Rule

I may have been trying to do the impossible here. The rule below is one of many similar variants that I tried.

The rule states:

Forall x,y,z,a,b,c

If agent x and y share an act a and a's consequence is in an unknown similie relation

Then

If x and y share an action b with the same consequence c,

Then

Make the consequence c a consequence for action a

I was trying to capture the idea that if x and y can proliferate and x and y can live then maybe, possibly the consequence for living is a consequence for proliferating.

;;(describe (assert forall ($x $y $z $a $b $c)
;;		  &ant((build &ant (build agent *x
;;			       		  act *a)
;;			      cq *z)
;;			(build &ant (build agent *y
;;			       		  act *a)
;;			      cq *z)
;;			(build object1 *z
;;			       rel  (build mod ("unknown concept")
;;	 		     		   head ("similie/metaphor action"))
;;			       object2 *z)
;;		  )
;;		  &cq(
;; 			(build forall ($b $c)
;;				&ant((build ant (build agent *x
;;				    	  	act *b)
;;					    cq *c)
;;				     (build ant (build agent *y
;;					      	  act *b)
;;			 		    cq *c))
;;		     		&cq((build ant (build agent *x
;;			       	       		act *a)
;;			      	           cq *c)
;;		     		    (build ant (build agent *y
;;			       	       		act *a)
;;			      	    	   cq *c)
;;			))))

There are two problems I can immediately see:

Problems

1.The rule is doing too much. It may be the case that the rule is too involved and should be broken down more. However I reread it numerous times and am convinced I had something that should work. However, Ehrlich's algorithm may have prevented this from ever happening because....

2. Ehrlichs algorithm only searches on CQ ! ANT variants. The consequence I create that links a new consequence to proliferate will likely be of the form CQ ANT and thus never seen by Ehrlich's algorithm.

3. Some vairants of the rule threw CASSIE into an infinite loop when I asked it to deduce information based on the rule :)