RE: XHTML and RDF; thinking about Trix, etc

At 09:30 27/02/04 +0200, Patrick.Stickler@nokia.com wrote:

>I share your concerns about confusing things by introducing yet
>another serialization for RDF. I myself don't consider TriX as a
>replacement for RDF/XML (at least not anytime soon) but rather
>as a tool for specific purposes, specifically use in an XML environment.

Good... that seems like the direction I was drawn to.

>There are two key shortcomings to RDF/XML that I see insofar as
>its usability with generic XML tools is concerned: (a) consistency
>of structure, and (b) open tagset vocabulary.

And, maybe, syntactic complexity... the simplicty of the abstract syntax is 
somewhat obscured by the concrete XML syntax.  There may be good reasons 
for this, but that does incir some difficulties.  This is maybe just a 
facet of your (a).

>The former could be addressed by defining a subset of RDF/XML which
>provides a highly consistent, canonical serialization which is still fully
>valid RDF/XML.

I used to favour that, but now I'm not so sure.

>The latter issue, the open tagset vocabulary, requires something other
>than RDF/XML, and it's IMO this latter issue that causes the greatest
>barrier to the use of generic XML tools and methods.

Ack.  (I wonder if there's a way this --the point about open tagsets-- can 
be made more obvious in the introductory sections of Trix?  I think it's an 
important point to articulate why Trix is fundamentally different from 
RDF/XML.  I'm sure it's in there, but I didn't notice it when I glanced 
through the Trix document.)

>For my own part, if RDF/XML can be successfully embedded in XHTML
>without too many challenges, then that is surely preferable (I haven't
>had time yet to read Mark's paper, but surely will).
>
>The greatest value of TriX, IMO (and Jeremy may think differently) is
>as a specialized serialization which can be used by the XML community as
>an XML friendly expression of RDF, not as a replacement for RDF/XML
>for the broader RDF community.

Mark's paper doesn't use the RDF/XML syntax, but rather adapts elements of 
(loosely) XHTML to convey RDF abstract syntax.  By comparison, you've taken 
a more generic approach to XML-friendly RDF.  So my suggestion was that 
maybe elements of these approaches can be combined to offer a more generic 
approach to RDF-in-XML, without risking the perception of creating yet 
another format for RDF documents.  In summary, I'd propose...

- Concerning Trix:  how easily can Trix be embedded into an arbitrary XML 
document?  Can it be adapted to be any more friendly to existing XML 
document formats?

- Concerning RDF-in-XHTML:  as presented, the ideas are quite 
XHTML-specific.  But are they necessarily so?  Can they (or some of them) 
be adapted to work with an arbitrary carrier format?

I haven't thought deeply about the specific technologies here;  these are 
just questions I'm posing, hopefully for consideration by those who have.

#g
--


>-----Original Message-----
>From:   www-rdf-interest-request@w3.org on behalf of ext Graham Klyne
>Sent:   Thu 2004-02-26 13:52
>To:     www-rdf-interest
>Cc:     Jos De_Roo
>Subject:        Re: XHTML and RDF; thinking about Trix, etc
>
>
>
>At 02:41 26/02/04 +0100, Jos De_Roo wrote:
>
> >it gives a very good feeling to read things like
> >http://www.w3.org/MarkUp/2004/02/xhtml-rdf.html
> >:)
>
>Hey, yes!  That's some nice work.  Thanks for the reference.
>
>I'd like to see a nice user interface for this kind of thing built into an
>HTML editor :-).  (It puts me in mind of a kind of "literate RDF"?)
>
>...
>
>Anyway, to broader topics.
>
>I've been mulling over the recent discussion of Trix [1] from Jeremy and
>Patrick.
>
>I have been thinking that it is a worthy piece of work, but have been
>struggling to find a coherent view about why, after several years, we want
>to start talking about /another/ XML syntax for RDF.  If the purpose of the
>existing RDF standard is to provide a single recommended way to exchange
>RDF between applications, then is yet another XML syntax for RDF not the
>last thing we want muddying the waters at this time?
>
>I've also been having some thoughts about how, now that RDF is a full
>Recommendation, I might be able to promote its adoption in real-world
>applications, which means articulating some benefits of using RDF.  My
>thoughts have been lead in part by a comment by Brian McBride at a meeting
>last year, roughly:  "the question we ask should not be 'how do we get (all
>this data) converted to RDF', but rather 'how do we bring the benefits of
>RDF processing to (all this data)'".  I surely misquote, but I hope the
>intent is not damaged.  Part if the value I see in Mark Birbeck's paper
>[2], cited by Jos, is the way it addresses this question (for XHTML).
>
>Reading Mark's paper [2] together with these other thoughts, lead me to a
>possible conclusion.  Maybe we don't really need another stand-alone
>RDF/XML format, but something we can use is a way to incrementally embed
>RDF in existing XML documents in a way that is amenable to processing with
>existing XML tools, and in particular easy isolation of the RDF into some
>representation of its abstract syntax.
>
>I haven't looked or thought deeply enough about the technical issues, but
>I'm wondering if some combination of ideas from Trix [1] and "XHTML and
>RDF" [2] might not provide a framework for such?
>
>#g
>--
>
>[1] http://www.hpl.hp.com/techreports/2003/HPL-2003-268.html
>      http://www-uk.hpl.hp.com/people/jjc/tmp/trix.pdf
>
>[2] http://www.w3.org/MarkUp/2004/02/xhtml-rdf.html
>
>
>
>------------
>Graham Klyne
>For email:
>http://www.ninebynine.org/#Contact

------------
Graham Klyne
For email:
http://www.ninebynine.org/#Contact

Received on Friday, 27 February 2004 06:58:16 UTC