Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!snorkelwacker.mit.edu!ira.uka.de!sun.rhrk.uni-kl.de!uklirb!shell
From: news@dftsrv.gsfc.nasa.gov
Newsgroups: comp.ai.shells
Subject: Re: CLIPS problem
Message-ID: <8058@uklirb.informatik.uni-kl.de>
Date: 11 Jun 91 22:33:12 GMT
References: <8023@uklirb.informatik.uni-kl.de>
Sender: shell@uklirb.informatik.uni-kl.de
Distribution: world
Organization: NASA GSFC
Lines: 53
Approved: shell@dfki.uni-kl.de
Posted-Date: Sun Jun 16 11:48:47 GMT 1991

In article <8023@uklirb.informatik.uni-kl.de>, causse@lifia.imag.fr (Olivier Causse) writes:
> Here is the problem, I'd like to express the following LHS condition
> 
> (defrule problem
>   [(<<preceding patterns>>)]
>   (not
>     (and 
> 	(A ?x ?y)
> 	(B ?x ?z&:(neq ?z ?y))
>     )
>   )
>   [(<<other patterns>>)]
> =>
>  ...
> )
> 
> I know that AND and OR logic blocks may not be placed inside of a
> NOT pattern.
> 
> In fact, I'd like to know
> 	
> 	1) Why such a restriction ?
> 	2) How to express the same condition in another (generic) way ?

Well the main purpose of the NOT operator is to test for the
non-existance of a pattern.

One good reason to have such a restriction is the ambiguity
of the statement.  Does the AND imply that you are testing for
the existing of both patterns?  If what you want to do is check
for multiple patterns that do not exist you can just have
multiple not statements as follows:

	...
	(not (A ? ?))
	(not (B ? ?))
 ...

Or in your case are you just using it as a negation 
operator?  If this is true, wouldn't the following be 
logically equivalent:

(defrule problem
   [(<<preceding patterns>>)]
 	(A ?x ?y)
 	(B ?x ?z&:(eq ?z ?y))
   [(<<other patterns>>)]
 =>

-------------
Robert Dominy
NASA Goddard Space Flight Center