On 9 Jul 2006, at 14:37, Bill de hÓra wrote:
Henry Story wrote:
On 9 Jul 2006, at 02:15, Bill de hÓra wrote:
RDF on the web has failed.
Not at all. The biggest xml data experiment on the web that chose
xml is failing: SOAP. [0]
SOAP is not a data model. It's not the biggest XML on the web
experiment. XML is not comparable to RDF. Pointing at SOAP and
shouting "fire" doesn't fool anyone.
It is the one most money has been poured into recently. And it is
using xml for something that is more than markup as understood by
xhtml and open office documents, which I will be the first to say are
GREAT successes. Kudos to Tim Bray and others in that room, plus the
SGML folk for this. Seriously.
But I don't want to be parsing through these WS-* docs. It just looks
insane.
The biggest rdf data experiment on the web is RSS1.0. Enough said.
Come on. That was one of the first attempts at working with rdf/xml.
I don't think it is such a problem. Are you sure atom is that much
better? I have not done the comparison yet.
There is a lot of usage of rdf in the life sciences. And that may
have a lot more impact than syndication.
I don't need to mistake mime types for data formats. Atom does
hardly anything and yet it has to invent a new mime type!
So do all the semweb technologies, and they still probably don't
work properly as a web interlingua. How should I interpret an OWL
document served with an RDF media type? Which theory will I apply?
I am not sure I understand your problem. Are you wondering if you
should apply OWL or RDF? That's not a problem OWL documents can and
usually will be served with mime type "application/rdf+xml". OWL is
just a semantic web vocabulary to define logically well understood
properties of relations and classes. It is RDF. There is no trouble
understanding the vocabulary it defines.
If you happen to really understand the OWL terms you get extra power
of course, as you will then be able to infer consequences of the
statements shipped. Just as a clever atom parser can infer things
from atom documents by the way. The difference is that OWL is backed
by mathematical logic. Atom is not. (But now with AtomOwl it is. :-)
But as Kuhn [2] and Feyerabend [3] have very well argued, these
types of debates are not won through argument. People work within
paradigms that forge their vision of the world. So that
counterexamples will always [...]
You should just say "you don't get it".
I am pointing out that if you don't get it, it is not necessarily a
detail that you are missing. It may require a little faith and good
will. A bit of a shift in perspective.
I am not the syndication geek you're looking for. I've worked with
RDF on and off for over half a decade, contributed to the standards
process. When I finish this email I'm going back to wrap up a
product to use SKOS in Plone before the world cup starts. If you
want a new paradigm, you'll have to do better than railing about
the current one.
Well I am thankful that people like you helped work out all the hard
bits first. I came later to the game. It helps to bring in new blood.
It is very tiring being out there first. And sometimes one gets
burned by the fire. I am coming to tell you that you did not waste
your time. A lot of things have been built since then upon those
foundations. (I got burned badly working with java when it was in
1.0. It took up to 1.3 for it to be ok, and 1.4 for it to be good.
Now it is excellent.)
* But of course, you're not /really/ using it.
BlogEd, which I completed in a simple form a year ago, is using it.
RDF isn't expressive enough for what you need to do. Look at your
code again and be assured you have gone well beyond RDF semantics.
You are right. I have gone somewhat beyond OWL in my N3 ontology:
http://bblfish.net/work/atom-owl/2006-06-06/AtomOwl.n3
I have used rules, which are allowed by N3. Rules are relations
between graphs. And graphs are now part of RDF since SPARQL. So rules
are just a logical consequence of that.
I have used rules to make some statements more precise, and to play
with those possibilities. They are not strictly needed. None of the
other serializations of AtomOwl use them (they can't since they don't
have a syntax for them). So a parser of AtomOwl.n3 that understands
those will be able to understand a little bit more than the other
parsers, but it should not find anything contradictory. I keep the
open world assumption I think)
http://www.amazon.co.uk/gp/product/0198248547/
read the essay by Drew McDermott. I think you need a crisis of faith.
I read that when it came out in the early 1990s. I remember one very
good article that distinguishes two types of AI: hard AI and soft AI.
One is trying to get computers to think and act like humans,
understand intuition, etc. The other is trying to solve current
problems by looking at how humans do things. The first are like early
everest climbers, or the people who tried to go to the moon. They are
trying something exceedingly hard. But humanity can be thankful for
their energy and the horizons they open up. The others are more
pragmatic, and have had a lot of success. One can automate quite a
few things, that humans can do. Often not to the level of experts but
often to a good approximation. Take the babelfish.yahoo.com for
example, which is based on technology from Systran from the 1960ies!
The translation are not very good, (though it improves a lot with
specialized dictionaries and well written text), but it saved people
in the CIA the need to read through every boring document that came
out of Moscow during the cold war. And now the free online version
can be very helpful if used correctly, (as well as a lot of fun if
used incorrectly).
Anyway. The Semantic Web is not about AI first of all. It is about
clear communication. Using URIs as names means you can find the
meaning of your terms at the click of a button. Think permalinks for
concepts and objects. It falls in the very pragmatic section of AI if
you wish. It is based on years of research into logic.
I like to build on solid ground. Others can build on soap bubbles if
they wish.
Henry
cheers
Bill
