W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > November 2001

RE: Issue rdfms-assertion

From: Dan Brickley <danbri@w3.org>
Date: Fri, 16 Nov 2001 11:51:41 -0500 (EST)
To: <w3c-rdfcore-wg@w3.org>
Message-ID: <Pine.LNX.4.30.0111161121520.15883-100000@tux.w3.org>

On Thu, 15 Nov 2001, Jeremy Carroll wrote:

> I agree with Brian,
> Dan:
> >
> > On the contrary: the RDF Core WG is in the W3C Technology
> > and Society domain for a reason: to make the connection
> > between bits on the wire and social obligations.
> >
> > There is some crazy laywer running around trying to encourage
> > web site owners to screw up their P3P policy files in such
> > a way as to be able to disclaim responsibility for their content.
> > [P3P isn't quite written in RDF, but it was supposed to
> > be, and should be, in a future revision, I hope].
> > I suggest that this WG has an obligation to say that no,
> > that's not consistent with the community's agreement
> > about how this technology works.
> The problem is that this is not about the meaning of the RDF, but about the
> meaning of the RDF in a particular context.

Yes, though we need to go a little further than we have to date in
claiming that RDF content is meaningful. We then need W3C et al to provide
document formatting and protocol contexts (XHTML, SVG, mime, HTTP,
SOAP...) within which this meaningful content can be exposed.

RDF data *isn't* an arbitrary graph structure, and DanC is write to say
that we need the RDF spec to be clear on this. We claim that RDF data can
be _about_ things in the world. The Model Theory alone doesn't give us
that. RDF uses the URI specification as its primary means of anchoring arbitrary
graphs to the physical and social world. The meaning of a piece of RDF
isn't fixed by the MT alone, but by the meanings that the URI spec
plus deployed Web practice gives to URI name strings. The meaning of RDF
content is furthermore fixed by the prose and rules associated with the
schemas/ontologies it draws upon. Where  those definitions are murky, the
meaning of RDF instance data is murky.

This is fine. We should say things to the effect that...:

 (i) RDF content has, on a good day, propositional content; it corresponds
     to statements or propositions about entities in the real social/legal world

(ii) the Model Theory assumes such a correspondence, but does not itself
     specify one.

(iii) the correspondence between RDF content and propositions about the
     world is fixed in part by facts about individual URI names (ie. what
     they denote); in part by facts about the URI specification (rules
     about how URI names can denote, (sometimes) fail to denote, etc.);
     in part by the machine-readable rules associated with classes,
     properties, and indivduals mentioned in the RDF content (or
     indirectly through the schemas/ontologies used); and in part by
     the natural language definitions specified in these schemas.

(iv) the formal components of RDF (MT etc.) do not guarantee that any
     piece of RDF instance data is meaningful, or that it corresponds to
     propositional content.  It may use incoherently
     designed or inadequately defined classes and properties; or it may
     use URI names whose referents are for some reason unclear or indeterminate.

(v)  While the RDF design explicitly notes that RDF is intended to have
propositional content, despite these various obstacles, the core RDF specs
are agnostic regarding the various options for deploying this content in
the Web. In particular, RDF core does not enumerate the XML vocabularies
within which it may be embedded and retain its propositional nature
nor define mechanisms that distinguish the roles played by the XML
elements that may contain it in wellformed XML  markup. RDF Core does not
claim that all XML document and data formats can be interpreted as having
propositional content, but notes for eg. that SVG's 'metadata' element and
(X)HTML's 'head' element are commonly understood in
this way. Similarly, the Core specs don't address protocols (HTTP) or
container formats (MIME), nor attempt to associate the propositional
content of RDF data with any particular real world entities such as
webmasters, document authors, editors, service providers.



Received on Friday, 16 November 2001 11:51:41 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:24:06 UTC