[XML4Lib] RDF or XML

[Alexander Johannesen]

On 5/15/07, Peter Kiraly <pkiraly at tesuji.eu> wrote:
> It depends on your usage plans. RDF is a special kind of XML
> designed for resource description.

That's a bit inaccurate; RDF in itself is a data-model with a
mini-ontology built in (meaning; a given "meaning" for a given set of
constructs) and has nothing to do with XML per se. However, in a lot
of plain talk, people refer to RDF when they really mean RDF/XML
(which is similar to when people say XSL, they really mean XSLT), and
indeed the W3C themselves use XML as the thruster-format for their
SemWeb initiative. RDF/XML is an XML representation of RDF statements.
RDF can also be represented in various other formats, such as N3 and
others.

Basically what happens is a representation of any piece of knowledge
(and I use that term lightly here :) as a triplet; [the_resource]
[the_connection] [the_thing_it_connects_to], so for example, in theory
;

  [Alexander] [works_for] [NLA]

and because RDF is contextual to the web (for the most part), same example ;

  [http://shelter.nu/me.html] [#works_for] [http://www.nla.gov.au/]

Of course, as you can see by that statememnt, there's tons of
ambiguity in it, such as when I say [http://shelter.nu/me.html], what
resource is that *really*? Me, the human, or my web page (as a web
page, and not a human representation), or the concept of my whole
website as a representation of me even if it's a single point of
entry? The same goes for the receptive end. And I've not really
defined a proper ontology here, so perhaps [#works_for] in reality
would be more like
[http://www.someplace.org/ont/work-relationships#works_for].

The idea here is that any SemWeb parser "understands" what all these
things mean (to a certain extent), especially the ontology bit (which,
if you're interested in these things, is where OWL and RDFS comes into
play, and will really screw with your mind) where the parser would
know how to handle resources with a [#works_for] relationship. Not
only do they have to handle that single item, but whole heaps of
similar statememnts made about both me and the NLA and all sorts of
things around it, both true and false. It can quickly get really big
and hairy, but it works ok for simple "Hello world" applications. :)

RDF unfortunately deals with a few issues about identity quite badly,
and introduces rather confusing things such as anonymous nodes,
strictly directional semantics, and necessary duplication because of
it, making some RDF representation rather crazy for people to
understand. The argument here is of course that tools are supposed to
understand this, not humans. Boy, I don't know how many technically
superiour technologies who have buckled under this very argument.

> [...] it is behind RSS news feeds etc.

No, that's not entirely correct; RSS version 2.0 is an RDF/XML
representaion, but it's far from the most popular RSS version. I think
0.94 and 1.0 are by far more popular, because, frankly, they're easier
to grasp.

As an example of the complexity involved in RDF and RDF/XML, when the
Atom community came together to create a better XML version of
syndication (at which point RSS 0.94, 1.0 and 2.0 all were used
haphazardly), they did specifically *not* recreate an RDF/XML thing.

> The Semantic Web
> initiative is greatly based upon this format because of its
> mathematical backround (description logic). The RDF has a
> 'meta language', which is the RDFS (RDF Schema language), with
> which you can define classes, properties and instances.
> The OWL language is another extension of RDF, which is used for
> ontology building.

Another way of saying this is that there's a basic simple datamodel at
the bottom, and you add as many layers of ontology as you feel you
need. RDF has a mini-ontology thrown in, RDFS adds a more data-centric
ontology layer, and then OWL itself has three levels of ontologies on
top of that. Make of that what you like.

> There are a lots of literature about the RDF, RDFS, OWL and the
> concerning mathematical-philosophical concepts, and there are
> some softwares as well (not too many i guess).

A handy book to understand all of this is "The Explorer's Guide to the
Semantic Web" (http://www.manning.com/passin/) [with the disclaimer
that I was the technical editor].


Alexander
-- 
 ---------------------------------------------------------------------------
 Project Wrangler, SOA, Information Alchymist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------