On 12/13/18 9:18 AM, Linas Vepstas wrote:
So maybe use GPN's, but then "hide" them behind "well-known"
DefinedPredicateNodes.
Since they're defined, they could be used for reasoning. I dunno. I
still don't really
know how to do reasoning ...
In principle reasoning can be performed on anything, even
GroundedPredicateNodes. All you need is to "axiomatize" them, for
instance if you want to say that `near` is symmetric you'd write
ImplicationScope (stv 1 1)
VariableList
Variable "$X"
Variable "$Y"
Evaluation
GroundedPredicate "near"
List
Variable "$X"
Variable "$Y"
Evaluation
GroundedPredicate "near"
List
Variable "$Y"
Variable "$X"
or if you already have higher knowledge about symmetry, maybe you'd
just write
Member (stv 1 1)
GroundedPredicate "near"
Concept "symmetric"
In practice I don't know how well the URE would chew on
GroundedPredicate though, as they have a special meaning to the
pattern matcher, as virtual clauses. Would need to try and fix what
needs to be fixed, quote what needs to be quoted, etc. Alexey and his
team have already stumbled on that kind of problems when trying to
incorporate neural nets to the URE.
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 [email protected].
To post to this group, send email to [email protected].
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/2dae083f-077c-1fe0-3d26-a84e8f7b5c2f%40gmail.com.
For more options, visit https://groups.google.com/d/optout.
