- From: Bullard, Claude L (Len) <clbullar@ingr.com>
- Date: Tue, 29 Jul 2003 09:11:56 -0500
- To: "'Graham Klyne'" <gk@ninebynine.org>, www-tag@w3.org, pat hayes <phayes@ihmc.us>
From: Graham Klyne [mailto:gk@ninebynine.org] >(1) An "information resource" is a one that returns a representation when >one does a GET on its URI. Which means as a distinction, it is observable and testable but transient. The results are only valid for the transaction (state returned). >(2) The exact nature of a resource is unknowable, in the sense that it's >not possible to fix a single denotation, but for an information resource >it is circumscribed by understanding the things that GET returns as >representations. Which again means it is transient. >This, I think, is the basis of shared understanding in >today's (pre semantic) Web. The more different representations there are, >the more precisely we might be able to pin down the underlying resource >(but never completely). Nor finally. The resource owner has a credibility issue with respect to reliably returning representations for which previous evaluations yielded the 'shared' understanding. As far as I can tell, the term 'information resource' doesn't buy the web architecture much except the distinction of negotiation of meaning, aka, ontological commitment. Fielding is right that clients are blind to the meaning, type or class of a resource per transaction. They can only establish a shared understanding that is as reliable as the resource is committed to maintaining it. So the distinction, 'information resource' is yet again, a social distinction. It provides a term for describing an operational contract, not a condition of the architecture per se. This is even more particularly true of the semantic web. In simple terms, trust but verify. len
Received on Tuesday, 29 July 2003 10:12:07 UTC