inference limitations

[email protected]
Newsgroups gmane.comp.ai.powerloom
Message-ID <[email protected]>
I was trying to create an indicator variable that would reify a
particular relationship, using a PowerLoom function:

(defrelation hostile-to (?hostile-er ?hostile-ee))

(defconcept affinity)
(assert (affinity friendly))
(assert (affinity hostile))
(assert (affinity neutral))
(assert (closed affinity))

(deffunction disposition (?g1 ?g2) :-> (?a affinity))
(assert (forall (?g1 ?g2)
		(<=> (hostile-to ?g1 ?g2)
		     (= (disposition ?g1 ?g2) hostile))))
(defconcept tribe-a (tribe)
  :axioms ((has-name tribe-a "Maquinna")
	   (operates-in tribe-a region-r)
	   (hostile-to tribe-a tribe-c)))

STELLA(4): (ask (hostile-to tribe-a tribe-c))
Processing check-types agenda...
TRUE

I was a little surprised to find that the functional synonym didn't
work properly (or what I thought of as properly):

STELLA(5): (ask (= (disposition tribe-a tribe-c) hostile))
TRUE
STELLA(6): (ask (= (disposition tribe-a tribe-c) friendly))
UNKNOWN

I tried using ask...not, to see if the second case above was a case
where limited effort was being devoted to disproof, but I don't
believe so:

STELLA(7): (ask (not (= (disposition tribe-a tribe-c) friendly)))
UNKNOWN

Interestingly, I am able to find the disposition using retrieve:

STELLA(8): (retrieve ?x (= (disposition tribe-a tribe-c) ?x))
There is 1 solution so far:
  #1: ?X=HOSTILE

[Note: I have found that since I am working with human social
groupings as entities, it's not at all obvious how I can use
individuals rather than concepts, since containment relations are
prominent.  This means that I end up having a lot of meta-relations,
e.g., tribe-a and tribe-c are concepts, rather than individuals.
Could this be responsible for some of my problems?]

Thanks,
R
-- 
Robert P. Goldman
Senior Scientist
Smart Information Flow Technologies (d/b/a SIFT, LLC)

211 N. First St., Suite 300
Minneapolis, MN 55401

Voice:	(612) 384-3454
Email:    [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.