RE: Why is RDF such a tough sell?

On Tue, 25 Jun 2002, Narahari, Sateesh wrote:

>> However, they do need to be able to deal with data in a unified fashion,
>> and perhaps understand the data just as far as it takes to filter it. RDF
>> facilitates that, XSchema doesn't.
>
>Why not?. Elaborate on it please...can you give some concrete examples?

As for the latter, not really. RDF hasn't spread far enough, yet. As for
the former, presume you have one XML schema which expresses references as
URI's inside element content, another which uses XLink, a third which uses
HTML-type namespace-free href's, and a fourth utilizing some variation of
HyTime. Then suppose your application just wants to gather references,
pure and simple. The four syntaxes will need four different (and, in the
last case, surprisingly complex) XSLT stylesheets to do the job. Now
consider mixing those together arbitrarily via namespaces. Then increase
the number of schemata to twenty or so. What you've got is a hideous mess.

In the corresponding RDF framework what you have is a bunch of statements.
Any reference is bound to appear as a subject/object, somewhere. You would
write a single N3 filter for each new namespace and merge those. Then
you'd cwm %1 -filter merged-filter.n3>result.n3 anything coming in. And
you're done. Seems somewhat simpler to me, even while my main expertise
lies squarely in XSLT.

>What about a simple XSLT filter which filters out all documents which do
>not belong to a given namespace?.

Who said anything about monolithic documents? What I meant was that those
1000 schemata will have been combined into a single document via
namespaces, arbitrarily. (XSchema does not support such a thing, of
course. But that is what your syndication framework will want to handle,
since its providers will want to extend any schema for their own needs.)
Do you really suppose anyone would want to code such a stylesheet, let
alone actually run it?

Of course in the end, the question isn't so much about what is possible.
It's about what is easy, clean and elegant. After all, anything that can
be expressed in RDF can be in XML, and vice versa.  (RDF can be serialized
as XML; any XML document can be described in an RDF vocabulary which
faithfully encodes all relationships expressed in an XML Infoset.) Before
we start referring to statements and doing higher order logic, no real
trouble emerges beyond programmatic complexity -- it's just as you say.

I'd say the only time RDF becomes truly essential is when you need to have
clearly defined semantics for a reference back to the logical relations
you're only just defining. There arbitrary XML simply becomes too
unwieldy, while RDF still behaves just as it did in the beginning.

Sampo Syreeni, aka decoy - mailto:decoy@iki.fi, tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2

Received on Tuesday, 25 June 2002 19:53:17 UTC