RE: NAF v. SNAF - where is this being addressed?

> Sorry if I wasn't clear. I was using "monotonic" in the 
> rather limited 
> sense of monotonic reasoning rather than the sense that facts on the 
> semantic web are/should be eternal and unchangeable.

These two senses are like two sides of the same coin. 
In mathematical logic, monotonic reasoning is only meaningful 
since true assertions are preserved eternally (that is, in
conjunction with monotonic change).
 
> Given a set of assertions and axioms an OWL reasoner will only make 
> deductions which would remain true if further such assertions were 
> added. This seems to be a useful feature given the fundamental open 
> world assumption behind the semantic web.

The "Open-World Assumtion (OWA)" is independent of the question 
of monotonicity. It just implies that failure to prove does not
amount to falsity. There are formalisms (such as extended
logic programs with two kinds of negation) where you can have 
the OWA, but still use NAF, e.g. for expressing heuristically
defined properties, or you can freely combine the OWA with 
the CWA.  

> > And again: the important point about a Web-oriented use
> > of NAF is not really "scope" (in the sense of provenance)
> > but its relationship to definitive knowledge (as captured
> > by N3's "definitiveDocument" construct).
> 
> I think the term "scope" is intended to convey things just like the 
> definitiveDocument construct.

But relativizing an inference to some source KB (as its "scope")
and declaring that source to represent definitive knowledge
about a property p are two different things.

-Gerd

Received on Wednesday, 6 July 2005 09:29:02 UTC