Hi Nil,
On Fri, Oct 6, 2017 at 6:30 AM, Nil Geisweiller <ngeis...@googlemail.com>
wrote:
>
>
> On 10/06/2017 02:10 AM, Linas Vepstas wrote:
>
>> it would be nice to have a fast crisp prover so that the system could
>> jump to conclusions, and pln more slowly in the background.
>>
>
> Yes, even for our rule engine alone there is a benefit to that. On top of
> being faster to evaluate, crisp rules tend to have less premises than their
> probabilistic counterparts.
>
> Then the question is how to set the TV of these conclusions. If the axioms
> are crisps with (stv 1 1) or (stv 0 1), then the conclusions would be (stv
> 1 1) or (stv 0 1). But if the axioms are non-crisp, then I guess the crisp
> rules could set (stv 1 Epsilon) or (stv 0 Epsilon), just to express that
> something is possibly true or false. Or else we can create a new TV type
> for it.
>
You don't need a new TV type, You can just store it in parallel, as just
another value on an atom. Recall that the current TV is stored by saying
atom->setValue (PredicateNode("*-TruthValueKey-*"), some_tv);
you could just store
atom->setValue (PredicateNode("*-CrispTruthKey-*"), crisp_tv);
and look it up that way, if/when you need it, for example to provide a
"backbone" around which fuzzy explorations can be done.
So here's a completely different but related idea: First, use a crisp
reasoner to deduce what happens whenever strength>0.9999. Next, do it
again, but now for strength>0.8. (but still using the crisp reasoner: just
take strength>0.8 to mean "true"). This should have a "broader" set of
consequences. Do it again for strength>0.6 - this causes even more
possibilities to be explored.
It seems like these three cases can be treated as "lower bounds" of what we
might expect PLN to find. That these could be used to guide/limit what
PLN explores.
Alternately, if this was fast enough, you could do this 100 times for 100
different truth cutoffs, and build up a distributional TV...
I find this idea exciting! It seems plausible, doable ...
--linas
> Nil
--
You received this message because you are subscribed to the Google Groups
"opencog" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to opencog+unsubscr...@googlegroups.com.
To post to this group, send email to opencog@googlegroups.com.
Visit this group at https://groups.google.com/group/opencog.
To view this discussion on the web visit
https://groups.google.com/d/msgid/opencog/CAHrUA36c9ShsmARYTzPs6JP61CPjL%2BvAz9zx4yHEG-34tkn%3DzA%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.
