using xmlns...
[...]
I think the way to look at it is much like the infoset in XML. An
underlying canonical form is defined, and once that is stable, you can
layer on top of it all sorts of semantic sugar to make things easier.
The RDFa TF has essentially been dealing with the underlying layer,
but we've always known that easier mark-up is a goal. But since we
also know that 'Microformats don't scale',
microformats *do* scale but only in a limited way by re-using existing
properties and design patterns in microformats typically used in
hcard... the only issue is that root class names in new uf's cant be
determined by old parsers which leads to scaling issues but is pretty
easy to fix in my view.
we can hardly expect to get
the same simplicity without first establishing this layer.
agreed....
This kind of simplicity -- built on a solid architecture -- is exactly
what was proposed in detail here:
<http://webbackplane.com/mark-birbeck/blog/2009/04/30/tokenising-the-semantic-web>
Note by the way that the attribute name being put forward in this
proposal is @token, because I believe we need to move away from *both*
the idea of a prefix, *and* the idea of 'I would like to use this
vocabulary', and instead look at a more generic solution (bear in mind
that CURIEs are also used in @role, for example, which is nothing to
do with RDFa).
So, my point is that neither the attribute name, nor the fact that we
would like a microformats-style syntax is actually the subject of this
discussion; the second is a given, and the first is pretty easy, when
we're ready.
The key thing is Ben's proposal at the top of this thread, which
suggests that the mapping between items should not occur at this
'token' level, but at the level of RDF itself, via OWL.
I tried a similar approach last year, which was workable but never
managed to finish the project. It used XMDP+RDFa Profiles pretty much in
the same way as Ben described see: http://purl.org/net/haudio, The RDFa
was used to describe classes and rel values and their mappings to RDF.
which meant that If I used mark-up like this:
<head profile="http://purl.org/net/haudio";>
....
</head>
<div id="track" class="haudio">
<span class="fn">Start wearing purple</span>
</div>
the RDF output was....
<?xml version="1.0" encoding="UTF-8"?>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#";>
<rdf:Description rdf:about="http://somedomain.com#track";>
<rdf:type rdf:resource="http://purl.org/net/haudio"/>
</rdf:Description>
<rdf:Description rdf:about="http://somedomain.com#track";>
<haudio:fn xmlns:haudio="http://purl.org/net/haudio";>Start wearing
purple</haudio:fn>
</rdf:Description>
</rdf:RDF>
That's what people need to be discussing here, since it's a completely
different way of approaching the issue to the one outlined in my blog
post.
Choosing a name for an attribute is not going to be an issue; but
unravelling the level of abstraction at which the mapping of terms
should take place, will be.
Regards,
Mark
Best wishes
--
Martin McEvoy
http://weborganics.co.uk/
