anomalous credulity

I have the following items in a PowerLoom knowledge base:

(defconcept nation (?n region human-social-grouping))
(assert (disjoint nation city))
(defconcept city (?c region))
(deffunction capital ((?n nation)) :-> (?c city))
(assert (total capital))

and I have 

(assert (city bogota))

(definstance colombia
    :axioms ((nation colombia)
	     (contained-in colombia south-america)
	     (= (capital colombia) bogota)))

when I was doing a query, I got the following odd results:

STELLA(7): (retrieve all ?x (and (region ?x) (contained-in ?x colombia)))
There are 5 solutions:
  #1: ?X=REGION-R
  #2: ?X=BOGOTA
  #3: ?X=COLOMBIA
  #4: ?X=|SK|(CAPITAL REGION-R)
  #5: ?X=|SK|(CAPITAL BOGOTA)

So I did a little further investigation:

STELLA(8): (ask (nation bogota))
UNKNOWN

OK, this doesn't surprise me, because I understand from earlier emails
that disjointness doesn't work properly.

But I'm still surprised that the skolemization proceeded anyway, on a
kind of credulous inference.  Then I saw the following:

STELLA(9): (ask (city (capital bogota)))
Processing check-types agenda...
WARNING: Type check violation on argument `@BOGOTA' in proposition
   (= (CAPITAL BOGOTA) ?v04).
   Argument must have type `NATION'.
   Warning occurred while parsing the proposition: 
   (KAPPA () (CITY (CAPITAL @BOGOTA)))
TRUE

and the following anomalous result:

STELLA(10): (ask (and (nation bogota)(city (capital bogota))))
Processing check-types agenda...
WARNING: Type check violation on argument `@BOGOTA' in proposition
   (= (CAPITAL BOGOTA) ?v04).
   Argument must have type `NATION'.
   Warning occurred while parsing the proposition: 
   (KAPPA () (AND (NATION @BOGOTA) (CITY (CAPITAL @BOGOTA))))
TRUE

Why does PowerLoom regard (nation bogota) as unknown in isolation, but
true in a conjunction?

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]