 | |
Network
configuration
with RDF | | |
 AI? Not necessarily
Isn't this what AI has been attempting for decades?
Yes and No.
Yes, in that RDF draws heavily on ideas of knowledge representation and formal semantics that have long
been used by the AI
community.
No, in that the useful applications of RDF seen today are much more prosaic. The base RDF specification
lacks many of the expressive
features that would be considered essential in an AI knowledge representation language. Consequently,
it is much simpler. The network
access application is just one example of an RDF application that does not depend on exotic reasoning
capabilityes. I see RDF as a
small step with some real near-term benefits (and considerable long-term potential).
The other thing that sets RDF apart from traditional approaches to Knowledge Representation is its use
of URIs as identifiers. The
meaning of these identifiers can be grounded in the Web. Within the Web, URIs are presumed to
be globally unique, so each occurrence
of a given URI can be presumed to denote the same thing across lexical contexts.
| 
