line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::JA::Categorize::Categorizer; |
2
|
1
|
|
|
1
|
|
588
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
34
|
|
3
|
1
|
|
|
1
|
|
5
|
use warnings; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
25
|
|
4
|
1
|
|
|
1
|
|
1483
|
use Algorithm::NaiveBayes; |
|
1
|
|
|
|
|
5720
|
|
|
1
|
|
|
|
|
26
|
|
5
|
1
|
|
|
1
|
|
589
|
use Lingua::JA::Categorize::Result; |
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
16
|
|
6
|
1
|
|
|
1
|
|
37
|
use base qw( Lingua::JA::Categorize::Base ); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
125
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
__PACKAGE__->mk_accessors($_) for qw( brain ); |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
sub import { |
11
|
1
|
|
|
1
|
|
12094
|
use Algorithm::NaiveBayes::Util; |
|
1
|
|
|
|
|
865
|
|
|
1
|
|
|
|
|
68
|
|
12
|
1
|
|
|
1
|
|
7
|
use List::Util qw(min max sum); |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
69
|
|
13
|
1
|
|
|
1
|
|
6
|
no warnings 'redefine'; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
501
|
|
14
|
|
|
|
|
|
|
*Algorithm::NaiveBayes::Util::rescale = sub { |
15
|
0
|
|
|
0
|
|
0
|
my ($scores) = @_; |
16
|
0
|
|
|
|
|
0
|
my $min = min( values %$scores ); |
17
|
0
|
|
|
|
|
0
|
my $sum = sum( values %$scores ); |
18
|
0
|
|
|
|
|
0
|
$sum -= $min * ( keys %$scores ); |
19
|
0
|
|
|
|
|
0
|
for ( sort { $scores->{$b} <=> $scores->{$a} } keys %$scores ) { |
|
0
|
|
|
|
|
0
|
|
20
|
0
|
|
|
|
|
0
|
$scores->{$_} = ( $scores->{$_} - $min ) / $sum; |
21
|
|
|
|
|
|
|
} |
22
|
0
|
|
|
|
|
0
|
my $max = max( values %$scores ); |
23
|
0
|
|
|
|
|
0
|
for ( sort { $scores->{$b} <=> $scores->{$a} } keys %$scores ) { |
|
0
|
|
|
|
|
0
|
|
24
|
0
|
|
|
|
|
0
|
$scores->{$_} = sprintf( "%0.2f", $scores->{$_} / $max ); |
25
|
|
|
|
|
|
|
} |
26
|
0
|
|
|
|
|
0
|
return $scores; |
27
|
1
|
|
|
1
|
|
39
|
}; |
28
|
|
|
|
|
|
|
} |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
sub new { |
31
|
1
|
|
|
1
|
1
|
14
|
my $class = shift; |
32
|
1
|
|
|
|
|
14
|
my $self = $class->SUPER::new(@_); |
33
|
1
|
|
|
|
|
10
|
$self->brain( Algorithm::NaiveBayes->new( purge => 0 ) ); |
34
|
|
|
|
|
|
|
#use Devel::Size qw(size total_size); |
35
|
|
|
|
|
|
|
#use Devel::Peek; |
36
|
|
|
|
|
|
|
{ |
37
|
1
|
|
|
1
|
|
18
|
no warnings 'redefine'; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
692
|
|
|
1
|
|
|
|
|
1913
|
|
38
|
|
|
|
|
|
|
*Algorithm::NaiveBayes::Model::Frequency::do_predict = sub { |
39
|
0
|
|
|
0
|
|
0
|
my ( $self, $m, $newattrs ) = @_; |
40
|
|
|
|
|
|
|
# print "IN :", total_size($m), "\n"; |
41
|
0
|
|
|
|
|
0
|
my %scores = %{ $m->{prior_probs} }; |
|
0
|
|
|
|
|
0
|
|
42
|
0
|
|
|
|
|
0
|
while ( my ( $feature, $value ) = each %$newattrs ) { |
43
|
0
|
0
|
|
|
|
0
|
unless ( exists $m->{attributes}{$feature} ) { |
44
|
0
|
|
|
|
|
0
|
push( @{ $self->{no_match_features} }, $feature ); |
|
0
|
|
|
|
|
0
|
|
45
|
0
|
|
|
|
|
0
|
next; |
46
|
|
|
|
|
|
|
} |
47
|
|
|
|
|
|
|
else { |
48
|
0
|
|
|
|
|
0
|
push( @{ $self->{match_features} }, $feature ); |
|
0
|
|
|
|
|
0
|
|
49
|
|
|
|
|
|
|
} |
50
|
0
|
|
|
|
|
0
|
while ( my ( $label, $attributes ) = each %{ $m->{probs} } ) { |
|
0
|
|
|
|
|
0
|
|
51
|
0
|
|
0
|
|
|
0
|
my $p = ($attributes->{$feature} || $m->{smoother}{$label}); |
52
|
0
|
|
|
|
|
0
|
$scores{$label} += $p * $value; |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
#$scores{$label} += |
55
|
|
|
|
|
|
|
# ( $attributes->{$feature} || $m->{smoother}->{$label} ) * |
56
|
|
|
|
|
|
|
# $value; |
57
|
|
|
|
|
|
|
} |
58
|
|
|
|
|
|
|
} |
59
|
|
|
|
|
|
|
# print "OUT:", total_size($m), "\n"; |
60
|
0
|
|
|
|
|
0
|
Algorithm::NaiveBayes::Util::rescale( \%scores ); |
61
|
|
|
|
|
|
|
|
62
|
0
|
|
|
|
|
0
|
return \%scores; |
63
|
1
|
|
|
|
|
23
|
}; |
64
|
|
|
|
|
|
|
} |
65
|
1
|
|
|
|
|
5
|
return $self; |
66
|
|
|
|
|
|
|
} |
67
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
sub categorize { |
69
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
70
|
0
|
|
|
|
|
|
my $word_set = shift; |
71
|
0
|
|
|
|
|
|
my $user_extention = shift; |
72
|
0
|
|
|
|
|
|
$self->brain->{no_match_features} = []; |
73
|
0
|
|
|
|
|
|
$self->brain->{match_features} = []; |
74
|
0
|
|
|
|
|
|
my $score = $self->brain->predict( attributes => $word_set ); |
75
|
0
|
|
|
|
|
|
my $no_matches = $self->brain->{no_match_features}; |
76
|
0
|
|
|
|
|
|
my $matches = $self->brain->{match_features}; |
77
|
0
|
|
|
|
|
|
my $result = Lingua::JA::Categorize::Result->new( |
78
|
|
|
|
|
|
|
context => $self->context, |
79
|
|
|
|
|
|
|
score => $score, |
80
|
|
|
|
|
|
|
matches => $matches, |
81
|
|
|
|
|
|
|
no_matches => $no_matches, |
82
|
|
|
|
|
|
|
word_set => $word_set, |
83
|
|
|
|
|
|
|
user_extention => $user_extention, |
84
|
|
|
|
|
|
|
); |
85
|
0
|
|
|
|
|
|
return $result; |
86
|
|
|
|
|
|
|
} |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
sub save { |
89
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
90
|
0
|
|
|
|
|
|
my $save_file = shift; |
91
|
0
|
|
|
|
|
|
$self->brain->save_state($save_file); |
92
|
|
|
|
|
|
|
} |
93
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
sub load { |
95
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
96
|
0
|
|
|
|
|
|
my $save_file = shift; |
97
|
0
|
|
|
|
|
|
my $brain = $self->brain; |
98
|
0
|
|
|
|
|
|
$brain = Algorithm::NaiveBayes->restore_state($save_file); |
99
|
0
|
|
|
|
|
|
$self->brain($brain); |
100
|
|
|
|
|
|
|
} |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
1; |
103
|
|
|
|
|
|
|
__END__ |