Lists Home |
Date Index |
- From: "W. E. Perry" <email@example.com>
- To: XML DEV <firstname.lastname@example.org>
- Date: Fri, 29 Dec 2000 16:05:26 -0500
Uche Ogbuji wrote:
> As I see it, the SW is all about *not* requiring fixity.
I am describing a very particular fixity, viz. agreement on the semantics of
> As assumptions change, assertions can be added to the graph to reflect this
Added, yes. This is what I mean by building upon the 'fixity' (i.e., the
shared or agreed semantics) of assertions already in place. This (rather
static concept of a) Semantic Web does not anticipate processes which will
elaborate new semantics from such pre-existing references within the immediate
local environment in which that reference data might participate on a
particular occasion. OTOH, a web of services or processes is defined by the
sum of the particular, perhaps unique, functionality available at its
constituent nodes. One by one, those nodes might individually elaborate--for
their own purposes, within their unique contexts--very different semantics
from the same referenced data. In my own real world experience, this is what
specialized expertise consists of. Conversely, I can think of no vertical
market, nor for that matter of any horizontal enterprise comprising individual
specialties, where the different users of variously overlapping bits of
referenced data agree on a fixed and limited canonical semantics as a
precondition to processing it.
> Inferencing tools can help automate navigating these changes.
Perhaps. My concern is that they appear to do so by first fixing the
semantics, and thereby the boundaries of that navigation.
> Constraints are always presented in context, which can itself be changed.
Does this mean that the context will (must?) change to accommodate inflexible
constraints imposed by the agreed semantics of the references upon which it
builds? If so, that is precisely what I argue is the cart-before-the-horse of
pre-agreed Semantic Web semantics. Surely the unique context of a instance
process is what should determine the specific outcome (i.e., the semantics
elaborated from) that process.