Author image Terrence Brannon

NAME

AI::Proplog - Propositional logic engine

SYNOPSIS

 use strict;
 use AI::Proplog;
 
 my $p = new AI::Proplog;
 #
 # assert some facts:
 #
 
 # cs requirements are basic cs, math, advanced cs, an engr rec and nat. sci.
 $p->a( cs_req => qw(basic_cs math_req advanced_cs engr_rec natural_science));
 
 # basic cs requires an intro req, comp org, advanced programming and theory
 $p->a( basic_cs  => qw(intro_req comp_org adv_prog theory) );
 
 # and so forth
 $p->a( intro_req => 'intro_cs');
 $p->a( intro_req => qw(introI introII) );
 $p->a( math_req  => qw(calc_req finite_req alg_req) );
 $p->a( calc_req  => qw(basic_calc adv_calc) );
 $p->a( basic_calc => qw(calcI calcII) );
 $p->a( basic_calc => qw(calcA calcB calcC) );
 $p->a( adv_calc   => 'lin_alg');
 $p->a( adv_calc   => 'honors_linalg');
 $p->a( finite_req => qw(fin_structI stat) );
 $p->a( alg_req    => 'fin_structII');
 $p->a( alg_req    => 'abs_alg');
 $p->a( alg_req    => 'abs_alg');
 
 # here we assert a bunch of facts:
 
 # the following things have been taken:
 # cs intro, computer org, advanced programming, and theory
 $p->apl( qw(intro_cs comp_org adv_prog theory) );
 
 # now do a bottom up search of the fact/rule space to see if the 
 # basic cs requirements have been met
 my $R = $p->bottom_up('basic_cs');
# or:  my $R = $p->top_down('basic_cs');
 print "The basic cs requirements have";
 print $R ? : ' not'
 print " been met\n";   

DESCRIPTION

This module allows one to assert propositions and rules based on propositions and then test new hypotheses based on the "universe" of existing propositions and rules. The implemented search algorithms are naive, inefficient and not-foolproof. For example the top-down engine will run without termination on the following code:

    $p->a( x => 'y' );
    $p->a( y => 'x' );

On the other hand, the code body is quite small (smaller than the original Pascal examples in the textbook) and thus serves as an excellent vehicle for study of such programs.

It is written based on the same interpreters described in "Computer with Logic: Logic Programming with Prolog", a book by David Maer and David S. Warren.

USAGE

First you use AI::Proplog then you assert rules and facts using a(). If you have a bunch of facts you want to assert, you can do so conveniently with apl() as exmplified in the SYNOPSIS.

Then you search your universe of truths in a top down or bottom-up manner, depending on what you want to do with the results. If you would like to know every truth that can result from your fact/rule space, then use bottom_up and inspect $p-{established}> with Data::Dumper to see all the derived propositions.

On the other hand, if you are more interested in simply finding out if a certain proposition holds based on the fact/rule base, then top_down() will likely be more efficient. This is because it will terminate as soon as a proposition is derived which matches your one of interest. However, top_down() does do backtracking, while bottom_up() simply plows through your fact/rule base one time finding every truth it can, and then testing your queried proposition.

API

  • a($then, @if)

    This function is used to model the human "if-then." It takes a list as its arguments. The first argument is the "then" and the remaining arguments are the "if", meaning the things which must be satisfied for the "then" to be true. Note: if there is no @if, then $then is true unconditionally. In other words, calling a with one argument asserts that argument to be true.

  • apl(@true)

    This function is used to state a bunch of facts. It is a convenience function. In other words, this:

      @fact = qw(smc.edu caltech.edu pcc.edu);
      $p->apl(@fact);
    
      is completely equivalent to this:
    
      @fact = qw(smc.edu caltech.edu pcc.edu);
      $p->a($_) for @fact;

TODO

Probablistic propositions:

 # there is a 0.6 chance that a souffle rises is beaten well and it is quiet
    $p->a( souffle_rises(0.6) => qw(beaten_well quiet_while_cooking) ) ;

 # there is a 0.1 chance that it will rise based on luck
    $p->a( souffle_rise(0.1 ) => 'have_luck');

 # there is thus 0.3 chance that it will NOT rise 

    $p->a( beaten_well(0.4)   => 'use_whisk');
    $p->a( beaten_well(0.6)   => 'use_mixer');

    $p->a( quiet_while_cooking(0.8) => 'kids_outside');
    $p->a( have_luck(0.3) => 'knock_on_wood');
 
 # tell me what the chances are of this souffle rising if the kids are
 # outside and I use whisk...
    $p->apl( qw(use_whisk kids_outside) );

 $p->top_down('souffle_rise')

use strict terms

Right now if a single term shows up there is no warning, thus if I mis-spell a term name (as I did when making the test suite for this), then the program will search for something that no-one intended it to.

AUTHOR

T. M. Brannon, <tbone@cpan.org>

I would like to thank nardo and IDStewart of www.perlmonks.org for their help in debugging my test suite for me. I would have never got this module out so fast if it weren't for thie speedy help.

SEE ALSO

  • "Computer with Logic: Logic Programming with Prolog", a book by David Maer and David S. Warren

  • Array::PatternMatcher

  • Quantum::Superpositions

  • Quantum::Entanglements