RE: Taking an axe RDF in XML? (no thank you)

Hash: SHA1

> -----Original Message-----
> From: 
> [] On Behalf Of Assini, 
> Pasqualino
> Sent: 23 May 2002 15:44
> To:
> Subject: RE: Taking an axe RDF in XML? (no thank you)
> Bill,
> I am not sure that the XML syntax is a major stumbling block 
> on the road to the adoption of RDF.
> The counterexample is XSLT that has an even more horrible 
> syntax but it has been widely picked up.
> My explanation of XSLT success vs RDF failure to gain 
> widespread acceptance is that the use case that XSLT is made 
> to provide, tranforming XML, is both widely needed and easily 
> understood.

XSLT and RDF are orthogonal. XSLT is a programming language. XSLT
also addressed a direct need (known from SGML work via DSSSL in any
case). I don't consider XSLT a counterexample.

If you identify a pressing need, you can get away with less than
beautiful work; often you need to just get it out the door or the
market will see you as the bottleneck and route around you (try
reflecting on the TAG and WS-IO in this light). DOM, XML Schema and
XSLT are examples of that in different ways. If you don’t identify
or catch the coattails of an urgency, you have an uphill battle.
Typically, you need something cool or disruptive or self evidently
valuable, and failing those, something way simple, to witness
adoption. RDF without tools is neither cool nor disruptive not
obviously valuable; the XML is not way simple (it is getting

> XSLT provide 'instant satisfaction': you write a transform 
> and some funny-looking XML gets transformed in human-readable 
> HTML. Now that's useful.

In some circles, that's known as the paradox of the active user :)

> On the contrary: you describe some resource in RDF and ... 
> not much happens.

RDF is data and as such is inert; nothing much happens without
running some code over it. That separation is very likely a good

> Are you aware of any XML syntax that people actually like to use?

RNG, Docbook, RSS, Jabber, XML-RPC, Ant, I find usable enough. It
is possible to produce a cleaner XML serialization for RDF, tho'
questioning the value of XML in itself is somewhat distracting (I'm
not trying to squabble, just pointing out that one can write
relatively clean XML). 

> XML fulfils a very important function in the semantic web as 
> a flexible, easy-to parse, transport syntax but it is far 
> from being easy on the eye. 

I'd very happy if the world had locked on s-expression notation,
but it didn't. In the meantime my interest is in seeing RDF used by
a lot of people. I suspect that happens easiest via piggybacking on
XML and web services infrastructure as that's a locus of interest.
Strategically there's no question that WS needs RDF or something
like it, both for describing services and eventually message

That approach might be considered pandering, but frankly my
understanding of RDF and the semantic web is too make this stuff
accessible at a web scale. I would say it's about maximising the
web's potential. In the meantime there exist barriers to adoption
of RDF. The Model Theory will doubtless do a fine job for removing
barriers where advanced semweb work and research is taking place.
My barrier of choice happens to be the syntax as I feel that's the
key to getting RDF out of the labs and making it consumer grade

> Problem: I don't think that you should be looking at XML for 
> a solution of this problem.

I'm glad you see a problem. As I see it WS middleware represent the
rich pickings if not quite low hanging fruit for RDF, notably the
WSFL, WSDL and XLANG stable of languages; that means an XML
baseline, not a notation that you or I might find pleasing. Make
RDF valuable and simple for the WS community/industry and you'll
see tools.

It's worth noting I don't really have any technical criticisms of
RDF. The only ones that ever stuck in my craw were reification and
literals and to some degree graph merging. I'm confident these are
getting ironed out and that RDF is good enough shape. Graph merging
is pretty much done at the MT level now.

> A much better starting point would be N3 or other non-XML
> syntaxes.  What about an official non-XML syntax for RDF
> optimised for 
> human readability?

I respectfully disagree for the reasons mentioned, though I do take
your point about XML being difficult to read. Ntriples is probably
what you're looking for.

Bill de hÓra

Version: PGP 7.0.4


Received on Thursday, 23 May 2002 12:56:10 UTC