[popfile-commit] engine/Classifier Bayes.pm,1.186,1.187

[<jgr...@us...>]

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv5019/Classifier

Modified Files:
	Bayes.pm 
Log Message:
Default values for the dummy unsure bucket

Index: Bayes.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/Bayes.pm,v
retrieving revision 1.186
retrieving revision 1.187
diff -C2 -d -r1.186 -r1.187
*** Bayes.pm	4 Sep 2003 16:30:41 -0000	1.186
--- Bayes.pm	5 Sep 2003 21:18:51 -0000	1.187
***************
*** 407,419 ****
  
      $self->{colors__}{unclassified} = 'black';
  
      # SLM for unclassified "bucket" will always match the global setting
  
      $self->{parameters__}{unclassified}{subject} = $self->global_config_('subject');
  
      # Quarantine for unclassified will be off:
  
      $self->{parameters__}{unclassified}{quarantine} = 0;
! 
  }
  
--- 407,421 ----
  
      $self->{colors__}{unclassified} = 'black';
+     $self->{colors__}{unsure}       = 'black';
  
      # SLM for unclassified "bucket" will always match the global setting
  
      $self->{parameters__}{unclassified}{subject} = $self->global_config_('subject');
+     $self->{parameters__}{unsure}{subject}       = $self->global_config_('subject');
  
      # Quarantine for unclassified will be off:
  
      $self->{parameters__}{unclassified}{quarantine} = 0;
!     $self->{parameters__}{unsure}{quarantine}       = 0;
  }
  
***************
*** 608,612 ****
      my $term = $sum;
  
!     for my $i (1..$free/2) {
          $term *= $m / $i;
          $sum  += $term;
--- 610,614 ----
      my $term = $sum;
  
!     for my $i (1..($free/2-1)) {
          $term *= $m / $i;
          $sum  += $term;
***************
*** 697,701 ****
--- 699,706 ----
      # P(word|bucket) ^ word count and multiply to the score
  
+     my $word_count = 0;
+ 
      foreach my $word (keys %{$self->{parser__}->{words__}}) {
+         $word_count += 2;
          my $wmax = -10000;
  
***************
*** 719,723 ****
  
      foreach my $bucket (@buckets) {
!         $chi{$bucket} = chi2( $score{$bucket}, 2 * $matchcount{$bucket}, -int($score{$ranking[0]}/log(10)) * log(10) );
      }
  
--- 724,728 ----
  
      foreach my $bucket (@buckets) {
!         $chi{$bucket} = chi2( $score{$bucket}, $word_count, -int($score{$ranking[0]}/log(10)) * log(10) );
      }
  
***************
*** 785,789 ****
          }
  
!         $self->{scores__} .= "<hr><b>$language{Scores}</b><p>\n<b>Verdict: <font color=\"$self->{colors__}{$class}\">$class ($certainty)</font></b><p>\n<table class=\"top20Words\">\n<tr>\n<th scope=\"col\">$language{Bucket}</th>\n<th>&nbsp;</th>\n";
          $self->{scores__} .= "<th scope=\"col\">$language{Count}&nbsp;&nbsp;</th><th scope=\"col\">$language{Probability}</th></tr>\n";
  
--- 790,794 ----
          }
  
!         $self->{scores__} .= "<hr><b>$language{Scores}</b><p>\n<b>Verdict: <font color=\"$self->{colors__}{$class}\">$class ($certainty  $chi{$ranking[0]} $chi{$ranking[1]})</font></b><p>\n<table class=\"top20Words\">\n<tr>\n<th scope=\"col\">$language{Bucket}</th>\n<th>&nbsp;</th>\n";
          $self->{scores__} .= "<th scope=\"col\">$language{Count}&nbsp;&nbsp;</th><th scope=\"col\">$language{Probability}</th></tr>\n";
  
***************
*** 1160,1164 ****
  
      # Add the Subject line modification or the original line back again
!     if ( $classification ne 'unclassified' ) {
          if ( $self->global_config_( 'subject' ) ) {
              # Don't add the classification unless it is not present
--- 1165,1169 ----
  
      # Add the Subject line modification or the original line back again
!     if ( ( $classification ne 'unclassified' ) && ( $classification ne 'unsure' ) ) {
          if ( $self->global_config_( 'subject' ) ) {
              # Don't add the classification unless it is not present
***************
*** 1198,1202 ****
          # information from POPFile and wrapping the original message in a MIME encoding
  
!         if ( $classification ne 'unclassified' ) {
              if ( $self->{parameters__}{$classification}{quarantine} == 1 ) {
                  print $client "From: " . $self->{parser__}->get_header( 'from' ) . "$eol";
--- 1203,1207 ----
          # information from POPFile and wrapping the original message in a MIME encoding
  
!         if ( ( $classification ne 'unclassified' ) && ( $classification ne 'unsure' ) ) {
              if ( $self->{parameters__}{$classification}{quarantine} == 1 ) {
                  print $client "From: " . $self->{parser__}->get_header( 'from' ) . "$eol";
***************
*** 1236,1240 ****
      my $before_dot = '';
  
!     if ( $classification ne 'unclassified' ) {
          if ( ( $self->{parameters__}{$classification}{quarantine} == 1 ) && $echo ) {
              $before_dot = "$eol--$nopath_temp_file--$eol";
--- 1241,1245 ----
      my $before_dot = '';
  
!     if ( ( $classification ne 'unclassified' ) && ( $classification ne 'unsure' ) ) {
          if ( ( $self->{parameters__}{$classification}{quarantine} == 1 ) && $echo ) {
              $before_dot = "$eol--$nopath_temp_file--$eol";