ines2hugin

数据挖掘中的一算法 ines算法 c下实现的。适合初学习数据挖掘者借鉴

#!/usr/bin/perl -w
########################################################################
# File:   ines2hugin.pl                                                #
# Author: D. Nauck, BT Labs                                            #
#----------------------------------------------------------------------#
# Description:                                                         #
# ============                                                         #
# This Perl script translates a INES network file                      #
# into a HUGIN network file.                                           #
# 2002.07.18 adapted to new INES version by Christian Borgelt          #
########################################################################

use strict;

my $script = "Ines2Hugin";      # name of script
my $command = "ines2hugin";     # string to invoke script
my $author = "Detlef Nauck";
my $last_change = "18 Jul 2002, Christian Borgelt";
my $description = "Convert a Bayesian network in INES format into HUGIN format.\n\n";

my $ines_file;
my $hugin_file;
my @ines = ();
my $lines;
my $line;
my $size;
my $node;
my $domain;
my @values;
my $x_pos = 50;
my $y_pos = 70;
my $x_inc = 50;
my $y_inc = 70;
my $indent = "  ";
my $nodes = 0;
my $zero_row_sums = 0;
my $distributions = 0;

my $nargs = @ARGV;

print ("$script by $author (last change: $last_change)\n");
print ($description);

if ($nargs != 2){
    usage();
    exit (-1);
}

$ines_file  = $ARGV[0];
$hugin_file = $ARGV[1];

-r $ines_file or die ("Cannot open $ines_file for reading.\n");
$size = -s $ines_file;

print ("Reading $ines_file ...\n");
open (INES, "$ines_file") or die ("Cannot open $ines_file for reading.\n");;
@ines = <INES>;
close (INES);
$lines = @ines;

print ("done ($lines lines, $size bytes).\n");

print ("Starting conversion, creating output file $hugin_file ...\n");

open (HUGIN, ">$hugin_file") or die ("Cannot open $hugin_file for output.\n");

#-----------------------------------------------------------------------
# Header of Hugin Network File
#-----------------------------------------------------------------------
  print(HUGIN "net\n", "{\n",
        "  node_size = (80 40);\n",
        "  HRUNTIME_Monitor_InitStates     = \"5\";\n",
        "  HRUNTIME_Monitor_OpenGraph      = \"0\";\n",
        "  HRUNTIME_Monitor_GraphPrecision = \"100\";\n",
        "  HRUNTIME_Monitor_AutoUpdGraph   = \"0\";\n",
        "  HRUNTIME_Compile_ApproxEpsilon  = \"0.00001\";\n",
        "  HRUNTIME_Compile_Approximate    = \"0\";\n",
        "  HRUNTIME_Compile_Compress       = \"0\";\n",
        "  HRUNTIME_Compile_TriangMethod   = \"0\";\n",
        "  HRUNTIME_Propagate_AutoNormal   = \"1\";\n",
        "  HRUNTIME_Propagate_AutoSum      = \"1\";\n",
        "  HRUNTIME_Propagate_Auto         = \"0\";\n",
        "  HRUNTIME_Font_Italic            = \"0\";\n",
        "  HRUNTIME_Font_Weight            = \"400\";\n",
        "  HRUNTIME_Font_Size              = \"-12\";\n",
        "  HRUNTIME_Font_Name              = \"Arial\";\n",
        "  HRUNTIME_Grid_GridShow          = \"0\";\n",
        "  HRUNTIME_Grid_GridSnap          = \"1\";\n",
        "  HRUNTIME_Grid_Y                 = \"10\";\n",
        "  HRUNTIME_Grid_X                 = \"10\";\n}\n\n\n");
#-----------------------------------------------------------------------
# Translate domain information into node information
#-----------------------------------------------------------------------

# skip until first dom line is found

$line = shift(@ines);

while($line !~ m/^dom/){
    $line = shift(@ines);
}

# convert dom lines

while($line =~ m/^dom/){
    $node = $line;
    $node =~ s/^dom\((.*)\).*\s*/$1/;
    print HUGIN ("node $node\n{\n", $indent, "label    = \"$node\";\n",
                 $indent, "position = ($x_pos $y_pos);\n",
                 $indent, "states   = (\"");
    $domain = $line;
    $domain =~ s/.*\{ (.*) \}.*\s*/$1/;
    $domain =~ s/, /" "/g;
    print HUGIN ("$domain\");\n}\n\n");
    $line = shift (@ines);
    $x_pos += $x_inc;
    $y_pos += $y_inc;
    $y_inc *= -1;
    $nodes++;
}

while (@ines > 0){
    $line = shift (@ines);
    if ($line =~ m/^  prob\(/){
        $line =~ s/  prob/potential/;
        $line =~ s/=/\{/;
        $line =~ s/,/ /g;
        $line =~ s/\|/ \| /;
        print HUGIN ("$line","data =\n");
        $line = shift (@ines);
        while (($line !~ m/^$/) && ($line !~ m/^\};/)) {
            $line =~ s/\([ |0-9]+\)//;
            $line =~ s/\{/\(/g;
            $line =~ s/\}/\)/g;
            $line =~ s/\w+\s*://g;
            $line =~ s/,//g;
            if ($line !~ m/[1-9]/){ # if the are only zeros
                $line =~ s/0/1/g;   # then replace them by ones (equal prob.)
                $zero_row_sums++;
            }
            print HUGIN ($line);
            $line = shift(@ines);
            $distributions++;
        }
        print HUGIN ("}\n\n");
    }
}
close (HUGIN);

print ("done.\n");
print ("The network has $nodes nodes.\n");
print ("There are $distributions conditional probability distributions.\n");
if ($zero_row_sums > 0){
    print ("$zero_row_sums distributions with zero row sum have been detected\n");
    print ("and have been converted to equal distributions.\n");
}

sub usage{
    print ("Usage: $command <ines file> <hugin file>\n");
}