RE: KR & W3C (was KR & Issue/bug tracking terms in RDFS?)

I wouldn't go as far as Joshua in his criticism of your posts, but I do
think the suggestion that you set up a list for your language (Yahoo groups
is the easiest) is a very good one. That way you have a place to offer your
solutions that are YKR-focussed (Your KR ;-), and rdf-interest stays clear
for purely RDF-related solutions, and you won't be subject to allegations of
spamming.

>The reasons I'm so certain my language is better:
>(1) I've looked at other AI languages for 30 years;
>(2) in the last 6 years I haven't found any "problem" my language couldn't
>handle;
>(3) the core of my language is context-dependent definitions & hierarchies.

Right, the way you are determining "better" here falls down somewhat because
you're not really comparing like with like. On your specific points :

1) Though RDF draws on previous AI material (many of the people working on
it have first-hand knowledge of the past 30 year's work), it is almost
certainly a big mistake to view it as an AI language - it is primarily a web
language, and is firmly grounded in existing web technologies. As Eric
Miller (W3C's Semantic Web Activity Lead) recently said, "it's not
artificial and it's not intelligent".  (can't find the ref, sorry)

2) This doesn't really count for much - it's possible to say the same about
e.g. comma separated values text. In terms of the ease of solving
web-related problems (RDF's main domain), then by design RDF has a head
start over most general purpose problem-solving languages.

3) I think it's a bit early to decide with any certainty what approaches
to/aspects of modelling are going to be most beneficial. Personally I would
expect that in the web environment some degree of context-independence would
be desirable, and that hierarchical modelling leaves a lot to be desired -
the graph model fits better than the tree.

Merely saying YKR is "better" isn't going to win anyone over. There is
however a simple way that you could state your case convincingly. Take some
of the tasks for which it has been shown that RDF is a good solution, and
provide online demonstrations of YKR doing the jobs better. For example, the
TAP project [1] uses an RDF-based knowledge base in concert with the most
commonly deployed web server (Apache). If YKR really is better, then it
should be little trouble for you to integrate your system with Apache and
set up an augmented Google search similar to the TAP 'Activity Based Search'
demo.

Discussion about such a project should of course take place on the
appropriate mailing list, but I don't think anyone on this list would object
to your announcing such a demonstration here.

Cheers,
Danny.

[1] http://tap.stanford.edu/tap/demos.html

Received on Saturday, 21 December 2002 07:28:11 UTC