Re: car color defaults: a story about Scoped Negation As Failure/log:notIncludes

On Thu, 2005-07-07 at 17:13 -0400, Peter F. Patel-Schneider wrote:
> > Here's a sort of story we've written to explain and motivate
> > log:notIncludes, aka Scoped Negation As Failure[1].
> > 
> > [[
> > Because a formula is a finite size,
> > [...]
> > ]]
> >  -- section "Implementing defaults and log:notIncludes" 
> >   of part "Reaching out onto the Web"
> >   of the Semantic Web Tutorial Using N3
> >   http://www.w3.org/2000/10/swap/doc/Reach#Implementi
> 
> 
> This isn't really the case in examples like:

Yes... cwm goes off into the weeds when you ask it to compute the
log:conclusion of things like "fred's a person;
every person has a father who is a person" but N3 semantics
are intended to work with things like Euler that use
backward chaining, so you're right to point out that
this is an open issue in the design of N3.

I just added it to the list of issues we track:

log:conclusion, log:notIncludes, and infinite formulas 
http://lists.w3.org/Archives/Public/public-cwm-bugs/2005Jul/thread.html

The use cases to date all involve finite formulas;
e.g. the ones written in documents that you can GET
via http, so the "you aren't gonna need it" principle
suggests that the domain of log:notIncludes be
constrained to finite formulas. But I don't think
I'm ready to say that's the answer just yet.


[...]
> Of course, it may be that cwm is so incomplete that a finite set of 
> premises always has a finite set of conclusions, but this is not true in 
> many formalisms, including the formalism (forward-chaining rules with 
> existentials in the consequent, as seen in 
> http://www.w3.org/2003/Talks/0520-www-tf1-b3-rules/slide15-0.html) that 
> appears to underly cwm.
> 
> [...]

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Thursday, 7 July 2005 22:38:32 UTC