CURIEs and alternative sources of prefix/string mappings [was Re: RDFa reliance on namespace declaration]

Hi Ian,

> Actually this thread doesn't concern eRDF at all. What I'm interested in
> is why RDFa uses qnames, which I've shown to be fragile...

I think that's stretching things, to put it bluntly! Each of the
'problems' you have raised I've shown to actually not be problems. I
agree that QNames in XML is quirky, and I don't think you'll find many
people to disagree with that. But it's an enormous leap to suggest not
using something as widespread and standard as QNames, and any
alternative would have to be extremely well worked through and
consistent, and not something that just evolves.


>... when the
> following would work just as well and be robust when used with all
> current XML tools:
>
> <span property="http://purl.org/dc/elements/1.1/date"
>        type="http://www.w3.org/2001/XMLSchema#date"
>        content="2006-01-02">

I agree with you here Ian, I really do. Indeed, Steven Pemberton had
to persuade me away from exactly this approach many moons ago in our
early work on RDFa!

But one of our 'communities' (as Karl rightly pointed out) is HTML
authors, and we really can't expect them to do this. I know the
approach you describe has been successfully used in Atom, but those
documents are invariably generated rather than being hand authored.

It's no accident that RDFa aims to be friendly to the community of
HTML authors. Getting access to the metadata that this 'community'
puts into their documents has been a primary motivation for RDFa since
the beginning. Here's a quote from the introduction to one of the
early drafts, February 2004 (it was called RDF/XHTML back then):

  We have two standards running parallel with each other; HTML is the
  de facto standard for document markup, accounting for millions of
  items on the web. RDF is a standard for expressing metadata, which
  in turn provides a foundation for making use of that metadata, such
  as reasoning about it. Yet the former is very rarely the subject of the
  latter; meta information placed in the HTML family of documents is
  often encoded in such a way as to make it difficult to extract by
  RDF-related parsers. And if it cannot be extracted, then it cannot be
  used.

  Our intention here is to make more of the information that is
  contained within HTML-family documents available to RDF tools, but
  without putting an unnecessary burden on authors familiar with HTML,
  but not with the subtleties of triples and statements. [1]

The final paragraph is probably key to the whole concept--it was being
suggested that we make the metadata in documents availabled to RDF
tools "but without putting an unnecessary burden on authors". This is
not just us being nice; it was a recognition that unless it is easy to
put metadata in documents, authors won't do it.

So, let's back up a little, and say that we would probably agree
*some* URI prefixing/abbreviation mechanism is desirable, but we lets
say that we don't quite know what it might be yet. I've proposed
CURIEs as a QName-like syntax, that uses XML namespaces as its source
of subsitution values. On a number of occassions myself and others
have discussed making this proposal open enough to cope with other
substitution mechanisms: so for example SPARQL could use CURIEs but
stipulate that its 'PREFIX' syntax is used to provide the strings for
subsititution; Misha and the IPTC could provide an additional
attribute in NewsML to provide the substitution strings, perhaps via
some external file; and yes, we could even support some other
mechanism than XML namespaces in XHTML to provide the substitution
strings.

But although I've done quite a lot of work in this area, try to make
CURIEs open in a similar way to that of XPath (which defines a
'context'), that doesn't necessarily mean that I think that XHTML
should use a mechanism other than namespaces! ;)

What I *am* getting at, though, is that if we successfully made CURIEs
open enough that it could support other syntaxes and methods of
substitution (in the way Misha would like), then if we do decide to
make 'the big change' in RDFa (i.e., to not use XML namespaces), then
it would just be a case of flicking a switch and using a different
'context' in CURIEs.

To put it a different way, we should try to crack the problem 'in the
abstract', since we know that regardless of our final decisions on
using QNames/CURIEs in XHTML, there are at least two groups that
cannot use namespaces as their source of 'string substitutions'--the
IPTC and those not using XML. If they are to be able to use CURIEs
then we've got to solve the problem anyway, and once we've solved the
problem we can separately make a decision as to whether to use the
same technique ourselves in RDFa/XHTML.

Regards,

Mark

[1] http://www.formsplayer.com/notes/xhtml-meta-data-03.html

-- 
Mark Birbeck
CEO
x-port.net Ltd.

e: Mark.Birbeck@x-port.net
t: +44 (0) 20 7689 9232
w: http://www.formsPlayer.com/
b: http://internet-apps.blogspot.com/

Download our XForms processor from
http://www.formsPlayer.com/

Received on Monday, 19 June 2006 12:07:54 UTC