Re: &

["Sean B. Palmer" <sean@mysterylights.com>]

Howdy,

> Maybe Sean could give us a clue or two, about how DAML+OIL
> extensions could help us at this point. I would also like to refine the
> namespaces framework, as it will be used by machines to understand
> what a document is about (via RDFS).

DAML+OIL is simply an ontology and inference language... with the term
"language" being used rather loosely there; it's more of a set of classes
and properties bundled together into a document. DAML+OIL is good, because
the people working on it (sponsered by DARPA, lead by Jim Hendler, worked
on by P. F. Patel-Schneider, L. Andrea Stein, Pat Hayes et al.) are
brilliant researchers.

HumanML can use DAML (and whatever WebONT produce) as a language for
defining taxonomies, data models, languages, or whatever else you want to
call them. DAML isn't quite as powerful as a FOPL, but it is good enough
for appplications like EARL (which indeed only uses a fraction of what DAML
has to offer), so it will be good enough for HumanML.

But DAML alone isn't the only specification that we can utilise to our
advantage. DAML itself is built on top of RDF Schema, and RDF Model and
Syntax. RDF Schema provides simple terms for classing and prototyping
languages... it's the most basic level of schema available on the Semantic
Web.

Some of the key parts of RDF are in the "wrong places". This isn't
something to be concerned about; it simply means that some terms were
invented long before they are really needed, and appear in the wrong
specifications. Somewhat akin to inventing a pen before there are any
words: harmless, but something to be aware of. In fact, this is probably
not something that we even need to consider, but I mention it in passing,
and in education.

We should also be prepared to have to invent our own ontology, proof, and
trust languages. The cool thing about the Semantic Web is that there is no
one central authority or set of specifications that really define how you
must use it. Obviously, we have to agree on soem very simple things, like
using RDF, but beyond that, you can do pretty much whatever you want. It
gives one the power to say anything about anything. Of course, the offset
is that you have to be careful how you develop things. There are some neat
Semantic Web hints and tips that can only really be learned from doing, and
that often involves messing it up once or twice. I'm currently in the
process of writing down some of these hints and tips.

Another thing to note is that RDF doesn't have any initial concept of what
a namespace is, and it doesn't require it. It uses namespaces in the syntax
to form URIs in the models, but neither the QNames or the namespaces are
preserved in the model. That isn't to say that they can't be represented in
the model, just that on converting the serialization to the model, the
namespaces vanish.

Well, it's nice to see that someone is getting into the Semantic Web :-)
The cool thing is, you (Manos) appear to be getting into the right-stuff,
and cutting through all the hype/junk. That's quite neat.

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .