Re: syntax question

>>>>> "TR" == Thomas Russ <[email protected]> writes:

    TR> On Feb 13, 2006, at 3:03 PM, [email protected] wrote:

    >> 
    >> Is there a convenient way of defining a concept as a PROPER
    >> sub-concept of another?  IIUC, defconcept with super-concepts makes no
    >> assumptions about the new concept being a proper sub-concept of its
    >> parent.  However, for a lot of modeling applications, this is the
    >> standard case.  It would be nice if there was a handy syntactic form
    >> for this.  Ordinarily, I would simply define a Lisp macro to generate
    >> an expansion that was what I wanted, but the combination of Stella +
    >> macros is a daunting one...

    TR> There is generally no need for this.

    TR> While it is true that PowerLoom doesn't make any particular
    TR> assumptions about whether any subconcept is, in fact, a proper
    TR> subconcept or an equivalent concept, any of the normal means
    TR> of defining a subconcept will give you what is operationally
    TR> the effect you want.

[...snip...]

    TR> If you really want to deny the equivalence, this is possible
    TR> by adding a particular rule that denies the subset-of relation:

    TR> (defconcept a)
    TR> |c|A
    TR> ? (defconcept b (?x a)
    TR>    :axioms (not (subset-of a b)))
    TR> |c|B
    TR> ? (defconcept c (?x b)
    TR>    :axioms (not (subset-of b c)))
    TR> |c|C
    TR> ? (retrieve all (proper-subrelation a ?x))
    TR> There are 2 solutions:
    TR>    #1: ?X=B
    TR>    #2: ?X=C
    TR> ? (retrieve all (proper-subrelation b ?x))
    TR> There is 1 solution:
    TR>    #1: ?X=C
    TR> ? (ask (not (subset-of a c)))
    TR> TRUE

Thank you very much.  I wasn't sure exactly how to do this, and you've
answered my question handily.

Now, if I were just writing common-lisp, I would do something like

(defmacro my-concept (conc-name supers &rest args)
  `(defconcept ',conc-name ',supers
     ,@args
     :axioms ,(loop for x in supers
                 collect `(not (subset-of ',conc-name ',x)))))

[I know that's not quite right, but you get the idea.]

Related to this, have you found a nice idiom for capturing disjoint
subclassing?  

I see that DISJOINT, COVERING, and DISJOINT-COVERING provide the raw
materials for this.  I was just wondering if you've found a
particularly convenient syntactic sugar for this, to make it easy to
develop an axiomatization.  Perhaps it's not necessary, and I just
need to accustom myself to the notation as is.

thanks,
Robert
     
-- 
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]