Re: Paradoxes are bugs on the SW was: Re: questions on assertion

Joshua Allen wrote:

[[
> So what is the meaning of assertion in RDF? If triples are, as you
say,
> unasserted, what is the meaning of

> "This document describes a model theory for RDF(S) which treats the
> language as simple assertional language, in which each triple makes a
> distinct assertion and the meaning of any triple is not changed by
adding
> other triples"

Hold on.  Just because someone makes an assertion does not mean that you
believe it.  How wonderful it would be if I could just drop one simple
RDF triple onto the Internet (joshuaa@microsoft.com isThe kingOfWorld)
and have that become true.

]]

True, true, but RDF does not address "belief" and "trust" being at the
bottom layer of the cake.

[[
Any legitimate system is going to have to decide which triples to trust
and which to distrust (and in some cases, place on a continuum of
trustworthiness).  Using triples to assert trustworthiness (or veracity)
of other triples does not at all change the meaning of the original
triple.  It simply gives you additional information for choosing which
triples to believe.
]]

By this critereon, RDF is not in and of itself a "legitimate system". I
agree to the extent that RDF does not _in and of itself_ make the Semantic
Web. This, however, has been a point of strong contention -- I suspect that
some folks _do believe_ that RDF alone can be used to construct the
"Semantic Web".

[[
Example:
A. Joe says "There are aliens in my house"
B. John says "Joe doesn't have a house"

The *meaning* of statement "A" is clear.
]]

Really? Translate into RDF please. "A" is not a simple RDF statement. Can
you construct an RDF graph that has precisely this meaning?

[[
Statement "B" does not in any
way affect the meaning of statement "A".  In fact, it doesn't even
affect the truth of statement A.  The listener is left to make a
judgment call about which assertion he accepts and which he rejects.
Rejecting a triple is not the same as changing its meaning.
]]

Presumably "A" and "B" label different RDF _graphs_ not single _triples_.
The RDF model theory alludes to multiple graphs, but doesn't directly or
definitively deal with such issues i.e. when there are multiple graphs
(corresponding perhaps to multiple documents on the Web) what is "true"?

I have suggested that "colored triples"
http://lists.w3.org/Archives/Public/www-rdf-logic/2002May/0123.html whereby
a triple is colored by the base URI of the document that contains it (note
that the base URI of a _document_ is not subject to change by any value of
an xml:base attribute within any XML that might be contained by the
document). This would be one direction toward sorting out these issues --
which I agree are important for any "real world" use of the SW.

Jonathan

Received on Tuesday, 9 July 2002 14:56:51 UTC