Lists Home |
Date Index |
On 2002-02-15 17:42, "ext Jonathan Borden" <firstname.lastname@example.org> wrote:
> Is RDDL a long term solution? I don't know. It will be useful as long as
> HTML browsers are the most popular types of user agents.
I have no problem with specific solutions for specific use cases.
It seems in general that we pretty much see things in a similar
light, though perhaps put different emphasis on what needs to be
addressed first, and how.
My concerns are about those specific use cases driving web architecture
in ways that will not be condusive to all reasonable use cases.
The browser+human context is certainly the most prevalent, and has
made the web what it is today -- but we need to be sure that we are
taking into account the future automated agent context when thinking
about official web (and thus semantic web) architecture.
See my recent comments on the TAG list about this:
> I am prefectly
> happy to accept that, and reevaluate the situation if the Web proves itself
> to radically change in the future.
> For the record I _am_ a proponent of the "Semantic Web" and not afraid to
> say so. I also firmly believe that "AI" become useful in the future, perhaps
> in the next decade if the available processing power continues to increase
> at the present rate.
I agree with you that real AI is not all pie in the sky, but...
I see there being several phases of development towards real AI on the web,
and the first phase is the consistent, explicit definition of knowledge in
a form that is easily interchangable and mergable. To do that, we have to
agree what kinds of things we are talking about, and how we are going to
talk about them. Thus these issues of what namespaces are, how we talk about
resources that utilize namespaces, what URIs actually denote, what are the
classes of URIs, can the same URI have multiple interpretations or denote
more than one thing, how we formally associate semantics with specific URIs,
or with all instances of a URI scheme, or with all instances of a URI class,
etc, etc, etc are the very meat and potatoes of that first phase.
The web won't "radically change". It will evolve. And that evolution has to
be actively happening now.
A RDDL kind of solution based on namespaces where namespace URIs are equated
to vocabularies, document models, etc. does not provide the needed accuracy
and resolution for us to move forward.
Though, a RDDL like solution, based on a more accurate resolution of
web resources could serve as a reasonable first step and provide useful
functionality now. C.f.
> So look what great things have been accomplished by the "lowly" HTTP
> protocol and HTML browser.
Sure. But it's apples and oranges. Web browsers are intended for
human consumption. I think it's fair to say that the web itself is
intended for human consumption.
The point of the semantic web is that it's intended for machine
consumption -- and if our web architecture is not going to provide
the resolution and formal definition needed by automated agents,
then how will we ever see the sw realized?
> I am more than happy to tie RDDL to that
> bandwagon, at least for the time being.
And I can fully appreciate such a view to a certain extent, but
I expect that you will agree that we cannot only look at the horizon
(the way things might be done in the future), nor can we only look at
our feet (how we're doing things now). If we are to get anywhere
interesting without falling into too many holes along the way, we have to
keep a view of both the future and the present. My recent proposal for
an RDDL+RDF approach is offered in that spirit.
Patrick Stickler Phone: +358 50 483 9453
Senior Research Scientist Fax: +358 7180 35409
Nokia Research Center Email: email@example.com