line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Lingua::EN::Segmenter::Evaluator; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
=head1 NAME |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
Lingua::EN::Segmenter::Evaluator - Evaluate a segmenting method |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
=head1 SYNOPSIS |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
my $tiling_segmenter = Lingua::EN::Segmenter::TextTiling->new(); |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
foreach (@ARGV) { |
12
|
|
|
|
|
|
|
my $input = read_file($_); |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
print "\nFile name: $_\n"; |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
printf "Results from TextTiling algorithm: |
17
|
|
|
|
|
|
|
Strict scoring: %2d%% recall, %2d%% precision |
18
|
|
|
|
|
|
|
Relaxed scoring: %2d%% recall, %2d%% precision |
19
|
|
|
|
|
|
|
V. relaxed scoring: %2d%% recall, %2d%% precision |
20
|
|
|
|
|
|
|
", calc_stats(evaluate_segmenter($tiling_segmenter,20,$input)); |
21
|
|
|
|
|
|
|
} |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
=head1 DESCRIPTION |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
See synopsis. |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
Also check out segmenter.pl in the eg directory. |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
=head1 BUGS |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
This module only works correctly when the segmenter has a MIN_SEGMENT_SIZE >= 2. |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
=head1 AUTHORS |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
David James |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
=head1 SEE ALSO |
38
|
|
|
|
|
|
|
|
39
|
|
|
|
|
|
|
L, L, |
40
|
|
|
|
|
|
|
L |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
=cut |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
$VERSION = 0.10; |
46
|
|
|
|
|
|
|
@EXPORT_OK = qw(evaluate_segmenter calc_stats); |
47
|
1
|
|
|
1
|
|
2262
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
37
|
|
48
|
1
|
|
|
1
|
|
6
|
use base 'Class::Exporter'; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
84
|
|
49
|
1
|
|
|
1
|
|
6
|
use Math::HashSum qw(hashsum); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
1011
|
|
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
# Create a new Evaluator object |
52
|
|
|
|
|
|
|
sub new { |
53
|
1
|
|
|
1
|
0
|
60
|
my $self = shift; |
54
|
1
|
|
|
|
|
7
|
bless { |
55
|
|
|
|
|
|
|
@_ |
56
|
|
|
|
|
|
|
}, $self |
57
|
|
|
|
|
|
|
} |
58
|
|
|
|
|
|
|
|
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
# Evaluate the segmenter on a particular input |
61
|
|
|
|
|
|
|
sub evaluate_segmenter { |
62
|
0
|
|
|
0
|
0
|
|
my ($self, $segmenter, $input, $num_segments) = @_; |
63
|
|
|
|
|
|
|
|
64
|
0
|
|
|
|
|
|
$self->{taken} = {}; |
65
|
|
|
|
|
|
|
|
66
|
0
|
|
|
|
|
|
my $num_paragraphs = @{$segmenter->{splitter}->paragraph_breaks($input)}; |
|
0
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
|
68
|
0
|
|
|
|
|
|
my $break = $self->{break} = $segmenter->{splitter}->segment_breaks($input); |
69
|
0
|
|
0
|
|
|
|
$num_segments ||= scalar keys %{$break}; |
|
0
|
|
|
|
|
|
|
70
|
0
|
|
|
|
|
|
my $assigned = $self->{assigned} = $segmenter->segment($num_segments, $input); |
71
|
|
|
|
|
|
|
|
72
|
0
|
|
0
|
|
|
|
my @description = map { { |
|
0
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
para=>$_, |
74
|
|
|
|
|
|
|
true=>exists $break->{$_}, |
75
|
|
|
|
|
|
|
label=>$assigned->{$_}, |
76
|
|
|
|
|
|
|
strict=>exists $break->{$_} && exists $assigned->{$_}, |
77
|
|
|
|
|
|
|
relaxed=>$self->relaxed_weight($_), |
78
|
|
|
|
|
|
|
very_relaxed=>$self->very_relaxed_weight($_), |
79
|
|
|
|
|
|
|
} } (0..$num_paragraphs-1); |
80
|
|
|
|
|
|
|
|
81
|
0
|
|
|
|
|
|
return @description; |
82
|
|
|
|
|
|
|
} |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
# Get the weight of a particular index based on a relaxed scheme |
85
|
|
|
|
|
|
|
# NOTE: Assumes that MIN_SEGMENT_SIZE >= 2 |
86
|
|
|
|
|
|
|
sub relaxed_weight { |
87
|
0
|
|
|
0
|
0
|
|
my ($self, $i) = @_; |
88
|
0
|
|
|
|
|
|
my $assigned = $self->{assigned}{$i}; |
89
|
0
|
|
|
|
|
|
my $break = $self->{break}{$i}; |
90
|
0
|
0
|
0
|
|
|
|
if ($assigned and $break) { |
91
|
0
|
|
|
|
|
|
$self->take(1,"break",$i); |
92
|
0
|
|
|
|
|
|
$self->take(1,"assigned",$i); |
93
|
0
|
|
|
|
|
|
return 1; |
94
|
|
|
|
|
|
|
} |
95
|
0
|
0
|
|
|
|
|
if (defined $assigned) { |
|
|
0
|
|
|
|
|
|
96
|
0
|
0
|
0
|
|
|
|
if ($assigned =~ /L/ and $self->take(1,"break",$i-1) or |
|
|
0
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
97
|
|
|
|
|
|
|
$assigned =~ /R/ and $self->take(1,"break",$i+1)) { |
98
|
0
|
|
|
|
|
|
return 0.8; |
99
|
|
|
|
|
|
|
} elsif ($self->take(1,"break",$i-1) or $self->take(1,"break",$i+1)) { |
100
|
0
|
|
|
|
|
|
return 0.4; |
101
|
|
|
|
|
|
|
} |
102
|
|
|
|
|
|
|
} elsif (exists $self->{break}{$i}) { |
103
|
0
|
0
|
0
|
|
|
|
if ($self->take(1,"assigned",$i-1,"R") or |
|
|
0
|
0
|
|
|
|
|
104
|
|
|
|
|
|
|
$self->take(1,"assigned",$i+1,"L")) { |
105
|
0
|
|
|
|
|
|
return 0.8; |
106
|
|
|
|
|
|
|
} elsif ($self->take(1,"assigned",$i-1) or |
107
|
|
|
|
|
|
|
$self->take(1,"assigned",$i+1)) { |
108
|
|
|
|
|
|
|
|
109
|
0
|
|
|
|
|
|
return 0.4; |
110
|
|
|
|
|
|
|
} |
111
|
|
|
|
|
|
|
} |
112
|
0
|
|
|
|
|
|
return 0; |
113
|
|
|
|
|
|
|
} |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
# Get the weight of a particular index based on a very relaxed scheme |
116
|
|
|
|
|
|
|
# NOTE: Assumes that MIN_SEGMENT_SIZE >= 2 |
117
|
|
|
|
|
|
|
sub very_relaxed_weight { |
118
|
0
|
|
|
0
|
0
|
|
my ($self, $i) = @_; |
119
|
0
|
|
|
|
|
|
my $assigned = $self->{assigned}{$i}; |
120
|
0
|
|
|
|
|
|
my $break = $self->{break}{$i}; |
121
|
|
|
|
|
|
|
|
122
|
0
|
0
|
0
|
|
|
|
if ($assigned or $break) { |
123
|
0
|
|
|
|
|
|
foreach (-2..2) { |
124
|
0
|
|
0
|
|
|
|
$assigned ||= $self->take(2,"assigned",$i+$_); |
125
|
0
|
|
0
|
|
|
|
$break ||= $self->take(2,"break",$i+$_); |
126
|
|
|
|
|
|
|
} |
127
|
|
|
|
|
|
|
} |
128
|
0
|
|
0
|
|
|
|
return ($assigned and $break); |
129
|
|
|
|
|
|
|
} |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
# Mark a particular index as used if it's not already used |
132
|
|
|
|
|
|
|
sub take { |
133
|
0
|
|
|
0
|
0
|
|
my ($self,$count,$which,$i,$req) = @_; |
134
|
0
|
0
|
0
|
|
|
|
if (!$self->{taken}{$count}{$which}{$i} and $self->{$which}{$i}) { |
135
|
0
|
0
|
0
|
|
|
|
if (!$req or $self->{$which}{$i} =~ /$req/) { |
136
|
0
|
|
|
|
|
|
$self->{taken}{$count}{$which}{$i}++; |
137
|
0
|
|
|
|
|
|
return 1; |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
} |
140
|
|
|
|
|
|
|
} |
141
|
0
|
|
|
|
|
|
return; |
142
|
|
|
|
|
|
|
} |
143
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
# Calculate precision and recall for strict, relaxed, very_relaxed |
145
|
|
|
|
|
|
|
sub calc_stats { |
146
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
147
|
0
|
|
|
|
|
|
my %sum = hashsum map { %$_ } @_; |
|
0
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
# Ensure "R" and "L" count as categories |
150
|
0
|
|
|
|
|
|
$sum{label} = grep { $_->{label} } @_; |
|
0
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
# Ensure relaxed counts don't double-count |
153
|
0
|
|
|
|
|
|
$sum{relaxed} -= ($sum{relaxed} - $sum{strict})/2; |
154
|
0
|
|
|
|
|
|
$sum{very_relaxed} -= ($sum{very_relaxed} - $sum{strict})/2; |
155
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
# Sanity checks |
157
|
0
|
0
|
|
|
|
|
if ($sum{true} == 0) { |
|
|
0
|
|
|
|
|
|
158
|
0
|
|
|
|
|
|
die "No segment_breaks found. Please label the true segments in the original text so that we can evaluate the performance of the Segmenting algorithm"; |
159
|
|
|
|
|
|
|
} elsif ($sum{label} == 0) { |
160
|
0
|
|
|
|
|
|
die "No segments labelled by Segmenting algorithm"; |
161
|
|
|
|
|
|
|
} |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
# Return results |
164
|
0
|
|
|
|
|
|
return map { 100*$sum{$_}/$sum{true}, 100*$sum{$_} / $sum{label} } |
|
0
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
qw(strict relaxed very_relaxed); |
166
|
|
|
|
|
|
|
} |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
1; |
169
|
|
|
|
|
|
|
|