- From: Graham Klyne <GK@NineByNine.org>
- Date: Tue, 14 Jan 2003 16:33:15 +0000
- To: RDF core WG <w3c-rdfcore-wg@w3.org>
I've (belatedly) been catching up and reading through rdf-comments, and
find myself wondering if we're somewhat overloading the concept of "normative".
I think I detect two uses:
1. Meaning that any RDF processor that doesn't recognize or honour the
semantics thus described is in some sense non conformant.
2. Meaning anything that is not informative, where informative material
exists solely to clarify that which is stated normatively, and can
otherwise be ignored without losing anything of the intended sense of RDF.
The dictionary definition I have is:
[[
normative adj.
of or establishing a norm.
]]
-- Concise Oxford Dictionary, ninth edition
which seems to apply in either case.
Of particular interest for me in this is the discussion of social meaning,
which is at heart intended to say that there are significant intended
aspects of RDF that are opaque to purely mechanical processes.
(This seems to mirror the distinction between any satisfying interpretation
and an intended interpretation.)
...
For the goals of RDF that have been set (e.g. [1][2]) I think we need a way
to distinguish between material that is normative in a formal,
machine-accessible fashion, and things that are normative in the sense of
intended interpretation by people.
At this point, one can play around with all sorts of words to describe
these cases. Maybe we should lake a lead from TimBL's Evolvability paper
[3] and talk in terms of "Normative{2}" (Human readable) and "Normative{6}"
(Expose logic of document)? Hmmm... I think the need here is to
distinguish case {6} from all the others {1}-{5}. I'm struggling for words
here, but I'm trying to draw a distinction something like:
Logically Normative or Formally Normative
Normative Intent
Informative
Is this a path we wish to consider? If so, what should we name the various
degrees of normative? If not, what exactly do we mean by
normative? Either way, I think we should probably be more explicit about
what the specification is saying.
#g
--
[1] http://www.w3.org/2000/03/rdf-tracking/#rdf-formal-semantics
[2] http://www.w3.org/2000/03/rdf-tracking/#rdfms-assertion
[3] http://www.w3.org/DesignIssues/Evolution.html
-------------------
Graham Klyne
<GK@NineByNine.org>
Received on Tuesday, 14 January 2003 12:19:14 UTC