W3C home > Mailing lists > Public > semantic-web@w3.org > July 2006

RE: Semantic content negotiation (was Re: expectations of vocabulary)

From: Xiaoshu Wang <wangxiao@musc.edu>
Date: Wed, 26 Jul 2006 14:15:16 -0400
To: "'Danny Ayers'" <danny.ayers@gmail.com>
Cc: "'Reto Bachmann-Gmür'" <reto@gmuer.ch>, <public-semweb-lifesci@w3.org>, "'Semantic Web'" <semantic-web@w3.org>
Message-ID: <000301c6b0df$72036ae0$4a741780@bioxiao>

> (I don't know either way if Accept-Vocabulary specifically is 
> a good idea, but it illustrates the point, there doesn't seem 
> to be any architectural obstacle to content negotiation at a 
> semantic level).

Any specification must have semantics to drive certain actions.  The
semantic domain dealt in HTTP protocol does not and should not overlap with
the semantic domain of the RDF model.  The semantic domain of HTTP is
Information Resources.  Each IR is treated as an atomic unit and HTTP GET
worries how to get a representation of that resource across the network.
Whether the representation is in what langauge and in what format does not
break that the assumption that the resource is atomic.  Accept langauge,
character encoding, Synonymy etc., is all fine because it does not break
that assumption.  But if you extend the HTTP in a way that you will break
the resource into various pieces, you goes out of the semantic domain of the
HTTP GET. You probably need to do an HTTP POST or use a different URI, such
as by appending some query parameters, to alter the state of the
representation, but not by HTTP GET.

It is a misguided argument to say: ... at a semantic level.  Without a
clearly defined semantic domain, this kind of sentences can be used to argue
for anything.  Is this what you want?

Received on Wednesday, 26 July 2006 18:24:59 UTC

This archive was generated by hypermail 2.4.0 : Tuesday, 5 July 2022 08:44:57 UTC