Lists Home |
Date Index |
- To: XML Developers List <email@example.com>
- Subject: RE: [xml-dev] Meta-somethingorother (was the semantic web mega-permathread thing)
- From: "Bullard, Claude L (Len)" <firstname.lastname@example.org>
- Date: Mon, 14 Jun 2004 14:26:37 -0500
Then it comes down to what one can believe about the power
of market forces to achieve optimum results (it's nonlinear
so no guarantees).
Which is why there will always be some suspicion of the results
of what the agents tell us. The semantic web will be just as
'superstitious' as the humans, but possibly able to detect that
better or at least, more willingly (needs vetting systems). If
the agent is constrained as to which other agents it can play
with, it has the same bias problems as human supply chains.
By the way, it has nothing at all to do with the ontology
being *proprietary* and everything to do with how many and who
agree that it is *right* given some question, e.g,;
a) Was Reagan a great President?
b) Who invented XML?
c) Was the Trojan Horse real?
d) What is the fate of Harry Potter?
What will it do with those? Of course, the ontology is just
the beginning for getting answers to these. Agents are more
than happy faces on ontologies. Applications to answer those
questions go beyond the power of ontologies. Yet I can:
a) There is no right answer. Only opinion.
b) There is an accurate answer and there is the market answer.
c) It hasn't been proven. Insufficient evidence.
d) It isn't known.
It will be interesting to see how aggregation of semantic metadata
by harvesting WinFS metadatabases works out. I guess it returns
something like what Google returns now.
Then it will be interesting to enable agents to simulate
situations and see if we get the same results we do for
human games for situational analysis. We've discussed
this here before: situational games are way better
predictors than statistics.
From: Mark Baker [mailto:email@example.com]
d) you want to develop a completely proprietary ontology and provide no
The Semantic Web - vision or technologies - doesn't depend on these
mappings existing since it has to (and does) support case d). Of
course, the more instances of case d) out there, the fewer cases of
open, ad-hoc integration we'll see, but that has little to do with the
technology, and everything to do with market forces.