line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package WordNet::SenseRelate::AllWords; |
2
|
|
|
|
|
|
|
|
3
|
|
|
|
|
|
|
# $Id: AllWords.pm,v 1.40 2009/05/27 20:58:27 kvarada Exp $ |
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
=head1 NAME |
6
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
WordNet::SenseRelate::AllWords - Disambiguate All Words in a Text based on semantic similarity and relatedness in WordNet |
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
=head1 SYNOPSIS |
10
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
use WordNet::SenseRelate::AllWords; |
12
|
|
|
|
|
|
|
use WordNet::QueryData; |
13
|
|
|
|
|
|
|
use WordNet::Tools; |
14
|
|
|
|
|
|
|
my $qd = WordNet::QueryData->new; |
15
|
|
|
|
|
|
|
defined $qd or die "Construction of WordNet::QueryData failed"; |
16
|
|
|
|
|
|
|
my $wntools = WordNet::Tools->new($qd); |
17
|
|
|
|
|
|
|
defined $wntools or die "\nCouldn't construct WordNet::Tools object"; |
18
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
my $wsd = WordNet::SenseRelate::AllWords->new (wordnet => $qd, |
20
|
|
|
|
|
|
|
wntools => $wntools, |
21
|
|
|
|
|
|
|
measure => 'WordNet::Similarity::lesk'); |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
my @context = qw/the bridge is held up by red tape/; |
24
|
|
|
|
|
|
|
my @results = $wsd->disambiguate (window => 3, |
25
|
|
|
|
|
|
|
context => [@context]); |
26
|
|
|
|
|
|
|
print "@results\n"; |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
=head1 DESCRIPTION |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
WordNet::SenseRelate::AllWords implements an algorithm for Word Sense |
31
|
|
|
|
|
|
|
Disambiguation that uses measures of semantic relatedness. The algorithm |
32
|
|
|
|
|
|
|
is an extension of an algorithm described by Pedersen, Banerjee, and |
33
|
|
|
|
|
|
|
Patwardhan[1]. This implementation is similar to the original SenseRelate |
34
|
|
|
|
|
|
|
package but disambiguates every word in the given context rather than just |
35
|
|
|
|
|
|
|
single word. |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
=head2 Methods |
38
|
|
|
|
|
|
|
|
39
|
|
|
|
|
|
|
Note: the methods below will die() on serious errors. Wrap calls to the |
40
|
|
|
|
|
|
|
methods in an eval BLOCK to catch the exceptions. See |
41
|
|
|
|
|
|
|
'perldoc -f eval' for more information. |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
Example: |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
my @res; |
46
|
|
|
|
|
|
|
eval {@res = $wsd->disambiguate (args...)} |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
if ($@){ |
49
|
|
|
|
|
|
|
print STDERR "An exception occurred ($@)\n"; |
50
|
|
|
|
|
|
|
} |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
=over |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
=cut |
55
|
|
|
|
|
|
|
|
56
|
3
|
|
|
3
|
|
3233
|
use 5.006; |
|
3
|
|
|
|
|
12
|
|
|
3
|
|
|
|
|
140
|
|
57
|
3
|
|
|
3
|
|
18
|
use strict; |
|
3
|
|
|
|
|
4
|
|
|
3
|
|
|
|
|
117
|
|
58
|
3
|
|
|
3
|
|
16
|
use warnings; |
|
3
|
|
|
|
|
17
|
|
|
3
|
|
|
|
|
111
|
|
59
|
3
|
|
|
3
|
|
167
|
use Carp; |
|
3
|
|
|
|
|
6
|
|
|
3
|
|
|
|
|
617
|
|
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
our @ISA = (); |
62
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
our $VERSION = '0.19'; |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
my %wordnet; |
66
|
|
|
|
|
|
|
my %wntools; |
67
|
|
|
|
|
|
|
my %simMeasure; # the similarity/relatedness measure |
68
|
|
|
|
|
|
|
my %stoplist; |
69
|
|
|
|
|
|
|
my %pairScore; |
70
|
|
|
|
|
|
|
my %contextScore; |
71
|
|
|
|
|
|
|
my %trace; |
72
|
|
|
|
|
|
|
my %outfile; |
73
|
|
|
|
|
|
|
my %forcepos; |
74
|
|
|
|
|
|
|
my %nocompoundify; |
75
|
|
|
|
|
|
|
my %usemono; |
76
|
|
|
|
|
|
|
my %backoff; |
77
|
|
|
|
|
|
|
my %wnformat; |
78
|
|
|
|
|
|
|
my %fixed; |
79
|
|
|
|
|
|
|
|
80
|
|
|
|
|
|
|
# closed class words |
81
|
3
|
|
|
|
|
275
|
use constant {CLOSED => 'c', |
82
|
3
|
|
|
3
|
|
27
|
NOINFO => 'f'}; |
|
3
|
|
|
|
|
6
|
|
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
# constants used to specify trace levels |
85
|
3
|
|
|
3
|
|
15
|
use constant TR_CONTEXT => 1; # show the context window |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
164
|
|
86
|
3
|
|
|
3
|
|
15
|
use constant TR_BESTSCORE => 2; # show the best score |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
149
|
|
87
|
3
|
|
|
3
|
|
13
|
use constant TR_ALLSCORES => 4; # show all non-zero scores |
|
3
|
|
|
|
|
7
|
|
|
3
|
|
|
|
|
135
|
|
88
|
3
|
|
|
3
|
|
14
|
use constant TR_PAIRWISE => 8; # show all the non-zero similarity scores |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
139
|
|
89
|
3
|
|
|
3
|
|
30
|
use constant TR_ZERO => 16; |
|
3
|
|
|
|
|
6
|
|
|
3
|
|
|
|
|
120
|
|
90
|
3
|
|
|
3
|
|
14
|
use constant TR_MEASURE => 32; # show similarity measure traces |
|
3
|
|
|
|
|
5
|
|
|
3
|
|
|
|
|
69206
|
|
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
# Penn tagset |
93
|
|
|
|
|
|
|
my %wnTag = ( |
94
|
|
|
|
|
|
|
JJ => 'a', |
95
|
|
|
|
|
|
|
JJR => 'a', |
96
|
|
|
|
|
|
|
JJS => 'a', |
97
|
|
|
|
|
|
|
CD => 'a', |
98
|
|
|
|
|
|
|
RB => 'r', |
99
|
|
|
|
|
|
|
RBR => 'r', |
100
|
|
|
|
|
|
|
RBS => 'r', |
101
|
|
|
|
|
|
|
RP => 'r', |
102
|
|
|
|
|
|
|
WRB => CLOSED, |
103
|
|
|
|
|
|
|
CC => CLOSED, |
104
|
|
|
|
|
|
|
IN => 'r', |
105
|
|
|
|
|
|
|
DT => CLOSED, |
106
|
|
|
|
|
|
|
PDT => CLOSED, |
107
|
|
|
|
|
|
|
CC => CLOSED, |
108
|
|
|
|
|
|
|
'PRP$' => CLOSED, |
109
|
|
|
|
|
|
|
PRP => CLOSED, |
110
|
|
|
|
|
|
|
WDT => CLOSED, |
111
|
|
|
|
|
|
|
'WP$' => CLOSED, |
112
|
|
|
|
|
|
|
NN => 'n', |
113
|
|
|
|
|
|
|
NNS => 'n', |
114
|
|
|
|
|
|
|
NNP => 'n', |
115
|
|
|
|
|
|
|
NNPS => 'n', |
116
|
|
|
|
|
|
|
PRP => CLOSED, |
117
|
|
|
|
|
|
|
WP => CLOSED, |
118
|
|
|
|
|
|
|
EX => CLOSED, |
119
|
|
|
|
|
|
|
VBP => 'v', |
120
|
|
|
|
|
|
|
VB => 'v', |
121
|
|
|
|
|
|
|
VBD => 'v', |
122
|
|
|
|
|
|
|
VBG => 'v', |
123
|
|
|
|
|
|
|
VBN => 'v', |
124
|
|
|
|
|
|
|
VBZ => 'v', |
125
|
|
|
|
|
|
|
VBP => 'v', |
126
|
|
|
|
|
|
|
MD => 'v', |
127
|
|
|
|
|
|
|
TO => CLOSED, |
128
|
|
|
|
|
|
|
POS => undef, |
129
|
|
|
|
|
|
|
UH => CLOSED, |
130
|
|
|
|
|
|
|
'.' => undef, |
131
|
|
|
|
|
|
|
':' => undef, |
132
|
|
|
|
|
|
|
',' => undef, |
133
|
|
|
|
|
|
|
_ => undef, |
134
|
|
|
|
|
|
|
'$' => undef, |
135
|
|
|
|
|
|
|
'(' => undef, |
136
|
|
|
|
|
|
|
')' => undef, |
137
|
|
|
|
|
|
|
'"' => undef, |
138
|
|
|
|
|
|
|
FW => NOINFO, |
139
|
|
|
|
|
|
|
SYM => undef, |
140
|
|
|
|
|
|
|
LS => undef, |
141
|
|
|
|
|
|
|
); |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
=item BZ<> |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
Z<>The constructor for this class. It will create a new instance and |
146
|
|
|
|
|
|
|
return a reference to the constructed object. |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
Parameters: |
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
wordnet => REFERENCE : WordNet::QueryData object |
151
|
|
|
|
|
|
|
wntools => REFERENCE : WordNet::Tools object |
152
|
|
|
|
|
|
|
measure => STRING : name of a WordNet::Similarity measure |
153
|
|
|
|
|
|
|
config => FILENAME : config file for above measure |
154
|
|
|
|
|
|
|
outfile => FILENAME : name of a file for output (optional) |
155
|
|
|
|
|
|
|
stoplist => FILENAME : file containing list of stop words |
156
|
|
|
|
|
|
|
pairScore => INTEGER : minimum pairwise score (default: 0) |
157
|
|
|
|
|
|
|
contextScore => INTEGER : minimum overall score (default: 0) |
158
|
|
|
|
|
|
|
trace => INTEGER : generate traces (default: 0) |
159
|
|
|
|
|
|
|
forcepos => INTEGER : do part-of-speech coercion (default: 0) |
160
|
|
|
|
|
|
|
nocompoundify => INTEGER : disable compoundify (default: 0) |
161
|
|
|
|
|
|
|
usemono => INTEGER : enable assigning the available sense to usemono (default: 0) |
162
|
|
|
|
|
|
|
backoff => INTEGER : enable assigning most frequent sense if the measure can't assign sense (default: 0) |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
Returns: |
165
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
A reference to the constructed object. |
167
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
Example: |
169
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
WordNet::SenseRelate::AllWords->new (wordnet => $query_data_obj, |
171
|
|
|
|
|
|
|
wntools => $wordnet_tools_obj, |
172
|
|
|
|
|
|
|
measure => 'WordNet::Similarity::lesk', |
173
|
|
|
|
|
|
|
trace => 1); |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
The trace levels are: |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
1 Show the context window for each pass through the algorithm. |
178
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
2 Display winning score for each pass (i.e., for each target word). |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
4 Display the non-zero scores for each sense of each target |
182
|
|
|
|
|
|
|
word (overrides 2). |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
8 Display the non-zero values from the semantic relatedness measures. |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
16 Show the zero values as well when combined with either 4 or 8. |
187
|
|
|
|
|
|
|
When not used with 4 or 8, this has no effect. |
188
|
|
|
|
|
|
|
|
189
|
|
|
|
|
|
|
32 Display traces from the semantic relatedness module. |
190
|
|
|
|
|
|
|
|
191
|
|
|
|
|
|
|
These trace levels can be added together. For example, by specifying |
192
|
|
|
|
|
|
|
a trace level of 3, the context window will be displayed along with |
193
|
|
|
|
|
|
|
the winning score for each pass. |
194
|
|
|
|
|
|
|
|
195
|
|
|
|
|
|
|
=cut |
196
|
|
|
|
|
|
|
|
197
|
|
|
|
|
|
|
sub new |
198
|
|
|
|
|
|
|
{ |
199
|
0
|
|
|
0
|
1
|
|
my $class = shift; |
200
|
0
|
|
|
|
|
|
my %args = @_; |
201
|
0
|
|
0
|
|
|
|
$class = ref $class || $class; |
202
|
|
|
|
|
|
|
|
203
|
0
|
|
|
|
|
|
my $qd; |
204
|
|
|
|
|
|
|
my $wnt; |
205
|
0
|
|
|
|
|
|
my $measure; |
206
|
0
|
|
|
|
|
|
my $measure_config; |
207
|
0
|
|
|
|
|
|
my $stoplist; |
208
|
0
|
|
|
|
|
|
my $pairScore = 0; |
209
|
0
|
|
|
|
|
|
my $contextScore = 0; |
210
|
0
|
|
|
|
|
|
my $trace; |
211
|
|
|
|
|
|
|
my $outfile; |
212
|
0
|
|
|
|
|
|
my $forcepos; |
213
|
0
|
|
|
|
|
|
my $nocompoundify=0; |
214
|
0
|
|
|
|
|
|
my $usemono=0; |
215
|
0
|
|
|
|
|
|
my $backoff=0; |
216
|
0
|
|
|
|
|
|
my $fixed = 0; |
217
|
0
|
|
|
|
|
|
my $wnformat = 0; |
218
|
|
|
|
|
|
|
|
219
|
0
|
|
|
|
|
|
while (my ($key, $val) = each %args) { |
220
|
0
|
0
|
|
|
|
|
if ($key eq 'wordnet') { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
221
|
0
|
|
|
|
|
|
$qd = $val; |
222
|
|
|
|
|
|
|
} |
223
|
|
|
|
|
|
|
elsif ($key eq 'wntools') |
224
|
|
|
|
|
|
|
{ |
225
|
0
|
|
|
|
|
|
$wnt = $val; |
226
|
|
|
|
|
|
|
} |
227
|
|
|
|
|
|
|
elsif ($key eq 'measure') { |
228
|
0
|
|
|
|
|
|
$measure = $val; |
229
|
|
|
|
|
|
|
} |
230
|
|
|
|
|
|
|
elsif ($key eq 'config') { |
231
|
0
|
|
|
|
|
|
$measure_config = $val; |
232
|
|
|
|
|
|
|
} |
233
|
|
|
|
|
|
|
elsif ($key eq 'stoplist') { |
234
|
0
|
|
|
|
|
|
$stoplist = $val; |
235
|
|
|
|
|
|
|
} |
236
|
|
|
|
|
|
|
elsif ($key eq 'pairScore') { |
237
|
0
|
|
|
|
|
|
$pairScore = $val; |
238
|
|
|
|
|
|
|
} |
239
|
|
|
|
|
|
|
elsif ($key eq 'contextScore') { |
240
|
0
|
|
|
|
|
|
$contextScore = $val; |
241
|
|
|
|
|
|
|
} |
242
|
|
|
|
|
|
|
elsif($key eq 'nocompoundify'){ |
243
|
0
|
|
|
|
|
|
$nocompoundify=$val; |
244
|
|
|
|
|
|
|
} |
245
|
|
|
|
|
|
|
elsif($key eq 'usemono'){ |
246
|
0
|
|
|
|
|
|
$usemono=$val; |
247
|
|
|
|
|
|
|
} |
248
|
|
|
|
|
|
|
elsif($key eq 'backoff'){ |
249
|
0
|
|
|
|
|
|
$backoff=$val; |
250
|
|
|
|
|
|
|
} |
251
|
|
|
|
|
|
|
elsif ($key eq 'trace') { |
252
|
0
|
|
|
|
|
|
$trace = $val; |
253
|
0
|
0
|
|
|
|
|
$trace = defined $trace ? $trace : 0; |
254
|
|
|
|
|
|
|
} |
255
|
|
|
|
|
|
|
elsif ($key eq 'outfile') { |
256
|
0
|
|
|
|
|
|
$outfile = $val; |
257
|
|
|
|
|
|
|
} |
258
|
|
|
|
|
|
|
elsif ($key eq 'forcepos') { |
259
|
0
|
|
|
|
|
|
$forcepos = $val; |
260
|
|
|
|
|
|
|
} |
261
|
|
|
|
|
|
|
elsif ($key eq 'fixed') { |
262
|
0
|
|
|
|
|
|
$fixed = $val; |
263
|
|
|
|
|
|
|
} |
264
|
|
|
|
|
|
|
elsif ($key eq 'wnformat') { |
265
|
0
|
|
|
|
|
|
$wnformat = $val; |
266
|
|
|
|
|
|
|
} |
267
|
|
|
|
|
|
|
else { |
268
|
0
|
|
|
|
|
|
croak "Unknown parameter type '$key'"; |
269
|
|
|
|
|
|
|
} |
270
|
|
|
|
|
|
|
} |
271
|
|
|
|
|
|
|
|
272
|
0
|
0
|
|
|
|
|
unless (ref $qd) { |
273
|
0
|
|
|
|
|
|
croak "No WordNet::QueryData object supplied"; |
274
|
|
|
|
|
|
|
} |
275
|
|
|
|
|
|
|
|
276
|
0
|
0
|
|
|
|
|
unless (ref $wnt) { |
277
|
0
|
|
|
|
|
|
croak "No WordNet::Tools object supplied"; |
278
|
|
|
|
|
|
|
} |
279
|
|
|
|
|
|
|
|
280
|
0
|
0
|
|
|
|
|
unless ($measure) { |
281
|
0
|
|
|
|
|
|
croak "No relatedness measure supplied"; |
282
|
|
|
|
|
|
|
} |
283
|
|
|
|
|
|
|
|
284
|
0
|
|
|
|
|
|
my $self = bless [], $class; |
285
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
# initialize tracing; |
287
|
0
|
0
|
|
|
|
|
if (defined $trace) { |
288
|
0
|
|
|
|
|
|
$trace{$self} = {level => $trace, string => ''}; |
289
|
|
|
|
|
|
|
} |
290
|
|
|
|
|
|
|
else { |
291
|
0
|
|
|
|
|
|
$trace{$self} = {level => 0, string => ''}; |
292
|
|
|
|
|
|
|
} |
293
|
|
|
|
|
|
|
|
294
|
|
|
|
|
|
|
# require the relatedness modules |
295
|
0
|
|
|
|
|
|
my $file = $measure; |
296
|
0
|
|
|
|
|
|
$file =~ s/::/\//g; |
297
|
0
|
|
|
|
|
|
require "${file}.pm"; |
298
|
|
|
|
|
|
|
|
299
|
|
|
|
|
|
|
# construct the relatedness object |
300
|
0
|
0
|
|
|
|
|
if (defined $measure_config) { |
301
|
0
|
|
|
|
|
|
$simMeasure{$self} = $measure->new ($qd, $measure_config); |
302
|
|
|
|
|
|
|
} |
303
|
|
|
|
|
|
|
else { |
304
|
0
|
|
|
|
|
|
$simMeasure{$self} = $measure->new ($qd); |
305
|
|
|
|
|
|
|
} |
306
|
|
|
|
|
|
|
|
307
|
|
|
|
|
|
|
# check for errors |
308
|
0
|
|
|
|
|
|
my ($errCode, $errStr) = $simMeasure{$self}->getError; |
309
|
0
|
0
|
|
|
|
|
if ($errCode) { |
310
|
0
|
|
|
|
|
|
carp $errStr; |
311
|
|
|
|
|
|
|
} |
312
|
|
|
|
|
|
|
|
313
|
|
|
|
|
|
|
# turn on traces in the relatedness measure if required |
314
|
0
|
0
|
|
|
|
|
if ($trace{$self}->{level} & TR_MEASURE) { |
315
|
0
|
|
|
|
|
|
$simMeasure{$self}->{trace} = 1; |
316
|
|
|
|
|
|
|
} |
317
|
|
|
|
|
|
|
else { |
318
|
0
|
|
|
|
|
|
$simMeasure{$self}->{trace} = 0; |
319
|
|
|
|
|
|
|
} |
320
|
|
|
|
|
|
|
|
321
|
|
|
|
|
|
|
|
322
|
|
|
|
|
|
|
# save ref to WordNet::QueryData obj |
323
|
0
|
|
|
|
|
|
$wordnet{$self} = $qd; |
324
|
|
|
|
|
|
|
|
325
|
|
|
|
|
|
|
# save ref to WordNet::Tools obj |
326
|
0
|
|
|
|
|
|
$wntools{$self} = $wnt; |
327
|
|
|
|
|
|
|
|
328
|
0
|
0
|
|
|
|
|
$self->_loadStoplist ($stoplist) if defined $stoplist; |
329
|
|
|
|
|
|
|
|
330
|
|
|
|
|
|
|
# store threshold values |
331
|
0
|
|
|
|
|
|
$pairScore{$self} = $pairScore; |
332
|
0
|
|
|
|
|
|
$contextScore{$self} = $contextScore; |
333
|
|
|
|
|
|
|
|
334
|
|
|
|
|
|
|
# save output file name |
335
|
0
|
|
|
|
|
|
$outfile{$self} = $outfile; |
336
|
0
|
0
|
0
|
|
|
|
if ($outfile and -e $outfile) { |
337
|
0
|
|
|
|
|
|
unlink $outfile; |
338
|
|
|
|
|
|
|
} |
339
|
|
|
|
|
|
|
|
340
|
0
|
0
|
|
|
|
|
if (defined $forcepos) { |
341
|
0
|
|
|
|
|
|
$forcepos{$self} = $forcepos; |
342
|
|
|
|
|
|
|
} |
343
|
|
|
|
|
|
|
else { |
344
|
0
|
|
|
|
|
|
$forcepos{$self} = 0; |
345
|
|
|
|
|
|
|
} |
346
|
|
|
|
|
|
|
|
347
|
0
|
0
|
|
|
|
|
if (defined $nocompoundify) { |
348
|
0
|
|
|
|
|
|
$nocompoundify{$self} = $nocompoundify; |
349
|
|
|
|
|
|
|
} |
350
|
|
|
|
|
|
|
else { |
351
|
0
|
|
|
|
|
|
$nocompoundify{$self} = 0; |
352
|
|
|
|
|
|
|
} |
353
|
|
|
|
|
|
|
|
354
|
0
|
0
|
|
|
|
|
if (defined $usemono) { |
355
|
0
|
|
|
|
|
|
$usemono{$self} = $usemono; |
356
|
|
|
|
|
|
|
} |
357
|
|
|
|
|
|
|
else { |
358
|
0
|
|
|
|
|
|
$usemono{$self} = 0; |
359
|
|
|
|
|
|
|
} |
360
|
|
|
|
|
|
|
|
361
|
0
|
0
|
|
|
|
|
if (defined $backoff) { |
362
|
0
|
|
|
|
|
|
$backoff{$self} = $backoff; |
363
|
|
|
|
|
|
|
} |
364
|
|
|
|
|
|
|
else { |
365
|
0
|
|
|
|
|
|
$backoff{$self} = 0; |
366
|
|
|
|
|
|
|
} |
367
|
|
|
|
|
|
|
|
368
|
|
|
|
|
|
|
|
369
|
|
|
|
|
|
|
|
370
|
|
|
|
|
|
|
|
371
|
0
|
|
|
|
|
|
$fixed{$self} = $fixed; |
372
|
|
|
|
|
|
|
|
373
|
0
|
|
|
|
|
|
$wnformat{$self} = $wnformat; |
374
|
|
|
|
|
|
|
|
375
|
0
|
|
|
|
|
|
return $self; |
376
|
|
|
|
|
|
|
} |
377
|
|
|
|
|
|
|
|
378
|
|
|
|
|
|
|
# the destructor for this class. You shouldn't need to call this |
379
|
|
|
|
|
|
|
# explicitly (but if you really want to, you can see what happens) |
380
|
|
|
|
|
|
|
sub DESTROY |
381
|
|
|
|
|
|
|
{ |
382
|
0
|
|
|
0
|
|
|
my $self = shift; |
383
|
0
|
|
|
|
|
|
delete $wordnet{$self}; |
384
|
0
|
|
|
|
|
|
delete $wntools{$self}; |
385
|
0
|
|
|
|
|
|
delete $simMeasure{$self}; |
386
|
0
|
|
|
|
|
|
delete $stoplist{$self}; |
387
|
0
|
|
|
|
|
|
delete $pairScore{$self}; |
388
|
0
|
|
|
|
|
|
delete $contextScore{$self}; |
389
|
0
|
|
|
|
|
|
delete $trace{$self}; |
390
|
0
|
|
|
|
|
|
delete $outfile{$self}; |
391
|
0
|
|
|
|
|
|
delete $forcepos{$self}; |
392
|
0
|
|
|
|
|
|
delete $nocompoundify{$self}; |
393
|
0
|
|
|
|
|
|
delete $usemono{$self}; |
394
|
0
|
|
|
|
|
|
delete $backoff{$self}; |
395
|
0
|
|
|
|
|
|
delete $wnformat{$self}; |
396
|
0
|
|
|
|
|
|
delete $fixed{$self}; |
397
|
|
|
|
|
|
|
|
398
|
0
|
|
|
|
|
|
1; |
399
|
|
|
|
|
|
|
} |
400
|
|
|
|
|
|
|
|
401
|
|
|
|
|
|
|
sub wordnet : lvalue |
402
|
|
|
|
|
|
|
{ |
403
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
404
|
0
|
|
|
|
|
|
$wordnet{$self}; |
405
|
|
|
|
|
|
|
} |
406
|
|
|
|
|
|
|
|
407
|
|
|
|
|
|
|
=item B |
408
|
|
|
|
|
|
|
|
409
|
|
|
|
|
|
|
Disambiguates all the words in the specified context and returns them |
410
|
|
|
|
|
|
|
as a list. If a word cannot be disambiguated, then it is returned "as is". |
411
|
|
|
|
|
|
|
A word cannot be disambiguated if it is not in WordNet or if no value |
412
|
|
|
|
|
|
|
exceeds the specified threshold. |
413
|
|
|
|
|
|
|
|
414
|
|
|
|
|
|
|
The context parameter specifies the |
415
|
|
|
|
|
|
|
words to be disambiguated. It treats the value as one sentence. To |
416
|
|
|
|
|
|
|
disambiguate a document with multiple sentences, make one call to |
417
|
|
|
|
|
|
|
disambiguate() for each sentence. |
418
|
|
|
|
|
|
|
|
419
|
|
|
|
|
|
|
Parameters: |
420
|
|
|
|
|
|
|
|
421
|
|
|
|
|
|
|
window => INTEGER : the window size to use. A window size of N means |
422
|
|
|
|
|
|
|
that the window will include N words, including |
423
|
|
|
|
|
|
|
the target word. If N is an even number, there |
424
|
|
|
|
|
|
|
will be one more word on the left side of the |
425
|
|
|
|
|
|
|
target word than on the right. |
426
|
|
|
|
|
|
|
tagged => BOOLEAN : true if the text is tagged, false otherwise |
427
|
|
|
|
|
|
|
scheme => normal|sense1|random|fixed : the disambiguation scheme to use |
428
|
|
|
|
|
|
|
context => ARRAY_REF : reference to an array of words to disambiguate |
429
|
|
|
|
|
|
|
|
430
|
|
|
|
|
|
|
Returns: An array of disambiguated words. |
431
|
|
|
|
|
|
|
|
432
|
|
|
|
|
|
|
Example: |
433
|
|
|
|
|
|
|
|
434
|
|
|
|
|
|
|
my @results = |
435
|
|
|
|
|
|
|
$wsd->disambiguate (window => 3, tagged => 0, context => [@words]); |
436
|
|
|
|
|
|
|
|
437
|
|
|
|
|
|
|
Rules for attaching suffixes: |
438
|
|
|
|
|
|
|
|
439
|
|
|
|
|
|
|
Suffixes are attached to the words in the context in order to ignore those while disambiguation. |
440
|
|
|
|
|
|
|
Note that after converting the tags to WordNet tags, tagged text is treated same as wntagged text. |
441
|
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
Below is the ordered enumeration of the words which are ignored for disambiguation and the suffixes attached to those words. |
443
|
|
|
|
|
|
|
|
444
|
|
|
|
|
|
|
Note that we check for such words in the order below: |
445
|
|
|
|
|
|
|
|
446
|
|
|
|
|
|
|
1 stopwords => #o |
447
|
|
|
|
|
|
|
|
448
|
|
|
|
|
|
|
2 Only for tagged text : |
449
|
|
|
|
|
|
|
|
450
|
|
|
|
|
|
|
i) Closed Class words => #CL |
451
|
|
|
|
|
|
|
|
452
|
|
|
|
|
|
|
ii) Invalid Tag => #IT |
453
|
|
|
|
|
|
|
|
454
|
|
|
|
|
|
|
iii) Missing Word => #MW |
455
|
|
|
|
|
|
|
|
456
|
|
|
|
|
|
|
3 For tagged and wntagged text: |
457
|
|
|
|
|
|
|
|
458
|
|
|
|
|
|
|
i) No Tag => #NT |
459
|
|
|
|
|
|
|
|
460
|
|
|
|
|
|
|
ii) Missing Word => #MW |
461
|
|
|
|
|
|
|
|
462
|
|
|
|
|
|
|
iii) Invalid Tag => #IT |
463
|
|
|
|
|
|
|
|
464
|
|
|
|
|
|
|
4 Not in WordNet => #ND |
465
|
|
|
|
|
|
|
|
466
|
|
|
|
|
|
|
5 No Relatedness found with the surrounding words => #NR |
467
|
|
|
|
|
|
|
|
468
|
|
|
|
|
|
|
=cut |
469
|
|
|
|
|
|
|
|
470
|
|
|
|
|
|
|
#The scheme can have three different values: |
471
|
|
|
|
|
|
|
# |
472
|
|
|
|
|
|
|
#=over |
473
|
|
|
|
|
|
|
# |
474
|
|
|
|
|
|
|
#=item normal |
475
|
|
|
|
|
|
|
# |
476
|
|
|
|
|
|
|
#This is the normal mode of operation, where disambiguation is done by |
477
|
|
|
|
|
|
|
#measuring the semantic relatedness of the senses of each word with the |
478
|
|
|
|
|
|
|
#surrounding words. |
479
|
|
|
|
|
|
|
# |
480
|
|
|
|
|
|
|
#=item sense1 |
481
|
|
|
|
|
|
|
# |
482
|
|
|
|
|
|
|
#In this mode, the first sense number (i.e., sense number 1) is assigned |
483
|
|
|
|
|
|
|
#to each word. In WordNet, the first sense of a word is I the |
484
|
|
|
|
|
|
|
#most frequent sense. |
485
|
|
|
|
|
|
|
# |
486
|
|
|
|
|
|
|
#=item random |
487
|
|
|
|
|
|
|
# |
488
|
|
|
|
|
|
|
#In this mode, sense numbers are randomly assigned to each word from the |
489
|
|
|
|
|
|
|
#set of valid sense numbers for each word. For example, the noun 'hart' |
490
|
|
|
|
|
|
|
#has three senses in WordNet 2.0, so the word would randomly be assigned |
491
|
|
|
|
|
|
|
#1, 2, or 3. This may be useful for comparison purposes when evaluating |
492
|
|
|
|
|
|
|
#experimental results. |
493
|
|
|
|
|
|
|
# |
494
|
|
|
|
|
|
|
# |
495
|
|
|
|
|
|
|
# |
496
|
|
|
|
|
|
|
# |
497
|
|
|
|
|
|
|
#=cut |
498
|
|
|
|
|
|
|
|
499
|
|
|
|
|
|
|
sub disambiguate |
500
|
|
|
|
|
|
|
{ |
501
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
502
|
0
|
|
|
|
|
|
my %options = @_; |
503
|
0
|
|
|
|
|
|
my $contextScore; |
504
|
|
|
|
|
|
|
my $pairScore; |
505
|
0
|
|
|
|
|
|
my $window = 3; # default the window to 3 to avoid failure if omitted |
506
|
0
|
|
|
|
|
|
my $tagged; |
507
|
|
|
|
|
|
|
my @context; |
508
|
0
|
|
|
|
|
|
my $scheme = 'normal'; |
509
|
|
|
|
|
|
|
|
510
|
0
|
|
|
|
|
|
while (my ($key, $value) = each %options){ |
511
|
0
|
0
|
|
|
|
|
if ($key eq 'window') { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
512
|
0
|
|
|
|
|
|
$window = $value; |
513
|
|
|
|
|
|
|
} |
514
|
|
|
|
|
|
|
elsif ($key eq 'tagged') { |
515
|
0
|
|
|
|
|
|
$tagged = $value; |
516
|
|
|
|
|
|
|
} |
517
|
|
|
|
|
|
|
elsif ($key eq 'context') { |
518
|
0
|
|
|
|
|
|
@context = @$value; |
519
|
|
|
|
|
|
|
} |
520
|
|
|
|
|
|
|
elsif ($key eq 'scheme') { |
521
|
0
|
|
|
|
|
|
$scheme = $value; |
522
|
|
|
|
|
|
|
} |
523
|
|
|
|
|
|
|
else { |
524
|
0
|
|
|
|
|
|
croak "Unknown option '$key'"; |
525
|
|
|
|
|
|
|
} |
526
|
|
|
|
|
|
|
} |
527
|
|
|
|
|
|
|
|
528
|
|
|
|
|
|
|
# _initializeContext method |
529
|
|
|
|
|
|
|
# 1) compoundifies the text |
530
|
|
|
|
|
|
|
# 2) checks if the word is a stopword. If it is a stopword, attaches \#o |
531
|
|
|
|
|
|
|
# 3) converts position tags if we have tagged text |
532
|
0
|
|
|
|
|
|
my @newcontext = $self->_initializeContext ($tagged, @context); |
533
|
|
|
|
|
|
|
|
534
|
0
|
0
|
0
|
|
|
|
if($tagged || $wnformat{$self}){ |
535
|
0
|
|
|
|
|
|
foreach my $word (@newcontext) { |
536
|
0
|
0
|
0
|
|
|
|
if ($word !~ /\#/) { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
537
|
0
|
|
|
|
|
|
$word = $word . "#NT"; |
538
|
|
|
|
|
|
|
} |
539
|
|
|
|
|
|
|
elsif( $word =~ /^#/) |
540
|
|
|
|
|
|
|
{ |
541
|
0
|
|
|
|
|
|
$word = $word . "#MW"; |
542
|
|
|
|
|
|
|
} |
543
|
|
|
|
|
|
|
elsif ( $word !~ /\#[nvar]$/ && $word !~ /\#o\b/ && $word !~ /\#CL\b/ && $word !~ /\#IT\b/) { |
544
|
0
|
|
|
|
|
|
$word = $word . "#IT"; |
545
|
|
|
|
|
|
|
} |
546
|
|
|
|
|
|
|
} |
547
|
|
|
|
|
|
|
} |
548
|
|
|
|
|
|
|
|
549
|
0
|
|
|
|
|
|
my @results; |
550
|
0
|
0
|
0
|
|
|
|
if ($scheme eq 'sense1') { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
551
|
0
|
|
|
|
|
|
@results = $self->doSense1 (@newcontext); |
552
|
|
|
|
|
|
|
} |
553
|
|
|
|
|
|
|
elsif ($scheme eq 'random') { |
554
|
0
|
|
|
|
|
|
@results = $self->doRandom (@newcontext); |
555
|
|
|
|
|
|
|
} |
556
|
|
|
|
|
|
|
elsif (($scheme eq 'normal') or ($scheme eq 'fixed')) { |
557
|
0
|
0
|
|
|
|
|
$fixed{$self} = 1 if $scheme eq 'fixed'; |
558
|
0
|
|
|
|
|
|
@results = $self->doNormal ($pairScore, $contextScore, $window, |
559
|
|
|
|
|
|
|
@newcontext); |
560
|
|
|
|
|
|
|
} |
561
|
|
|
|
|
|
|
else { |
562
|
0
|
|
|
|
|
|
croak ("Bad scheme '$scheme'.\n", |
563
|
|
|
|
|
|
|
"Scheme must be 'normal', 'sense1', 'random', or 'fixed'"); |
564
|
|
|
|
|
|
|
} |
565
|
|
|
|
|
|
|
|
566
|
|
|
|
|
|
|
# my @rval = map {s/\#o//; $_} @results; |
567
|
0
|
|
|
|
|
|
my @rval = @results; |
568
|
|
|
|
|
|
|
|
569
|
0
|
0
|
|
|
|
|
if ($outfile{$self}) { |
570
|
0
|
0
|
|
|
|
|
open OFH, '>>', $outfile{$self} or croak "Cannot open outfile: $!"; |
571
|
0
|
|
|
|
|
|
print OFH "\n\n"; |
572
|
0
|
|
|
|
|
|
print OFH "Results after disambiguation...\n"; |
573
|
0
|
|
|
|
|
|
for my $i (0..$#newcontext) { |
574
|
0
|
|
|
|
|
|
my $orig_word = $newcontext[$i]; |
575
|
0
|
|
|
|
|
|
my $new_word = $rval[$i]; |
576
|
0
|
|
|
|
|
|
my ($w, $p, $s) = $new_word =~ /([^\#]+)(?:\#([^\#]+)(?:\#([^\#]+))?)?/; |
577
|
0
|
|
|
|
|
|
printf OFH "%25s", $orig_word; |
578
|
0
|
|
|
|
|
|
printf OFH " %24s", $w; |
579
|
0
|
0
|
|
|
|
|
printf OFH "%3s", $p if defined $p; |
580
|
0
|
0
|
|
|
|
|
printf OFH "%3s", $s if defined $s; |
581
|
0
|
|
|
|
|
|
print OFH "\n"; |
582
|
|
|
|
|
|
|
} |
583
|
|
|
|
|
|
|
|
584
|
0
|
|
|
|
|
|
close OFH; |
585
|
|
|
|
|
|
|
} |
586
|
|
|
|
|
|
|
|
587
|
0
|
|
|
|
|
|
return @rval; |
588
|
|
|
|
|
|
|
} |
589
|
|
|
|
|
|
|
|
590
|
|
|
|
|
|
|
sub _initializeContext |
591
|
|
|
|
|
|
|
{ |
592
|
0
|
|
|
0
|
|
|
my $self = shift; |
593
|
0
|
|
|
|
|
|
my $tagged = shift; |
594
|
0
|
|
|
|
|
|
my $wn = $wordnet{$self}; |
595
|
0
|
|
|
|
|
|
my $wnt = $wntools{$self}; |
596
|
0
|
|
|
|
|
|
my $nocompoundify = $nocompoundify{$self}; |
597
|
|
|
|
|
|
|
|
598
|
0
|
|
|
|
|
|
my @context = @_; |
599
|
|
|
|
|
|
|
|
600
|
|
|
|
|
|
|
# compoundify the words (if the text is raw) |
601
|
0
|
0
|
0
|
|
|
|
if (defined $wnt and $nocompoundify == 0 and defined !$tagged and !$wnformat{$self} ) { |
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
602
|
0
|
|
|
|
|
|
@context = split(/ +/,$wnt->compoundify("@context")); |
603
|
|
|
|
|
|
|
} |
604
|
|
|
|
|
|
|
|
605
|
0
|
|
|
|
|
|
my @newcontext; |
606
|
|
|
|
|
|
|
# do stoplisting |
607
|
0
|
0
|
|
|
|
|
if ($stoplist{$self}) { |
608
|
0
|
|
|
|
|
|
foreach my $word (@context) { |
609
|
0
|
0
|
|
|
|
|
if ($self->isStop ($word)) { |
610
|
0
|
|
|
|
|
|
push @newcontext, $word."#o"; |
611
|
|
|
|
|
|
|
} |
612
|
|
|
|
|
|
|
else { |
613
|
0
|
|
|
|
|
|
push @newcontext, $word; |
614
|
|
|
|
|
|
|
} |
615
|
|
|
|
|
|
|
} |
616
|
|
|
|
|
|
|
} |
617
|
|
|
|
|
|
|
else { |
618
|
0
|
|
|
|
|
|
@newcontext = @context; |
619
|
|
|
|
|
|
|
} |
620
|
|
|
|
|
|
|
|
621
|
|
|
|
|
|
|
# convert POS tags, if we have tagged text |
622
|
0
|
0
|
|
|
|
|
if ($tagged) { |
623
|
0
|
|
|
|
|
|
foreach my $wpos (@newcontext) { |
624
|
0
|
|
|
|
|
|
$wpos = $self->convertTag ($wpos); |
625
|
0
|
0
|
|
|
|
|
if (!defined $wpos) { |
626
|
0
|
|
|
|
|
|
$wpos="#MW"; |
627
|
|
|
|
|
|
|
} |
628
|
|
|
|
|
|
|
} |
629
|
|
|
|
|
|
|
} |
630
|
|
|
|
|
|
|
|
631
|
0
|
|
|
|
|
|
return @newcontext; |
632
|
|
|
|
|
|
|
} |
633
|
|
|
|
|
|
|
|
634
|
|
|
|
|
|
|
sub doNormal { |
635
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
636
|
0
|
|
|
|
|
|
my $pairScore = shift; |
637
|
0
|
|
|
|
|
|
my $contextScore = shift; |
638
|
0
|
|
|
|
|
|
my $window = shift; |
639
|
0
|
|
|
|
|
|
my @context = @_; |
640
|
|
|
|
|
|
|
|
641
|
0
|
|
|
|
|
|
my $lwindow = $window >> 1; # simply divide by 2 & throw away remainder |
642
|
0
|
|
|
|
|
|
my $rwindow = $window - $lwindow - 1; |
643
|
|
|
|
|
|
|
|
644
|
|
|
|
|
|
|
# get all the senses for each word |
645
|
0
|
|
|
|
|
|
my @senses = $self->_getSenses (\@context); |
646
|
|
|
|
|
|
|
|
647
|
|
|
|
|
|
|
|
648
|
|
|
|
|
|
|
# disambiguate |
649
|
0
|
|
|
|
|
|
my @results; |
650
|
|
|
|
|
|
|
|
651
|
0
|
|
|
|
|
|
local $| = 1; |
652
|
|
|
|
|
|
|
|
653
|
0
|
|
|
|
|
|
my $sense1firstword = 0; |
654
|
|
|
|
|
|
|
|
655
|
|
|
|
|
|
|
# for each word in the context, disambiguate the (target) word |
656
|
0
|
|
|
|
|
|
for my $targetIdx (0..$#context) { |
657
|
0
|
|
|
|
|
|
my @target_scores; |
658
|
|
|
|
|
|
|
|
659
|
0
|
0
|
|
|
|
|
unless (ref $senses[$targetIdx]) { |
660
|
0
|
|
|
|
|
|
$results[$targetIdx] = $context[$targetIdx]; |
661
|
0
|
|
|
|
|
|
next; |
662
|
|
|
|
|
|
|
} |
663
|
|
|
|
|
|
|
|
664
|
|
|
|
|
|
|
|
665
|
|
|
|
|
|
|
# figure out which words are in the window |
666
|
0
|
|
|
|
|
|
my $lower = $targetIdx - $lwindow; |
667
|
0
|
0
|
|
|
|
|
$lower = 0 if $lower < 0; |
668
|
0
|
|
|
|
|
|
my $upper = $targetIdx + $rwindow; |
669
|
0
|
0
|
|
|
|
|
$upper = $#context if $upper > $#context; |
670
|
|
|
|
|
|
|
|
671
|
|
|
|
|
|
|
# expand context window to the left, if necessary |
672
|
0
|
|
|
|
|
|
my $i = $targetIdx - 1; |
673
|
0
|
|
|
|
|
|
while ($i >= $lower) { |
674
|
0
|
0
|
|
|
|
|
last if $lower == 0; |
675
|
0
|
0
|
0
|
|
|
|
unless (defined $senses[$i] and (scalar @{$senses[$i]} > 0)) { |
|
0
|
|
|
|
|
|
|
676
|
0
|
|
|
|
|
|
$lower--; |
677
|
|
|
|
|
|
|
} |
678
|
0
|
|
|
|
|
|
$i--; |
679
|
|
|
|
|
|
|
} |
680
|
|
|
|
|
|
|
|
681
|
|
|
|
|
|
|
# expand context window to the right, if necessary |
682
|
0
|
|
|
|
|
|
my $j = $targetIdx + 1; |
683
|
0
|
|
|
|
|
|
while ($j <= $upper) { |
684
|
0
|
0
|
|
|
|
|
last if $upper >= scalar $#context; |
685
|
0
|
0
|
0
|
|
|
|
unless (defined $senses[$j] and (scalar @{$senses[$j]} > 0)) { |
|
0
|
|
|
|
|
|
|
686
|
0
|
|
|
|
|
|
$upper++; |
687
|
|
|
|
|
|
|
} |
688
|
0
|
|
|
|
|
|
$j++; |
689
|
|
|
|
|
|
|
} |
690
|
|
|
|
|
|
|
|
691
|
|
|
|
|
|
|
# If it is the first word in a sentence and the window size is 2, we'll |
692
|
|
|
|
|
|
|
# consider a word at the right of the target word. Otherwise it will not be |
693
|
|
|
|
|
|
|
# assigned any sense. In the previous version we were simply doing sense1 |
694
|
|
|
|
|
|
|
# which gave a boost to window=2 results. |
695
|
|
|
|
|
|
|
|
696
|
0
|
0
|
0
|
|
|
|
if ($targetIdx==0 && $window == 2){ |
697
|
0
|
|
|
|
|
|
$upper=1; |
698
|
|
|
|
|
|
|
} |
699
|
|
|
|
|
|
|
|
700
|
|
|
|
|
|
|
# do some tracing |
701
|
0
|
0
|
0
|
|
|
|
if ($trace{$self} and ($trace{$self}->{level} & TR_CONTEXT)) { |
702
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= "Context: "; |
703
|
0
|
0
|
|
|
|
|
if ($lower < $targetIdx) { |
704
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= |
705
|
|
|
|
|
|
|
join (' ', @context[$lower..$targetIdx-1]) . ' '; |
706
|
|
|
|
|
|
|
|
707
|
|
|
|
|
|
|
} |
708
|
|
|
|
|
|
|
|
709
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= |
710
|
|
|
|
|
|
|
"$context[$targetIdx]"; |
711
|
|
|
|
|
|
|
|
712
|
0
|
0
|
|
|
|
|
if ($targetIdx < $upper) { |
713
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= ' ' . |
714
|
|
|
|
|
|
|
join (' ', @context[($targetIdx+1)..$upper]); |
715
|
|
|
|
|
|
|
} |
716
|
|
|
|
|
|
|
|
717
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= "\n"; |
718
|
|
|
|
|
|
|
} |
719
|
|
|
|
|
|
|
|
720
|
0
|
|
|
|
|
|
my $result; |
721
|
0
|
0
|
|
|
|
|
if ($sense1firstword) { |
722
|
|
|
|
|
|
|
########################## |
723
|
0
|
|
|
|
|
|
my $word = $context[$targetIdx]; |
724
|
|
|
|
|
|
|
|
725
|
0
|
|
|
|
|
|
my $t = $self->getSense1 (\$context[$targetIdx]); |
726
|
0
|
0
|
|
|
|
|
if (defined $t) { |
727
|
0
|
|
|
|
|
|
$sense1firstword = 0; |
728
|
0
|
|
|
|
|
|
$result = $t; |
729
|
|
|
|
|
|
|
} |
730
|
|
|
|
|
|
|
else { |
731
|
0
|
|
|
|
|
|
$result = $context[$targetIdx]; |
732
|
|
|
|
|
|
|
} |
733
|
|
|
|
|
|
|
} |
734
|
|
|
|
|
|
|
else { |
735
|
0
|
0
|
|
|
|
|
if ($forcepos{$self}) { |
736
|
0
|
|
|
|
|
|
$result = $self->_forcedPosDisambig ($lower, $targetIdx, |
737
|
|
|
|
|
|
|
$upper, \@senses, |
738
|
|
|
|
|
|
|
\@context); |
739
|
|
|
|
|
|
|
} |
740
|
|
|
|
|
|
|
else { |
741
|
0
|
|
|
|
|
|
$result = $self->_normalDisambig ($lower, $targetIdx, $upper, |
742
|
|
|
|
|
|
|
\@senses, \@context); |
743
|
|
|
|
|
|
|
} |
744
|
|
|
|
|
|
|
} |
745
|
|
|
|
|
|
|
|
746
|
0
|
0
|
|
|
|
|
if ($fixed{$self}) { |
747
|
0
|
0
|
|
|
|
|
if ($result =~ /\#[nvars]\#\d/) { |
748
|
0
|
|
|
|
|
|
$senses[$targetIdx] = [$result]; |
749
|
|
|
|
|
|
|
} |
750
|
|
|
|
|
|
|
} |
751
|
|
|
|
|
|
|
|
752
|
0
|
|
|
|
|
|
push @results, $result; |
753
|
|
|
|
|
|
|
} |
754
|
|
|
|
|
|
|
|
755
|
0
|
|
|
|
|
|
return @results; |
756
|
|
|
|
|
|
|
} |
757
|
|
|
|
|
|
|
|
758
|
|
|
|
|
|
|
=item B |
759
|
|
|
|
|
|
|
|
760
|
|
|
|
|
|
|
Gets the current trace string and resets it to "". |
761
|
|
|
|
|
|
|
|
762
|
|
|
|
|
|
|
Parameters: |
763
|
|
|
|
|
|
|
|
764
|
|
|
|
|
|
|
None |
765
|
|
|
|
|
|
|
|
766
|
|
|
|
|
|
|
Returns: |
767
|
|
|
|
|
|
|
|
768
|
|
|
|
|
|
|
The current trace string (before resetting it). If the returned string |
769
|
|
|
|
|
|
|
is not empty, it will end with a newline. |
770
|
|
|
|
|
|
|
|
771
|
|
|
|
|
|
|
Example: |
772
|
|
|
|
|
|
|
|
773
|
|
|
|
|
|
|
my $str = $wsd->getTrace (); |
774
|
|
|
|
|
|
|
print $str; |
775
|
|
|
|
|
|
|
|
776
|
|
|
|
|
|
|
=cut |
777
|
|
|
|
|
|
|
|
778
|
|
|
|
|
|
|
sub getTrace |
779
|
|
|
|
|
|
|
{ |
780
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
781
|
0
|
|
|
|
|
|
my $str = $trace{$self}->{string}; |
782
|
0
|
|
|
|
|
|
$trace{$self}->{string} = ''; |
783
|
0
|
|
|
|
|
|
return $str; |
784
|
|
|
|
|
|
|
} |
785
|
|
|
|
|
|
|
|
786
|
|
|
|
|
|
|
# does sense 1 disambiguation |
787
|
|
|
|
|
|
|
sub doSense1 |
788
|
|
|
|
|
|
|
{ |
789
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
790
|
0
|
|
|
|
|
|
my @words = @_; |
791
|
0
|
|
|
|
|
|
my $wn = $wordnet{$self}; |
792
|
|
|
|
|
|
|
|
793
|
0
|
|
|
|
|
|
my @disambiguated; |
794
|
|
|
|
|
|
|
|
795
|
0
|
|
|
|
|
|
foreach my $word (@words) { |
796
|
0
|
|
|
|
|
|
my $tmp = $self->getSense1 (\$word); |
797
|
0
|
0
|
|
|
|
|
if (defined $tmp) { |
798
|
0
|
|
|
|
|
|
push @disambiguated, $tmp; |
799
|
|
|
|
|
|
|
} |
800
|
|
|
|
|
|
|
else { |
801
|
0
|
|
|
|
|
|
push @disambiguated, $word; |
802
|
|
|
|
|
|
|
} |
803
|
|
|
|
|
|
|
} |
804
|
|
|
|
|
|
|
|
805
|
0
|
|
|
|
|
|
return @disambiguated; |
806
|
|
|
|
|
|
|
} |
807
|
|
|
|
|
|
|
|
808
|
|
|
|
|
|
|
# gets sense number 1 for the specified word. If the word has multiple forms, |
809
|
|
|
|
|
|
|
# then the most frequent sense is returned. If there is more than one |
810
|
|
|
|
|
|
|
# most frequent sense with sense number 1, a sense is chosen at random. |
811
|
|
|
|
|
|
|
# |
812
|
|
|
|
|
|
|
# this is not quite the same as choosing the most frequent sense of a word. |
813
|
|
|
|
|
|
|
# The sense number 1 in wordnet is often the most frequent but not always. |
814
|
|
|
|
|
|
|
sub getSense1 |
815
|
|
|
|
|
|
|
{ |
816
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
817
|
0
|
|
|
|
|
|
my $word_ref = shift; |
818
|
0
|
|
|
|
|
|
my $wn = $wordnet{$self}; |
819
|
0
|
|
|
|
|
|
my %senses; |
820
|
|
|
|
|
|
|
|
821
|
|
|
|
|
|
|
# check if word has error suffix in it, if it does, we can't do anything with it |
822
|
0
|
0
|
|
|
|
|
if (${$word_ref} =~ /\#o|\#IT|\#CL|\#NT|\#MW/) { |
|
0
|
|
|
|
|
|
|
823
|
0
|
|
|
|
|
|
return undef; |
824
|
|
|
|
|
|
|
} |
825
|
|
|
|
|
|
|
|
826
|
0
|
|
|
|
|
|
my @forms; |
827
|
0
|
0
|
|
|
|
|
unless ($wnformat{$self}) { |
828
|
0
|
|
|
|
|
|
@forms = $wn->validForms (${$word_ref}); |
|
0
|
|
|
|
|
|
|
829
|
|
|
|
|
|
|
} |
830
|
|
|
|
|
|
|
else { |
831
|
0
|
|
|
|
|
|
@forms = ${$word_ref}; |
|
0
|
|
|
|
|
|
|
832
|
|
|
|
|
|
|
} |
833
|
0
|
0
|
|
|
|
|
if (scalar @forms == 0) { |
834
|
0
|
|
|
|
|
|
${$word_ref}= "${$word_ref}"."#ND"; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
835
|
|
|
|
|
|
|
} |
836
|
|
|
|
|
|
|
else{ |
837
|
0
|
|
|
|
|
|
foreach my $form (@forms) { |
838
|
0
|
|
|
|
|
|
my @t = $wn->querySense ($form); |
839
|
0
|
0
|
|
|
|
|
if (scalar @t > 0) { |
840
|
0
|
|
|
|
|
|
$senses{$form} = $t[0]; |
841
|
|
|
|
|
|
|
} |
842
|
|
|
|
|
|
|
} |
843
|
|
|
|
|
|
|
} |
844
|
0
|
|
|
|
|
|
my @best_senses; |
845
|
|
|
|
|
|
|
|
846
|
0
|
|
|
|
|
|
foreach my $key (keys %senses) { |
847
|
0
|
|
|
|
|
|
my $sense = $senses{$key}; |
848
|
|
|
|
|
|
|
|
849
|
0
|
|
|
|
|
|
my $freq = $wn->frequency ($sense); |
850
|
|
|
|
|
|
|
|
851
|
0
|
0
|
|
|
|
|
if ($#best_senses < 0) { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
852
|
0
|
|
|
|
|
|
push @best_senses, [$sense, $freq]; |
853
|
|
|
|
|
|
|
} |
854
|
|
|
|
|
|
|
elsif ($best_senses[$#best_senses]->[1] < $freq) { |
855
|
0
|
|
|
|
|
|
@best_senses = ([$sense, $freq]); |
856
|
|
|
|
|
|
|
} |
857
|
|
|
|
|
|
|
elsif ($best_senses[$#best_senses]->[1] == $freq) { |
858
|
0
|
|
|
|
|
|
push @best_senses, [$sense, $freq]; |
859
|
|
|
|
|
|
|
} |
860
|
|
|
|
|
|
|
else { |
861
|
|
|
|
|
|
|
# do nothing |
862
|
|
|
|
|
|
|
} |
863
|
|
|
|
|
|
|
} |
864
|
|
|
|
|
|
|
|
865
|
0
|
0
|
|
|
|
|
if (scalar @best_senses) { |
866
|
0
|
|
|
|
|
|
my $i = int (rand (scalar @best_senses)); |
867
|
|
|
|
|
|
|
|
868
|
0
|
|
|
|
|
|
return $best_senses[$i]->[0]; |
869
|
|
|
|
|
|
|
} |
870
|
|
|
|
|
|
|
|
871
|
0
|
|
|
|
|
|
return undef; |
872
|
|
|
|
|
|
|
} |
873
|
|
|
|
|
|
|
|
874
|
|
|
|
|
|
|
# does random guessing. This could be considered a baseline approach |
875
|
|
|
|
|
|
|
# of sorts. Also try running normal disambiguation using the |
876
|
|
|
|
|
|
|
# WordNet::Similarity::random measure |
877
|
|
|
|
|
|
|
sub doRandom |
878
|
|
|
|
|
|
|
{ |
879
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
880
|
0
|
|
|
|
|
|
my @words = @_; |
881
|
0
|
|
|
|
|
|
my $wn = $wordnet{$self}; |
882
|
|
|
|
|
|
|
|
883
|
0
|
|
|
|
|
|
my $datapath = $wn->dataPath; |
884
|
|
|
|
|
|
|
|
885
|
0
|
|
|
|
|
|
my @disambiguated; |
886
|
|
|
|
|
|
|
|
887
|
0
|
|
|
|
|
|
foreach my $word (@words) { |
888
|
0
|
0
|
|
|
|
|
if ( $word =~ /\#o|\#IT|\#CL|\#NT|\#MW/) { |
889
|
|
|
|
|
|
|
# push the string into the array |
890
|
0
|
|
|
|
|
|
push @disambiguated, $word; |
891
|
0
|
|
|
|
|
|
next; |
892
|
|
|
|
|
|
|
} |
893
|
|
|
|
|
|
|
|
894
|
0
|
|
|
|
|
|
my @forms; |
895
|
0
|
0
|
|
|
|
|
unless ($wnformat{$self}) { |
896
|
0
|
|
|
|
|
|
@forms = $wn->validForms ($word); |
897
|
|
|
|
|
|
|
} |
898
|
|
|
|
|
|
|
else { |
899
|
0
|
|
|
|
|
|
@forms = $word; |
900
|
|
|
|
|
|
|
} |
901
|
0
|
|
|
|
|
|
my @senses; |
902
|
0
|
0
|
|
|
|
|
if (scalar @forms == 0) { |
903
|
0
|
|
|
|
|
|
$word= "$word"."#ND"; |
904
|
|
|
|
|
|
|
} |
905
|
|
|
|
|
|
|
else { |
906
|
0
|
|
|
|
|
|
foreach my $form (@forms) { |
907
|
0
|
|
|
|
|
|
my @t = $wn->querySense ($form); |
908
|
0
|
0
|
|
|
|
|
if (scalar @t > 0) { |
909
|
0
|
|
|
|
|
|
push @senses, @t; |
910
|
|
|
|
|
|
|
} |
911
|
|
|
|
|
|
|
} |
912
|
|
|
|
|
|
|
} |
913
|
0
|
0
|
|
|
|
|
if (scalar @senses) { |
914
|
0
|
|
|
|
|
|
my $i = int (rand (scalar @senses)); |
915
|
0
|
|
|
|
|
|
push @disambiguated, $senses[$i]; |
916
|
|
|
|
|
|
|
} |
917
|
|
|
|
|
|
|
else { |
918
|
0
|
|
|
|
|
|
push @disambiguated, $word; |
919
|
|
|
|
|
|
|
} |
920
|
|
|
|
|
|
|
|
921
|
|
|
|
|
|
|
|
922
|
|
|
|
|
|
|
} |
923
|
0
|
|
|
|
|
|
return @disambiguated; |
924
|
|
|
|
|
|
|
} |
925
|
|
|
|
|
|
|
|
926
|
|
|
|
|
|
|
sub _forcedPosDisambig |
927
|
|
|
|
|
|
|
{ |
928
|
0
|
|
|
0
|
|
|
my $self = shift; |
929
|
0
|
|
|
|
|
|
my $lower = shift; |
930
|
0
|
|
|
|
|
|
my $targetIdx = shift; |
931
|
0
|
|
|
|
|
|
my $upper = shift; |
932
|
0
|
|
|
|
|
|
my $senses_ref = shift; |
933
|
0
|
|
|
|
|
|
my $context_ref = shift; |
934
|
0
|
|
|
|
|
|
my $measure = $simMeasure{$self}; |
935
|
0
|
|
|
|
|
|
my $result; |
936
|
|
|
|
|
|
|
my @traces; |
937
|
0
|
|
|
|
|
|
my @target_scores; |
938
|
|
|
|
|
|
|
|
939
|
|
|
|
|
|
|
|
940
|
|
|
|
|
|
|
# for each sense of the target word ... |
941
|
0
|
|
|
|
|
|
for my $i (0..$#{$senses_ref->[$targetIdx]}) { |
|
0
|
|
|
|
|
|
|
942
|
0
|
0
|
0
|
|
|
|
unless (ref $senses_ref->[$targetIdx] |
943
|
|
|
|
|
|
|
and defined $senses_ref->[$targetIdx][$i]) { |
944
|
0
|
|
|
|
|
|
$target_scores[$i] = -1; |
945
|
0
|
|
|
|
|
|
next; |
946
|
|
|
|
|
|
|
} |
947
|
|
|
|
|
|
|
|
948
|
0
|
|
|
|
|
|
$target_scores[$i] = 0; |
949
|
|
|
|
|
|
|
|
950
|
0
|
|
|
|
|
|
my $target_pos = getPos ($senses_ref->[$targetIdx][$i]); |
951
|
|
|
|
|
|
|
|
952
|
|
|
|
|
|
|
# for each (context) word in the window around the target word |
953
|
0
|
|
|
|
|
|
for my $contextIdx ($lower..$upper) { |
954
|
0
|
0
|
|
|
|
|
next if $contextIdx == $targetIdx; |
955
|
0
|
0
|
|
|
|
|
next unless ref $senses_ref->[$contextIdx]; |
956
|
|
|
|
|
|
|
|
957
|
0
|
|
|
|
|
|
my @tempScores; |
958
|
|
|
|
|
|
|
|
959
|
|
|
|
|
|
|
my @goodsenses; |
960
|
|
|
|
|
|
|
# * check if senses for context word work with target word * |
961
|
0
|
0
|
|
|
|
|
if (needCoercePos ($target_pos, $senses_ref->[$contextIdx])) { |
962
|
0
|
|
|
|
|
|
@goodsenses = $self->coercePos ($context_ref->[$contextIdx], |
963
|
|
|
|
|
|
|
$target_pos); |
964
|
|
|
|
|
|
|
} |
965
|
|
|
|
|
|
|
else { |
966
|
0
|
|
|
|
|
|
@goodsenses = @{$senses_ref->[$contextIdx]}; |
|
0
|
|
|
|
|
|
|
967
|
|
|
|
|
|
|
} |
968
|
|
|
|
|
|
|
|
969
|
|
|
|
|
|
|
# for each sense of the context word in the window |
970
|
0
|
|
|
|
|
|
for my $k (0..$#{$senses_ref->[$contextIdx]}) { |
|
0
|
|
|
|
|
|
|
971
|
0
|
0
|
|
|
|
|
unless (defined $senses_ref->[$contextIdx][$k]) { |
972
|
0
|
|
|
|
|
|
$tempScores[$k] = -1; |
973
|
0
|
|
|
|
|
|
next; |
974
|
|
|
|
|
|
|
} |
975
|
|
|
|
|
|
|
|
976
|
0
|
|
|
|
|
|
$tempScores[$k] = |
977
|
|
|
|
|
|
|
$measure->getRelatedness ($senses_ref->[$targetIdx][$i], |
978
|
|
|
|
|
|
|
$senses_ref->[$contextIdx][$k]); |
979
|
|
|
|
|
|
|
|
980
|
0
|
0
|
|
|
|
|
if ($trace{$self}->{level} & TR_PAIRWISE) { |
981
|
|
|
|
|
|
|
# only trace zero values if TR_ZERO is specified |
982
|
0
|
0
|
0
|
|
|
|
if ((defined $tempScores[$k] and $tempScores[$k] > 0) |
|
|
|
0
|
|
|
|
|
983
|
|
|
|
|
|
|
or ($trace{$self}->{level} & TR_ZERO)) { |
984
|
0
|
0
|
|
|
|
|
my $s = " " |
985
|
|
|
|
|
|
|
. $senses_ref->[$targetIdx][$i] . ' ' |
986
|
|
|
|
|
|
|
. $senses_ref->[$contextIdx][$k] . ' ' |
987
|
|
|
|
|
|
|
. (defined $tempScores[$k] |
988
|
|
|
|
|
|
|
? $tempScores[$k] |
989
|
|
|
|
|
|
|
: 'undef'); |
990
|
0
|
|
|
|
|
|
push @{$traces[$i]}, $s; |
|
0
|
|
|
|
|
|
|
991
|
|
|
|
|
|
|
} |
992
|
|
|
|
|
|
|
} |
993
|
|
|
|
|
|
|
|
994
|
0
|
0
|
0
|
|
|
|
if ($trace{$self}->{level} & TR_MEASURE |
|
|
|
0
|
|
|
|
|
995
|
|
|
|
|
|
|
and ((defined $tempScores[$k] and $tempScores[$k] > 0) |
996
|
|
|
|
|
|
|
or ($trace{$self}->{level} & TR_ZERO))) { |
997
|
0
|
|
|
|
|
|
push @{$traces[$i]}, $measure->getTraceString (); |
|
0
|
|
|
|
|
|
|
998
|
|
|
|
|
|
|
} |
999
|
|
|
|
|
|
|
# clear errors in Similarity object |
1000
|
0
|
0
|
|
|
|
|
$measure->getError () unless defined $tempScores[$k]; |
1001
|
|
|
|
|
|
|
} |
1002
|
0
|
|
|
|
|
|
my $best = -2; |
1003
|
0
|
|
|
|
|
|
foreach my $temp (@tempScores) { |
1004
|
0
|
0
|
|
|
|
|
next unless defined $temp; |
1005
|
0
|
0
|
|
|
|
|
$best = $temp if $temp > $best; |
1006
|
|
|
|
|
|
|
} |
1007
|
|
|
|
|
|
|
|
1008
|
0
|
0
|
|
|
|
|
if ($best > $pairScore{$self}) { |
1009
|
0
|
|
|
|
|
|
$target_scores[$i] += $best; |
1010
|
|
|
|
|
|
|
} |
1011
|
|
|
|
|
|
|
} |
1012
|
|
|
|
|
|
|
} |
1013
|
|
|
|
|
|
|
|
1014
|
|
|
|
|
|
|
# find the best score for this sense of the target word |
1015
|
|
|
|
|
|
|
|
1016
|
|
|
|
|
|
|
# first, do a bit of tracing |
1017
|
0
|
0
|
0
|
|
|
|
if (ref $trace{$self} and ($trace{$self}->{level} & TR_ALLSCORES)) { |
1018
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " Scores for $context_ref->[$targetIdx]\n"; |
1019
|
|
|
|
|
|
|
} |
1020
|
|
|
|
|
|
|
|
1021
|
|
|
|
|
|
|
# now find the best sense |
1022
|
0
|
|
|
|
|
|
my $best_tscore = -1; |
1023
|
0
|
|
|
|
|
|
foreach my $i (0..$#target_scores) { |
1024
|
0
|
|
|
|
|
|
my $tscore = $target_scores[$i]; |
1025
|
0
|
0
|
|
|
|
|
next unless defined $tscore; |
1026
|
|
|
|
|
|
|
|
1027
|
0
|
0
|
0
|
|
|
|
if ($trace{$self}->{level} & TR_ALLSCORES |
|
|
|
0
|
|
|
|
|
1028
|
|
|
|
|
|
|
and (($tscore > 0) or ($trace{$self}->{level} & TR_ZERO))) { |
1029
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " $senses_ref->[$targetIdx][$i]: $tscore\n"; |
1030
|
|
|
|
|
|
|
} |
1031
|
|
|
|
|
|
|
|
1032
|
0
|
0
|
0
|
|
|
|
if (($trace{$self}->{level} & TR_MEASURE |
|
|
|
0
|
|
|
|
|
1033
|
|
|
|
|
|
|
or $trace{$self}->{level} & TR_PAIRWISE) |
1034
|
|
|
|
|
|
|
and defined $traces[$i]) { |
1035
|
0
|
|
|
|
|
|
foreach my $str (@{$traces[$i]}) { |
|
0
|
|
|
|
|
|
|
1036
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= $str . "\n"; |
1037
|
|
|
|
|
|
|
} |
1038
|
|
|
|
|
|
|
} |
1039
|
|
|
|
|
|
|
|
1040
|
|
|
|
|
|
|
# ignore scores less than the threshold |
1041
|
0
|
0
|
|
|
|
|
next unless $tscore > $contextScore{$self}; |
1042
|
|
|
|
|
|
|
|
1043
|
0
|
0
|
|
|
|
|
if ($tscore > $best_tscore) { |
1044
|
0
|
|
|
|
|
|
$result = $senses_ref->[$targetIdx][$i]; |
1045
|
0
|
|
|
|
|
|
$best_tscore = $tscore; |
1046
|
|
|
|
|
|
|
} |
1047
|
|
|
|
|
|
|
} |
1048
|
|
|
|
|
|
|
|
1049
|
0
|
0
|
|
|
|
|
if ($best_tscore < 0) { |
1050
|
0
|
|
|
|
|
|
$result = $context_ref->[$targetIdx]; |
1051
|
|
|
|
|
|
|
} |
1052
|
|
|
|
|
|
|
|
1053
|
0
|
0
|
0
|
|
|
|
if (ref $trace{$self} and $trace{$self}->{level} & TR_BESTSCORE) { |
1054
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " Winning score: $best_tscore\n"; |
1055
|
|
|
|
|
|
|
} |
1056
|
|
|
|
|
|
|
|
1057
|
0
|
|
|
|
|
|
return $result; |
1058
|
|
|
|
|
|
|
} |
1059
|
|
|
|
|
|
|
|
1060
|
|
|
|
|
|
|
sub _normalDisambig |
1061
|
|
|
|
|
|
|
{ |
1062
|
0
|
|
|
0
|
|
|
my $self = shift; |
1063
|
0
|
|
|
|
|
|
my $lower = shift; |
1064
|
0
|
|
|
|
|
|
my $targetIdx = shift; |
1065
|
0
|
|
|
|
|
|
my $upper = shift; |
1066
|
0
|
|
|
|
|
|
my $senses_ref = shift; |
1067
|
0
|
|
|
|
|
|
my $context_ref = shift; |
1068
|
0
|
|
|
|
|
|
my $measure = $simMeasure{$self}; |
1069
|
0
|
|
|
|
|
|
my $usemono = $usemono{$self}; |
1070
|
0
|
|
|
|
|
|
my $backoff = $backoff{$self}; |
1071
|
|
|
|
|
|
|
|
1072
|
0
|
|
|
|
|
|
my $result; |
1073
|
|
|
|
|
|
|
|
1074
|
|
|
|
|
|
|
my @traces; |
1075
|
0
|
|
|
|
|
|
my @target_scores; |
1076
|
|
|
|
|
|
|
|
1077
|
|
|
|
|
|
|
# for each sense of the target word ... |
1078
|
0
|
|
|
|
|
|
for my $i (0..$#{$senses_ref->[$targetIdx]}) { |
|
0
|
|
|
|
|
|
|
1079
|
0
|
0
|
0
|
|
|
|
unless (ref $senses_ref->[$targetIdx] |
1080
|
|
|
|
|
|
|
and defined $senses_ref->[$targetIdx][$i]) { |
1081
|
0
|
|
|
|
|
|
$target_scores[$i] = -1; |
1082
|
0
|
|
|
|
|
|
next; |
1083
|
|
|
|
|
|
|
} |
1084
|
0
|
|
|
|
|
|
$target_scores[$i] = 0; |
1085
|
|
|
|
|
|
|
# If --usemono flag is on and the word has only one sense then assign it. |
1086
|
|
|
|
|
|
|
# This flag will be off by default. |
1087
|
0
|
0
|
0
|
|
|
|
if($usemono == 1 && $#{$senses_ref->[$targetIdx]} == 0){ |
|
0
|
|
|
|
|
|
|
1088
|
0
|
|
|
|
|
|
$result = $senses_ref->[$targetIdx][0]; |
1089
|
0
|
|
|
|
|
|
return $result; |
1090
|
|
|
|
|
|
|
} |
1091
|
|
|
|
|
|
|
#my @tempScores; |
1092
|
|
|
|
|
|
|
|
1093
|
|
|
|
|
|
|
|
1094
|
|
|
|
|
|
|
# for each (context) word in the window around the target word |
1095
|
0
|
|
|
|
|
|
for my $contextIdx ($lower..$upper) { |
1096
|
0
|
|
|
|
|
|
my @tempScores = (); |
1097
|
0
|
0
|
|
|
|
|
next if $contextIdx == $targetIdx; |
1098
|
0
|
0
|
|
|
|
|
next unless ref $senses_ref->[$contextIdx]; |
1099
|
|
|
|
|
|
|
|
1100
|
|
|
|
|
|
|
# for each sense of the context word in the window |
1101
|
0
|
|
|
|
|
|
for my $k (0..$#{$senses_ref->[$contextIdx]}) { |
|
0
|
|
|
|
|
|
|
1102
|
0
|
0
|
|
|
|
|
unless (defined $senses_ref->[$contextIdx][$k]) { |
1103
|
0
|
|
|
|
|
|
$tempScores[$k] = -1; |
1104
|
0
|
|
|
|
|
|
next; |
1105
|
|
|
|
|
|
|
} |
1106
|
|
|
|
|
|
|
|
1107
|
0
|
|
|
|
|
|
$tempScores[$k] = |
1108
|
|
|
|
|
|
|
$measure->getRelatedness ($senses_ref->[$targetIdx][$i], |
1109
|
|
|
|
|
|
|
$senses_ref->[$contextIdx][$k]); |
1110
|
|
|
|
|
|
|
|
1111
|
0
|
0
|
|
|
|
|
if ($trace{$self}->{level} & TR_PAIRWISE) { |
1112
|
|
|
|
|
|
|
# only trace zero values if TR_ZERO is specified |
1113
|
0
|
0
|
0
|
|
|
|
if ((defined $tempScores[$k] and $tempScores[$k] > 0) |
|
|
|
0
|
|
|
|
|
1114
|
|
|
|
|
|
|
or ($trace{$self}->{level} & TR_ZERO)) { |
1115
|
0
|
0
|
|
|
|
|
my $s = " " |
1116
|
|
|
|
|
|
|
.$senses_ref->[$targetIdx][$i] . ' ' |
1117
|
|
|
|
|
|
|
.$senses_ref->[$contextIdx][$k] . ' ' |
1118
|
|
|
|
|
|
|
. (defined $tempScores[$k] |
1119
|
|
|
|
|
|
|
? $tempScores[$k] |
1120
|
|
|
|
|
|
|
: 'undef'); |
1121
|
|
|
|
|
|
|
|
1122
|
0
|
|
|
|
|
|
push @{$traces[$i]}, $s; |
|
0
|
|
|
|
|
|
|
1123
|
|
|
|
|
|
|
} |
1124
|
|
|
|
|
|
|
} |
1125
|
|
|
|
|
|
|
|
1126
|
0
|
0
|
0
|
|
|
|
if ($trace{$self}->{level} & TR_MEASURE |
|
|
|
0
|
|
|
|
|
1127
|
|
|
|
|
|
|
and ((defined $tempScores[$k] and $tempScores[$k] > 0) |
1128
|
|
|
|
|
|
|
or ($trace{$self}->{level} & TR_ZERO))) { |
1129
|
0
|
|
|
|
|
|
push @{$traces[$i]}, $measure->getTraceString (); |
|
0
|
|
|
|
|
|
|
1130
|
|
|
|
|
|
|
} |
1131
|
|
|
|
|
|
|
|
1132
|
|
|
|
|
|
|
# clear errors in Similarity object |
1133
|
0
|
0
|
|
|
|
|
$measure->getError () unless defined $tempScores[$k]; |
1134
|
|
|
|
|
|
|
} |
1135
|
0
|
|
|
|
|
|
my $best = -2; |
1136
|
0
|
|
|
|
|
|
foreach my $temp (@tempScores) { |
1137
|
0
|
0
|
|
|
|
|
next unless defined $temp; |
1138
|
|
|
|
|
|
|
|
1139
|
0
|
0
|
|
|
|
|
$best = $temp if $temp > $best; |
1140
|
|
|
|
|
|
|
} |
1141
|
|
|
|
|
|
|
|
1142
|
0
|
0
|
|
|
|
|
if ($best > $pairScore{$self}) { |
1143
|
0
|
|
|
|
|
|
$target_scores[$i] += $best; |
1144
|
|
|
|
|
|
|
} |
1145
|
|
|
|
|
|
|
} |
1146
|
|
|
|
|
|
|
} |
1147
|
|
|
|
|
|
|
|
1148
|
|
|
|
|
|
|
# find the best score for this sense of the target word |
1149
|
|
|
|
|
|
|
|
1150
|
|
|
|
|
|
|
# first, do a bit of tracing |
1151
|
0
|
0
|
0
|
|
|
|
if (ref $trace{$self} and ($trace{$self}->{level} & TR_ALLSCORES)) { |
1152
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " Scores for $context_ref->[$targetIdx]\n"; |
1153
|
|
|
|
|
|
|
} |
1154
|
|
|
|
|
|
|
|
1155
|
|
|
|
|
|
|
# now find the best sense |
1156
|
0
|
|
|
|
|
|
my $best_tscore = -1; |
1157
|
|
|
|
|
|
|
|
1158
|
0
|
|
|
|
|
|
foreach my $i (0..$#target_scores) { |
1159
|
0
|
|
|
|
|
|
my $tscore = $target_scores[$i]; |
1160
|
0
|
0
|
|
|
|
|
next unless defined $tscore; |
1161
|
|
|
|
|
|
|
|
1162
|
0
|
0
|
0
|
|
|
|
if ($trace{$self}->{level} & TR_ALLSCORES |
|
|
|
0
|
|
|
|
|
1163
|
|
|
|
|
|
|
&& (($tscore > 0) or ($trace{$self}->{level} & TR_ZERO))) { |
1164
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " $senses_ref->[$targetIdx][$i]: $tscore\n"; |
1165
|
|
|
|
|
|
|
} |
1166
|
|
|
|
|
|
|
|
1167
|
0
|
0
|
0
|
|
|
|
if (($trace{$self}->{level} & TR_MEASURE |
|
|
|
0
|
|
|
|
|
1168
|
|
|
|
|
|
|
or $trace{$self}->{level} & TR_PAIRWISE) |
1169
|
|
|
|
|
|
|
and defined $traces[$i]) { |
1170
|
0
|
|
|
|
|
|
foreach my $str (@{$traces[$i]}) { |
|
0
|
|
|
|
|
|
|
1171
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= $str . "\n"; |
1172
|
|
|
|
|
|
|
} |
1173
|
|
|
|
|
|
|
} |
1174
|
|
|
|
|
|
|
|
1175
|
|
|
|
|
|
|
# ignore scores less than the threshold |
1176
|
0
|
0
|
|
|
|
|
next unless $tscore > $contextScore{$self}; |
1177
|
|
|
|
|
|
|
|
1178
|
0
|
0
|
|
|
|
|
if ($tscore > $best_tscore) { |
1179
|
0
|
|
|
|
|
|
$result = $senses_ref->[$targetIdx][$i]; |
1180
|
0
|
|
|
|
|
|
$best_tscore = $tscore; |
1181
|
|
|
|
|
|
|
} |
1182
|
|
|
|
|
|
|
} |
1183
|
|
|
|
|
|
|
|
1184
|
0
|
0
|
|
|
|
|
if ($best_tscore < 0) { |
1185
|
|
|
|
|
|
|
#$result = $context_ref->[$targetIdx]; |
1186
|
0
|
|
|
|
|
|
$result = "$context_ref->[$targetIdx]"."#NR"; |
1187
|
0
|
0
|
|
|
|
|
if($backoff == 1){ |
1188
|
0
|
|
|
|
|
|
$result = $self->getSense1(\$context_ref->[$targetIdx]); |
1189
|
|
|
|
|
|
|
} |
1190
|
|
|
|
|
|
|
} |
1191
|
|
|
|
|
|
|
|
1192
|
0
|
0
|
0
|
|
|
|
if (ref $trace{$self} and $trace{$self}->{level} & TR_BESTSCORE) { |
1193
|
0
|
|
|
|
|
|
$trace{$self}->{string} .= " Winning score: $best_tscore\n"; |
1194
|
|
|
|
|
|
|
} |
1195
|
|
|
|
|
|
|
|
1196
|
|
|
|
|
|
|
# if ($trace{$self}->{level} & 8) { # | $trace{$self}->{level} & TR_PAIRWISE) { |
1197
|
|
|
|
|
|
|
# foreach my $str (@traces) { |
1198
|
|
|
|
|
|
|
# $trace{$self}->{string} .= "$str\n"; |
1199
|
|
|
|
|
|
|
# } |
1200
|
|
|
|
|
|
|
# @traces = (); |
1201
|
|
|
|
|
|
|
# } |
1202
|
0
|
|
|
|
|
|
return $result; |
1203
|
|
|
|
|
|
|
} |
1204
|
|
|
|
|
|
|
|
1205
|
|
|
|
|
|
|
sub isStop |
1206
|
|
|
|
|
|
|
{ |
1207
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
1208
|
0
|
|
|
|
|
|
my $word = shift; |
1209
|
0
|
|
|
|
|
|
foreach my $re (@{$stoplist{$self}}) { |
|
0
|
|
|
|
|
|
|
1210
|
0
|
0
|
|
|
|
|
if ($word =~ /^$re$/) { |
1211
|
0
|
|
|
|
|
|
return 1; |
1212
|
|
|
|
|
|
|
} |
1213
|
|
|
|
|
|
|
} |
1214
|
0
|
|
|
|
|
|
return 0; |
1215
|
|
|
|
|
|
|
} |
1216
|
|
|
|
|
|
|
|
1217
|
|
|
|
|
|
|
# checks to see if the POS of at least one word#pos#sense string in $aref |
1218
|
|
|
|
|
|
|
# is $pos |
1219
|
|
|
|
|
|
|
sub needCoercePos |
1220
|
|
|
|
|
|
|
{ |
1221
|
0
|
|
|
0
|
0
|
|
my $pos = shift; |
1222
|
|
|
|
|
|
|
|
1223
|
|
|
|
|
|
|
# Only coerce if target POS is noun or verb. |
1224
|
|
|
|
|
|
|
# The measures that take advantage of POS coercion only work with |
1225
|
|
|
|
|
|
|
# nouns and verbs. |
1226
|
0
|
0
|
0
|
|
|
|
unless ($pos eq 'n' or $pos eq 'v') { |
1227
|
0
|
|
|
|
|
|
return 0; |
1228
|
|
|
|
|
|
|
} |
1229
|
|
|
|
|
|
|
|
1230
|
0
|
|
|
|
|
|
my $aref = shift; |
1231
|
0
|
|
|
|
|
|
foreach my $wps (@$aref) { |
1232
|
0
|
0
|
|
|
|
|
if ($pos eq getPos ($wps)) { |
1233
|
0
|
|
|
|
|
|
return 0; |
1234
|
|
|
|
|
|
|
} |
1235
|
|
|
|
|
|
|
} |
1236
|
0
|
|
|
|
|
|
return 1; |
1237
|
|
|
|
|
|
|
} |
1238
|
|
|
|
|
|
|
|
1239
|
|
|
|
|
|
|
sub convertTag |
1240
|
|
|
|
|
|
|
{ |
1241
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
1242
|
0
|
|
|
|
|
|
my $wordpos = shift; |
1243
|
0
|
|
|
|
|
|
my $index = index $wordpos, "/"; |
1244
|
|
|
|
|
|
|
|
1245
|
0
|
0
|
|
|
|
|
if ($index < 0) { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
1246
|
0
|
|
|
|
|
|
return $wordpos; |
1247
|
|
|
|
|
|
|
} |
1248
|
|
|
|
|
|
|
elsif ($index == 0) { |
1249
|
0
|
|
|
|
|
|
return undef; |
1250
|
|
|
|
|
|
|
} |
1251
|
|
|
|
|
|
|
elsif (index ($wordpos, "'") == 0) { |
1252
|
|
|
|
|
|
|
# we have a contraction |
1253
|
0
|
|
|
|
|
|
my $word = substr $wordpos, 0, $index; |
1254
|
0
|
|
|
|
|
|
my $tag = substr $wordpos, $index + 1; |
1255
|
0
|
|
|
|
|
|
return $self->convertContraction ($word, $tag); |
1256
|
|
|
|
|
|
|
} |
1257
|
|
|
|
|
|
|
else { |
1258
|
0
|
|
|
|
|
|
my $word = substr $wordpos, 0, $index; |
1259
|
0
|
|
|
|
|
|
my $old_pos_tag = substr $wordpos, $index + 1; |
1260
|
0
|
|
|
|
|
|
my $new_pos_tag = $wnTag{$old_pos_tag}; |
1261
|
|
|
|
|
|
|
|
1262
|
0
|
0
|
0
|
|
|
|
if ((defined $new_pos_tag) and ($new_pos_tag =~ /[nvar]/)) { |
|
|
0
|
0
|
|
|
|
|
|
|
0
|
|
|
|
|
|
1263
|
0
|
|
|
|
|
|
return $word . '#' . $new_pos_tag; |
1264
|
|
|
|
|
|
|
} |
1265
|
|
|
|
|
|
|
elsif((defined $new_pos_tag) and ($new_pos_tag =~ /[cf]/)){ |
1266
|
0
|
|
|
|
|
|
return $word . '#CL'; |
1267
|
|
|
|
|
|
|
} |
1268
|
|
|
|
|
|
|
elsif(!(defined $new_pos_tag)){ |
1269
|
0
|
|
|
|
|
|
return $word . '#IT'; |
1270
|
|
|
|
|
|
|
} |
1271
|
|
|
|
|
|
|
else { |
1272
|
0
|
|
|
|
|
|
return $word; |
1273
|
|
|
|
|
|
|
} |
1274
|
|
|
|
|
|
|
} |
1275
|
|
|
|
|
|
|
} |
1276
|
|
|
|
|
|
|
|
1277
|
|
|
|
|
|
|
|
1278
|
|
|
|
|
|
|
sub convertContraction |
1279
|
|
|
|
|
|
|
{ |
1280
|
0
|
|
|
0
|
0
|
|
my ($self, $word, $tag) = @_; |
1281
|
0
|
0
|
|
|
|
|
if ($word eq "'s") { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
1282
|
0
|
0
|
|
|
|
|
if ($tag =~ /^V/) { |
1283
|
0
|
|
|
|
|
|
return "is#v"; |
1284
|
|
|
|
|
|
|
} |
1285
|
|
|
|
|
|
|
else { |
1286
|
0
|
|
|
|
|
|
return ""; |
1287
|
|
|
|
|
|
|
} |
1288
|
|
|
|
|
|
|
} |
1289
|
|
|
|
|
|
|
elsif ($word eq "'re") { |
1290
|
0
|
|
|
|
|
|
return "are#v"; |
1291
|
|
|
|
|
|
|
} |
1292
|
|
|
|
|
|
|
elsif ($word eq "'d") { |
1293
|
0
|
|
|
|
|
|
return "had#v"; # actually this could be would as well |
1294
|
|
|
|
|
|
|
} |
1295
|
|
|
|
|
|
|
elsif ($word eq "'ll") { |
1296
|
0
|
|
|
|
|
|
return "will#v"; |
1297
|
|
|
|
|
|
|
} |
1298
|
|
|
|
|
|
|
elsif ($word eq "'em") { |
1299
|
0
|
|
|
|
|
|
return ""; |
1300
|
|
|
|
|
|
|
} |
1301
|
|
|
|
|
|
|
elsif ($word eq "'ve") { |
1302
|
0
|
|
|
|
|
|
return "have#v"; |
1303
|
|
|
|
|
|
|
} |
1304
|
|
|
|
|
|
|
elsif ($word eq "'m") { |
1305
|
0
|
|
|
|
|
|
return "am#v"; |
1306
|
|
|
|
|
|
|
} |
1307
|
|
|
|
|
|
|
elsif ($word eq "'t") { # HELP should be n't |
1308
|
0
|
|
|
|
|
|
return "not"; |
1309
|
|
|
|
|
|
|
} |
1310
|
|
|
|
|
|
|
else { |
1311
|
0
|
|
|
|
|
|
return "$word#$tag"; |
1312
|
|
|
|
|
|
|
} |
1313
|
|
|
|
|
|
|
|
1314
|
|
|
|
|
|
|
} |
1315
|
|
|
|
|
|
|
|
1316
|
|
|
|
|
|
|
# noun to non-noun ptr symbols, with frequencies |
1317
|
|
|
|
|
|
|
# -u 329 (dmnu) - cf. domn (all domains) |
1318
|
|
|
|
|
|
|
# -r 80 (dmnr) |
1319
|
|
|
|
|
|
|
# = 648 (attr) |
1320
|
|
|
|
|
|
|
# -c 2372 (dmnc) |
1321
|
|
|
|
|
|
|
# + 21390 (deri) lexical |
1322
|
|
|
|
|
|
|
|
1323
|
|
|
|
|
|
|
# verb to non-verb ptr symbols, with frequencies |
1324
|
|
|
|
|
|
|
# ;u 16 (dmtu) - cf. domt (all domains) |
1325
|
|
|
|
|
|
|
# ;c 1213 (dmtc) |
1326
|
|
|
|
|
|
|
# ;r 2 (dmtr) |
1327
|
|
|
|
|
|
|
# + 21095 (deri) lexical |
1328
|
|
|
|
|
|
|
|
1329
|
|
|
|
|
|
|
# adj to non-adj |
1330
|
|
|
|
|
|
|
# \ 4672 (pert) pertains to noun ; lexical |
1331
|
|
|
|
|
|
|
# ;u 233 |
1332
|
|
|
|
|
|
|
# ;c 1125 |
1333
|
|
|
|
|
|
|
# = 648 (attr) |
1334
|
|
|
|
|
|
|
# < 124 (part) particple of verb ; lexical |
1335
|
|
|
|
|
|
|
# ;r 76 |
1336
|
|
|
|
|
|
|
|
1337
|
|
|
|
|
|
|
# adv to non-adv |
1338
|
|
|
|
|
|
|
# \ 3208 (derived from adj) |
1339
|
|
|
|
|
|
|
# ;u 74 |
1340
|
|
|
|
|
|
|
# ;c 37 |
1341
|
|
|
|
|
|
|
# ;r 2 |
1342
|
|
|
|
|
|
|
|
1343
|
|
|
|
|
|
|
sub coercePos |
1344
|
|
|
|
|
|
|
{ |
1345
|
0
|
|
|
0
|
0
|
|
my $self = shift; |
1346
|
0
|
|
|
|
|
|
my $word = shift; |
1347
|
0
|
|
|
|
|
|
my $pos = shift; |
1348
|
0
|
|
|
|
|
|
my $wn = $wordnet{$self}; |
1349
|
|
|
|
|
|
|
|
1350
|
|
|
|
|
|
|
# remove pos tag, if present |
1351
|
0
|
|
|
|
|
|
$word =~ s/\#.*//; |
1352
|
|
|
|
|
|
|
|
1353
|
0
|
|
|
|
|
|
my @forms = $wn->validForms ($word); |
1354
|
|
|
|
|
|
|
|
1355
|
0
|
0
|
|
|
|
|
if (0 >= scalar @forms) { |
1356
|
0
|
|
|
|
|
|
return undef; |
1357
|
|
|
|
|
|
|
} |
1358
|
|
|
|
|
|
|
|
1359
|
|
|
|
|
|
|
# pre-compile the pattern |
1360
|
0
|
|
|
|
|
|
my $cpattern = qr/\#$pos/; |
1361
|
|
|
|
|
|
|
|
1362
|
0
|
|
|
|
|
|
foreach my $form (@forms) { |
1363
|
0
|
0
|
|
|
|
|
if ($form =~ /$cpattern/) { |
1364
|
0
|
|
|
|
|
|
return $form; |
1365
|
|
|
|
|
|
|
} |
1366
|
|
|
|
|
|
|
} |
1367
|
|
|
|
|
|
|
|
1368
|
|
|
|
|
|
|
# didn't find a surface match, look along cross-pos relations |
1369
|
|
|
|
|
|
|
|
1370
|
0
|
|
|
|
|
|
my @goodforms; |
1371
|
0
|
|
|
|
|
|
foreach my $form (@forms) { |
1372
|
0
|
|
|
|
|
|
my @cands = $wn->queryWord ($form, "deri"); |
1373
|
0
|
|
|
|
|
|
foreach my $candidate (@cands) { |
1374
|
0
|
0
|
|
|
|
|
if ($candidate =~ /$cpattern/) { |
1375
|
0
|
|
|
|
|
|
push @goodforms, $candidate; |
1376
|
|
|
|
|
|
|
} |
1377
|
|
|
|
|
|
|
} |
1378
|
|
|
|
|
|
|
} |
1379
|
|
|
|
|
|
|
|
1380
|
0
|
|
|
|
|
|
return @goodforms; |
1381
|
|
|
|
|
|
|
} |
1382
|
|
|
|
|
|
|
|
1383
|
|
|
|
|
|
|
# get all senses for each context word |
1384
|
|
|
|
|
|
|
sub _getSenses |
1385
|
|
|
|
|
|
|
{ |
1386
|
0
|
|
|
0
|
|
|
my $self = shift; |
1387
|
0
|
|
|
|
|
|
my $context_ref = shift; |
1388
|
0
|
|
|
|
|
|
my @senses; |
1389
|
0
|
|
|
|
|
|
for my $i (0..$#{$context_ref}){ |
|
0
|
|
|
|
|
|
|
1390
|
|
|
|
|
|
|
# first get all forms for each POS |
1391
|
0
|
0
|
|
|
|
|
if ( (${$context_ref}[$i] =~ /\#o|\#IT|\#CL|\#NT|\#MW/) ) { |
|
0
|
|
|
|
|
|
|
1392
|
0
|
|
|
|
|
|
$senses[$i] = undef; |
1393
|
|
|
|
|
|
|
} |
1394
|
|
|
|
|
|
|
else { |
1395
|
0
|
|
|
|
|
|
my @forms; |
1396
|
0
|
0
|
|
|
|
|
unless ($wnformat{$self}) { |
1397
|
0
|
|
|
|
|
|
@forms = $self->wordnet->validForms (${$context_ref}[$i]); |
|
0
|
|
|
|
|
|
|
1398
|
|
|
|
|
|
|
} |
1399
|
|
|
|
|
|
|
else { |
1400
|
0
|
|
|
|
|
|
@forms = ${$context_ref}[$i]; |
|
0
|
|
|
|
|
|
|
1401
|
|
|
|
|
|
|
} |
1402
|
0
|
0
|
|
|
|
|
if (scalar @forms == 0) { |
1403
|
0
|
|
|
|
|
|
${$context_ref}[$i]= "${$context_ref}[$i]"."#ND"; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
1404
|
|
|
|
|
|
|
} |
1405
|
|
|
|
|
|
|
else { |
1406
|
|
|
|
|
|
|
# now get all the senses for each form |
1407
|
0
|
|
|
|
|
|
foreach my $form (@forms) { |
1408
|
0
|
|
|
|
|
|
my @temps = $self->wordnet->querySense ($form); |
1409
|
0
|
|
|
|
|
|
push @{$senses[$i]}, @temps; |
|
0
|
|
|
|
|
|
|
1410
|
|
|
|
|
|
|
} |
1411
|
|
|
|
|
|
|
} |
1412
|
|
|
|
|
|
|
} |
1413
|
|
|
|
|
|
|
} |
1414
|
0
|
|
|
|
|
|
return @senses; |
1415
|
|
|
|
|
|
|
} |
1416
|
|
|
|
|
|
|
|
1417
|
|
|
|
|
|
|
sub _loadStoplist |
1418
|
|
|
|
|
|
|
{ |
1419
|
0
|
|
|
0
|
|
|
my $self = shift; |
1420
|
0
|
|
|
|
|
|
my $file = shift; |
1421
|
0
|
0
|
|
|
|
|
open SFH, '<', $file or die "Cannot open stoplist $file: $!"; |
1422
|
0
|
|
|
|
|
|
$stoplist{$self} = []; |
1423
|
0
|
|
|
|
|
|
while (my $line = ) { |
1424
|
0
|
|
|
|
|
|
chomp $line; |
1425
|
0
|
0
|
|
|
|
|
if ($line =~ m|/(.*)/|) { |
1426
|
0
|
|
|
|
|
|
push @{$stoplist{$self}}, qr/$1/; |
|
0
|
|
|
|
|
|
|
1427
|
|
|
|
|
|
|
} |
1428
|
|
|
|
|
|
|
else { |
1429
|
0
|
|
|
|
|
|
warn "Line $. of the stoplist '$file' is malformed\n"; |
1430
|
|
|
|
|
|
|
} |
1431
|
|
|
|
|
|
|
} |
1432
|
0
|
|
|
|
|
|
close SFH; |
1433
|
|
|
|
|
|
|
} |
1434
|
|
|
|
|
|
|
|
1435
|
|
|
|
|
|
|
sub getPos |
1436
|
|
|
|
|
|
|
{ |
1437
|
0
|
|
|
0
|
0
|
|
my $string = shift; |
1438
|
0
|
|
|
|
|
|
my $p = index $string, "#"; |
1439
|
0
|
0
|
|
|
|
|
return undef if $p < 0; |
1440
|
0
|
|
|
|
|
|
my $pos = substr $string, $p+1, 1; |
1441
|
0
|
|
|
|
|
|
return $pos; |
1442
|
|
|
|
|
|
|
} |
1443
|
|
|
|
|
|
|
|
1444
|
|
|
|
|
|
|
1; |
1445
|
|
|
|
|
|
|
|
1446
|
|
|
|
|
|
|
__END__ |