[popfile-commit] engine/Classifier Bayes.pm,1.131,1.132

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

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv28528

Modified Files:
	Bayes.pm 
Log Message:
Merge biljir that improves the display of probabilities for an individual message

Index: Bayes.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/Bayes.pm,v
retrieving revision 1.131
retrieving revision 1.132
diff -C2 -d -r1.131 -r1.132
*** Bayes.pm	20 Apr 2003 21:09:06 -0000	1.131
--- Bayes.pm	20 Apr 2003 21:12:33 -0000	1.132
***************
*** 235,238 ****
--- 235,253 ----
  }
  
+ # ---------------------------------------------------------------------------------------------
+ #
+ # get_sort_value_ behaves the same as get_value_, except that it returns not_likely__ rather
+ # than 0 if the word is not found.  This makes its result more suitable as a sort key for bucket
+ # ranking.
+ #
+ # ---------------------------------------------------------------------------------------------
+ sub get_sort_value_
+ {
+     my ($self, $bucket, $word) = @_;
+     my $v = get_value_($self, $bucket, $word);
+     return $self->{not_likely__} if $v == 0;
+     return $v;
+ }
+ 
  sub set_value_
  {
***************
*** 545,551 ****
--- 560,568 ----
  
      my %score;
+     my %matchcount;
  
      for my $bucket (@buckets) {
          $score{$bucket} = $self->{bucket_start__}{$bucket};
+         $matchcount{$bucket} = 0;
      }
  
***************
*** 575,578 ****
--- 592,596 ----
              my $probability = get_value_( $self, $bucket, $word );
  
+             $matchcount{$bucket} += $self->{parser__}{words__}{$word} if ($probability != 0);
              $probability = $self->{not_likely__} if ( $probability == 0 );
              $wmax = $probability if ( $wmax < $probability );
***************
*** 646,650 ****
  
          $self->{scores__} .= "<hr><b>$language{Scores}</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{Probability}</th></tr>\n";
  
          foreach my $b (@ranking) {
--- 664,668 ----
  
          $self->{scores__} .= "<hr><b>$language{Scores}</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";
  
          foreach my $b (@ranking) {
***************
*** 658,662 ****
               }
  
!              $self->{scores__} .= "<tr>\n<td><font color=\"$self->{colors__}{$b}\"><b>$b</b></font></td>\n<td>&nbsp;</td>\n<td>$probstr</td>\n</tr>\n";
          }
  
--- 676,680 ----
               }
  
!              $self->{scores__} .= "<tr>\n<td><font color=\"$self->{colors__}{$b}\"><b>$b</b></font></td>\n<td>&nbsp;</td>\n<td align=\"right\">$matchcount{$b}&nbsp;&nbsp;&nbsp;&nbsp;</td>\n<td>$probstr</td>\n</tr>\n";
          }
  
***************
*** 665,669 ****
          $self->{scores__} .= "<tr>\n<th scope=\"col\">$language{Word}</th><th>&nbsp;</th><th scope=\"col\">$language{Count}</th><th>&nbsp;</th>\n";
  
!         foreach my $bucket (@buckets) {
              my $bucketcolor  = $self->get_bucket_color( $bucket );
              $self->{scores__} .= "<th><font color=\"$bucketcolor\">$bucket</font></th><th>&nbsp;</th>";
--- 683,688 ----
          $self->{scores__} .= "<tr>\n<th scope=\"col\">$language{Word}</th><th>&nbsp;</th><th scope=\"col\">$language{Count}</th><th>&nbsp;</th>\n";
  
!         foreach my $ix (0..(@buckets > 7? 7: @buckets)) {
!             my $bucket = $ranking[$ix];
              my $bucketcolor  = $self->get_bucket_color( $bucket );
              $self->{scores__} .= "<th><font color=\"$bucketcolor\">$bucket</font></th><th>&nbsp;</th>";
***************
*** 672,681 ****
          $self->{scores__} .= "</tr>";
  
!         my @ranked_words = sort {$self->get_value_( $ranking[0], $b ) <=> $self->get_value_( $ranking[0], $a )} keys %{$self->{parser__}->{words__}};
  
          foreach my $word (@ranked_words) {
              my $known = 0;
  
!             foreach my $bucket (@buckets) {
                  if ( $self->get_value_( $bucket, $word ) != 0 ) {
                     $known = 1;
--- 691,700 ----
          $self->{scores__} .= "</tr>";
  
!         my @ranked_words = sort {$self->get_sort_value_( $ranking[0], $b ) <=> $self->get_sort_value_( $ranking[0], $a )} keys %{$self->{parser__}->{words__}};
  
          foreach my $word (@ranked_words) {
              my $known = 0;
  
!             foreach my $bucket (@ranking) {
                  if ( $self->get_value_( $bucket, $word ) != 0 ) {
                     $known = 1;
***************
*** 692,700 ****
                  my $base_probability = $self->get_value_( $ranking[0], $word );
  
!                     foreach my $bucket (@buckets) {
                      my $probability  = get_value_( $self, $bucket, $word );
                      my $color        = 'black';
  
!                     if ( $probability >= $base_probability ) {
                          $color = $self->get_bucket_color( $bucket );
                      }
--- 711,720 ----
                  my $base_probability = $self->get_value_( $ranking[0], $word );
  
!                 foreach my $ix (0..(@buckets > 7? 7: @buckets)) {
!                     my $bucket = $ranking[$ix];
                      my $probability  = get_value_( $self, $bucket, $word );
                      my $color        = 'black';
  
!                     if ( $probability >= $base_probability || $base_probability == 0 ) {
                          $color = $self->get_bucket_color( $bucket );
                      }