[
Lists Home |
Date Index |
Thread Index
]
That was my conclusion too, last year. It is why
Gruber coined the concept "ontological commitment"
and we refer to the occasional test of commitment.
Ontological communities are a fact of life and
in fact, one way to deduce that some person or
thing is a member of a reference group is to
check the vocabulary usage. No news there, but
of consequence to people who make secondary
inferences for behavioral systems. There are
governance issues which are now being dimly
understood as America rushes to provide
profiling systems with global reach.
In my article in Markup Languages (2-4), I concluded:
"In contrast to other descriptions of web systems
as services, a semantic web is not simply a web of
discovery of resources. With its dependence on
ontology as the means, a semantic web is a web of
agreement of means for choices. While its goal is
to enable agents to process descriptions and offer
choices, it can become a means to discard any message
considered by a dominant community to be irrelevant
or noisy. Agents filter opinion, winnow classifications,
and ultimately are a means of stratification of information
and sources of information. In essence, the concept
of the right to choose the means to choose is becoming
the dominant issue of the next generation of web evolution."
Note that web services (UDDI) also depend on ontologies.
The devil is in the details and depth, but ultimately
on the freedom of the described to commit to the description.
As long as John Cowan assents to the description on a web page
he controls, the system works just fine as a means of
harvesting values by which he can be identified.
len
-----Original Message-----
From: Joshua Allen [mailto:joshuaa@microsoft.com]
Balkanization is more of a danger in semantic web, in my opinion, since
there is a lot more room for individual metadata communities to form and
thrive before they end up having to interoperate. And it is harder for
a few large vendors to ever have significant enough control over all of
the metadata publishing and consumption pieces to be able to force some
level of sameness from the start. So I think semantic web will require
*more* vigilance and work than RPC to maintain openness.
|