[
Lists Home |
Date Index |
Thread Index
]
> That is why Joshua's example of USENET is
> interesting. A global ontology. One assumes
> the users have to police it themselves.
There are two pieces to it:
* Global and public distribution/aggregation network: When you publish
something on USENET, it is immediately available to anyone who
subscribes to USENET. The medium is public -- anyone can publish into
USENET. Web pages, by contrast, require you to publish to a
privately-owned site, and require the reader to deliberately connect to
your private site to consume what you published. In such a system, the
metadata is only available to the person who can "crawl" for it. Push
vs. Pull. Pull is ideal for the web, but push is ideal for semantic
web.
* Subject-Sensitive Filtering: In a "push" model, the amount of metadata
being pushed around globally will be overwhelming. People need the
ability to filter out stuff they don't care about. The newsgroup
hierarchy could be seen as an ontology, but I think of it as a very weak
one. More like a set of "channels" that people can tune into or
broadcast on. It's possible to post a message about Oracle in the
Microsoft newsgroup, and there is a chance that someone will be able to
answer your question. But the channels exist so that
people/applications can voluntarily provide some filtering information.
I'm not saying that such "Channels" are *necessary* to a semantic web,
since ideally you could establish filters automatically based on any
metadata characteristic. But USENET is an example of where someone
already solved the problem (albeit in a limited and low-tech way)
Architecturally, I think decentralized "push" and subject-oriented
"filtering" are essential for the semantic web. And USENET (or IRC for
low-latency) is the definitive subject-oriented push communication
system.
|
