line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package RDF::NLP::SPARQLQuery::Question; |
2
|
|
|
|
|
|
|
|
3
|
11
|
|
|
11
|
|
63
|
use utf8; |
|
11
|
|
|
|
|
22
|
|
|
11
|
|
|
|
|
88
|
|
4
|
11
|
|
|
11
|
|
302
|
use strict; |
|
11
|
|
|
|
|
27
|
|
|
11
|
|
|
|
|
353
|
|
5
|
11
|
|
|
11
|
|
62
|
use warnings; |
|
11
|
|
|
|
|
22
|
|
|
11
|
|
|
|
|
381
|
|
6
|
11
|
|
|
11
|
|
56
|
use Data::Dumper; |
|
11
|
|
|
|
|
19
|
|
|
11
|
|
|
|
|
636
|
|
7
|
|
|
|
|
|
|
|
8
|
11
|
|
|
11
|
|
9190
|
use Mouse; |
|
11
|
|
|
|
|
448850
|
|
|
11
|
|
|
|
|
66
|
|
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
|
11
|
11
|
|
|
11
|
|
13805
|
use RDF::NLP::SPARQLQuery::Query; |
|
11
|
|
|
|
|
39
|
|
|
11
|
|
|
|
|
80568
|
|
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
our $VERSION='0.1'; |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
has 'docId' => (is => 'rw', isa => 'Str'); |
16
|
|
|
|
|
|
|
has 'verbose' => (is => 'rw', isa => 'Int'); |
17
|
|
|
|
|
|
|
has 'language' => (is => 'rw', isa => 'Str'); |
18
|
|
|
|
|
|
|
has 'sentences' => (is => 'rw', isa => 'ArrayRef'); |
19
|
|
|
|
|
|
|
has 'postags' => (is => 'rw', isa => 'ArrayRef'); |
20
|
|
|
|
|
|
|
has 'semanticUnits' => (is => 'rw', isa => 'ArrayRef'); |
21
|
|
|
|
|
|
|
has 'semanticCorrespondance' => (is => 'rw', isa => 'HashRef'); |
22
|
|
|
|
|
|
|
has 'aggregation' => (is => 'rw', isa => 'HashRef'); |
23
|
|
|
|
|
|
|
# has 'conjunction' => (is => 'rw', isa => 'Int'); |
24
|
|
|
|
|
|
|
has 'negation' => (is => 'rw', isa => 'HashRef'); |
25
|
|
|
|
|
|
|
# has 'varPrefix' => (is => 'rw', isa => 'Str'); |
26
|
|
|
|
|
|
|
# has 'variableCounter' => (is => 'rw', isa => 'Int'); |
27
|
|
|
|
|
|
|
# has 'variableSet' => (is => 'rw', isa => 'HashRef'); |
28
|
|
|
|
|
|
|
has 'union' => (is => 'rw', isa => 'ArrayRef'); |
29
|
|
|
|
|
|
|
has 'query' => (is => 'rw', isa => 'Object'); |
30
|
|
|
|
|
|
|
has 'questionTopic' => (is => 'rw', isa => 'Str'); |
31
|
|
|
|
|
|
|
has 'semFeaturesIndex' => (is => 'rw', isa => 'HashRef'); |
32
|
|
|
|
|
|
|
has 'config' => (is => 'rw', isa => 'HashRef'); |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
# has '' => (is => 'rw', isa => 'HashRef'); |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
# DOC |
38
|
|
|
|
|
|
|
sub new { |
39
|
10
|
|
|
10
|
1
|
22
|
my $class = shift; |
40
|
10
|
|
|
|
|
80
|
my %args = @_; |
41
|
10
|
|
|
|
|
18
|
my $i; |
42
|
|
|
|
|
|
|
|
43
|
10
|
|
|
|
|
221
|
my $Question = { |
44
|
|
|
|
|
|
|
"verbose" => 0, |
45
|
|
|
|
|
|
|
"docId" => undef, |
46
|
|
|
|
|
|
|
"language" => undef, |
47
|
|
|
|
|
|
|
"sentences" => undef, |
48
|
|
|
|
|
|
|
"postags" => undef, |
49
|
|
|
|
|
|
|
"semanticUnits" => undef, |
50
|
|
|
|
|
|
|
# "selectPart" => [], |
51
|
|
|
|
|
|
|
# "wherePart" => [], |
52
|
|
|
|
|
|
|
"sortedSemanticUnits" => undef, |
53
|
|
|
|
|
|
|
"semanticCorrespondance" => undef, |
54
|
|
|
|
|
|
|
"aggregation" => {'TERM' => { |
55
|
|
|
|
|
|
|
'count' => {}, |
56
|
|
|
|
|
|
|
'max' => {}, |
57
|
|
|
|
|
|
|
'min' => {}, |
58
|
|
|
|
|
|
|
'distinct' => {}, |
59
|
|
|
|
|
|
|
'per' => {}, |
60
|
|
|
|
|
|
|
}, |
61
|
|
|
|
|
|
|
'QT' => {}, |
62
|
|
|
|
|
|
|
'QTVAR' => {}, |
63
|
|
|
|
|
|
|
'PREDICATE' => {}, |
64
|
|
|
|
|
|
|
'ASK' => 0, |
65
|
|
|
|
|
|
|
}, |
66
|
|
|
|
|
|
|
# "conjunction" => 0, |
67
|
|
|
|
|
|
|
"negation" => {}, |
68
|
|
|
|
|
|
|
"config" => undef, |
69
|
|
|
|
|
|
|
# "varPrefix" => "?v", |
70
|
|
|
|
|
|
|
# "variableCounter" => 0, |
71
|
|
|
|
|
|
|
# "variableSet" => {}, |
72
|
|
|
|
|
|
|
"union" => [], |
73
|
|
|
|
|
|
|
"questionTopic" => undef, |
74
|
|
|
|
|
|
|
'semFeaturesIndex' => {} |
75
|
|
|
|
|
|
|
}; |
76
|
10
|
|
|
|
|
38
|
bless $Question, $class; |
77
|
|
|
|
|
|
|
|
78
|
10
|
|
|
|
|
163
|
$Question->verbose($args{'verbose'}); |
79
|
10
|
|
|
|
|
69
|
$Question->docId($args{'docId'}); |
80
|
10
|
|
|
|
|
51
|
$Question->language($args{'language'}); |
81
|
10
|
|
|
|
|
17
|
$Question->sentences([@{$args{'sentences'}}]); |
|
10
|
|
|
|
|
71
|
|
82
|
10
|
|
|
|
|
16
|
$Question->postags([@{$args{'postags'}}]); |
|
10
|
|
|
|
|
64
|
|
83
|
10
|
|
|
|
|
18
|
$Question->semanticUnits([@{$args{'semanticUnits'}}]); |
|
10
|
|
|
|
|
58
|
|
84
|
10
|
|
|
|
|
87
|
$Question->config($args{'config'}); |
85
|
|
|
|
|
|
|
|
86
|
10
|
|
|
|
|
15
|
my $term; |
87
|
10
|
|
|
|
|
17
|
foreach $term (@{$Question->semanticUnits}) { |
|
10
|
|
|
|
|
41
|
|
88
|
35
|
|
|
|
|
140
|
for($i=$term->{'start_offset'}; $i <= $term->{'end_offset'};$i++) { |
89
|
279
|
|
|
|
|
645
|
$Question->_termIndex($i, $term); |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
# warn "=> " . $term->{'semanticUnit'} . "\n"; |
92
|
|
|
|
|
|
|
# warn "\t" . join(",", keys %{$term->{'semanticTypes'}}) . "\n"; |
93
|
|
|
|
|
|
|
} |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
# foreach $term (@{$Question->sortedSemanticUnits}) { |
96
|
|
|
|
|
|
|
# warn "=> " . $term->{'semanticUnit'} . "\n"; |
97
|
|
|
|
|
|
|
|
98
|
|
|
|
|
|
|
# } |
99
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
|
101
|
10
|
|
|
|
|
56
|
return($Question); |
102
|
|
|
|
|
|
|
} |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
sub _regexForm { |
105
|
1
|
|
|
1
|
|
3
|
my $self = shift; |
106
|
|
|
|
|
|
|
|
107
|
1
|
|
|
|
|
14
|
return($self->config->{'NLQUESTION'}->{'language="' . uc($self->language) . '"'}->{'REGEXFORM'}); |
108
|
|
|
|
|
|
|
} |
109
|
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
sub _reset_sortedSemanticUnits { |
112
|
9
|
|
|
9
|
|
17
|
my $self = shift; |
113
|
|
|
|
|
|
|
|
114
|
9
|
|
|
|
|
24
|
$self->{'sortedSemanticUnits'} = undef; |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
} |
117
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
sub _sortedSemanticUnits { |
119
|
331
|
|
|
331
|
|
4657
|
my $self = shift; |
120
|
|
|
|
|
|
|
|
121
|
331
|
100
|
|
|
|
584
|
if (@_) { |
122
|
9
|
|
|
|
|
18
|
$self->{'sortedSemanticUnits'} = shift; |
123
|
|
|
|
|
|
|
} else { |
124
|
322
|
100
|
|
|
|
726
|
if (!defined $self->{'sortedSemanticUnits'}) { |
125
|
18
|
|
|
|
|
29
|
$self->{'sortedSemanticUnits'} = [sort {$a->{'start_offset'} <=> $b->{'start_offset'}} @{$self->semanticUnits}]; |
|
68
|
|
|
|
|
184
|
|
|
18
|
|
|
|
|
140
|
|
126
|
|
|
|
|
|
|
} |
127
|
|
|
|
|
|
|
} |
128
|
331
|
|
|
|
|
1497
|
return($self->{'sortedSemanticUnits'}); |
129
|
|
|
|
|
|
|
} |
130
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
sub _getTerms { |
132
|
12
|
|
|
12
|
|
13
|
my $self = shift; |
133
|
|
|
|
|
|
|
|
134
|
12
|
|
|
|
|
14
|
my $offset = shift; |
135
|
12
|
|
|
|
|
22
|
return($self->_termIndex($offset)); |
136
|
|
|
|
|
|
|
} |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
sub _termIndex { |
139
|
291
|
|
|
291
|
|
317
|
my $self = shift; |
140
|
291
|
|
|
|
|
308
|
my $offset = shift; |
141
|
|
|
|
|
|
|
|
142
|
291
|
100
|
|
|
|
468
|
if (@_) { |
143
|
279
|
|
|
|
|
287
|
my $term = shift; |
144
|
279
|
100
|
|
|
|
727
|
if (!exists $self->{'termIndex'}->{$offset}) { |
145
|
269
|
|
|
|
|
736
|
$self->{'termIndex'}->{$offset} = []; |
146
|
|
|
|
|
|
|
} |
147
|
279
|
|
|
|
|
346
|
push @{$self->{'termIndex'}->{$offset}}, $term; |
|
279
|
|
|
|
|
614
|
|
148
|
|
|
|
|
|
|
} else { |
149
|
12
|
50
|
|
|
|
28
|
if (exists $self->{'termIndex'}->{$offset}) { |
150
|
12
|
|
|
|
|
38
|
return($self->{'termIndex'}->{$offset}); |
151
|
|
|
|
|
|
|
} else { |
152
|
0
|
|
|
|
|
0
|
return(exists($self->{'termIndex'}->{$offset})); |
153
|
|
|
|
|
|
|
} |
154
|
|
|
|
|
|
|
} |
155
|
279
|
|
|
|
|
1001
|
return($self->{'termIndex'}); |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
} |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
sub _delSemanticUnit { |
160
|
0
|
|
|
0
|
|
0
|
my $self = shift; |
161
|
0
|
|
|
|
|
0
|
my $term = shift; |
162
|
0
|
|
|
|
|
0
|
my $max; |
163
|
|
|
|
|
|
|
my $i; |
164
|
0
|
|
|
|
|
0
|
my $deleted; |
165
|
|
|
|
|
|
|
# del in semanticUnits |
166
|
|
|
|
|
|
|
# $self->semanticUnits->[$term |
167
|
|
|
|
|
|
|
|
168
|
0
|
|
|
|
|
0
|
$max = scalar(@{$self->semanticUnits}); |
|
0
|
|
|
|
|
0
|
|
169
|
0
|
|
|
|
|
0
|
$i=0; |
170
|
0
|
|
|
|
|
0
|
$deleted = 0; |
171
|
0
|
|
0
|
|
|
0
|
do { |
172
|
0
|
0
|
|
|
|
0
|
if ($self->semanticUnits->[$i]->{'id'} == $term->{'id'}) { |
173
|
0
|
|
|
|
|
0
|
splice(@{$self->semanticUnits}, $i, 1); |
|
0
|
|
|
|
|
0
|
|
174
|
0
|
|
|
|
|
0
|
$deleted = 1; |
175
|
|
|
|
|
|
|
} |
176
|
0
|
|
|
|
|
0
|
$i++; |
177
|
|
|
|
|
|
|
} while(($i < $max)&&($deleted==0)); |
178
|
0
|
0
|
|
|
|
0
|
if ($deleted==0) { |
179
|
0
|
|
|
|
|
0
|
$self->_printVerbose("term (" . $term->{'id'} . ") to delete not found in SemanticUnits\n",2); |
180
|
|
|
|
|
|
|
} |
181
|
|
|
|
|
|
|
|
182
|
|
|
|
|
|
|
# del in index |
183
|
|
|
|
|
|
|
# warn "$term\n"; |
184
|
|
|
|
|
|
|
# warn "\n". $term->{'start_offset'} . "\n"; |
185
|
|
|
|
|
|
|
# warn $self->_termIndex($term->{'start_offset'}); |
186
|
0
|
|
|
|
|
0
|
$max = scalar(@{$self->_termIndex($term->{'start_offset'})}); |
|
0
|
|
|
|
|
0
|
|
187
|
0
|
|
|
|
|
0
|
$deleted = 0; |
188
|
0
|
|
|
|
|
0
|
$i=0; |
189
|
0
|
|
0
|
|
|
0
|
do { |
190
|
0
|
0
|
|
|
|
0
|
if ($self->_termIndex($term->{'start_offset'})->[$i]->{'id'} == $term->{'id'}) { |
191
|
0
|
|
|
|
|
0
|
splice(@{$self->_termIndex($term->{'start_offset'})}, $i, 1); |
|
0
|
|
|
|
|
0
|
|
192
|
0
|
|
|
|
|
0
|
$deleted=1; |
193
|
|
|
|
|
|
|
} |
194
|
0
|
|
|
|
|
0
|
$i++; |
195
|
|
|
|
|
|
|
} while(($i < $max)&&($deleted==0)); |
196
|
0
|
0
|
|
|
|
0
|
if ($deleted==0) { |
197
|
0
|
|
|
|
|
0
|
$self->_printVerbose("term (" . $term->{'id'} . ") to delete not found in index\n",2); |
198
|
|
|
|
|
|
|
} |
199
|
0
|
|
|
|
|
0
|
return(0); |
200
|
|
|
|
|
|
|
} |
201
|
|
|
|
|
|
|
|
202
|
|
|
|
|
|
|
sub _delSemanticType { |
203
|
0
|
|
|
0
|
|
0
|
my $self = shift; |
204
|
0
|
|
|
|
|
0
|
my $term = shift; |
205
|
0
|
|
|
|
|
0
|
my $semf = shift; |
206
|
|
|
|
|
|
|
|
207
|
0
|
|
|
|
|
0
|
delete $term->{'semanticTypes'}->{$semf}; |
208
|
0
|
|
|
|
|
0
|
return(0); |
209
|
|
|
|
|
|
|
} |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
sub _modifySemanticType { |
212
|
3
|
|
|
3
|
|
3
|
my $self = shift; |
213
|
3
|
|
|
|
|
4
|
my $term = shift; |
214
|
3
|
|
|
|
|
4
|
my $oldsemf = shift; |
215
|
3
|
|
|
|
|
4
|
my $newsemf = shift; |
216
|
3
|
|
|
|
|
3
|
my $semf; |
217
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
# warn $term->{'semanticUnit'}; |
219
|
|
|
|
|
|
|
# if (exists $term->{'semanticTypes'}->{$oldsemf}) { |
220
|
|
|
|
|
|
|
# warn $oldsemf; |
221
|
|
|
|
|
|
|
|
222
|
|
|
|
|
|
|
# } |
223
|
3
|
|
|
|
|
7
|
delete $term->{'semanticTypes'}->{$oldsemf}; |
224
|
3
|
|
|
|
|
8
|
foreach $semf (split /;/, $newsemf) { |
225
|
4
|
|
|
|
|
9
|
$self->_addSemanticType($term, $semf); |
226
|
|
|
|
|
|
|
# $term->{'semanticTypes'}->{$semf} = [split /\//, $semf]; |
227
|
|
|
|
|
|
|
} |
228
|
3
|
|
|
|
|
4
|
return($newsemf); |
229
|
|
|
|
|
|
|
} |
230
|
|
|
|
|
|
|
|
231
|
|
|
|
|
|
|
sub _addSemanticType { |
232
|
5
|
|
|
5
|
|
3
|
my $self = shift; |
233
|
5
|
|
|
|
|
6
|
my $term = shift; |
234
|
5
|
|
|
|
|
6
|
my $newsemf = shift; |
235
|
|
|
|
|
|
|
|
236
|
5
|
|
|
|
|
19
|
$term->{'semanticTypes'}->{$newsemf} = [split /\//, $newsemf]; |
237
|
5
|
|
|
|
|
11
|
return($newsemf); |
238
|
|
|
|
|
|
|
} |
239
|
|
|
|
|
|
|
|
240
|
|
|
|
|
|
|
# DOC |
241
|
|
|
|
|
|
|
sub Question2Query { |
242
|
9
|
|
|
9
|
1
|
19
|
my $self = shift; |
243
|
|
|
|
|
|
|
# my $format = shift; |
244
|
9
|
|
|
|
|
19
|
my $semanticCorrespondance = shift; |
245
|
|
|
|
|
|
|
# my $outStr = shift; |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
# warn $semanticCorrespondance; |
248
|
9
|
|
|
|
|
80
|
$self->semanticCorrespondance($semanticCorrespondance); |
249
|
|
|
|
|
|
|
|
250
|
9
|
|
|
|
|
244
|
$self->query(RDF::NLP::SPARQLQuery::Query->new( |
251
|
|
|
|
|
|
|
'verbose' => $self->verbose, |
252
|
|
|
|
|
|
|
'docId' => $self->docId, |
253
|
|
|
|
|
|
|
'language' => $self->language, |
254
|
|
|
|
|
|
|
'sentences' => $self->sentences, |
255
|
|
|
|
|
|
|
'negation' => $self->negation, |
256
|
|
|
|
|
|
|
'union' => $self->union, |
257
|
|
|
|
|
|
|
'aggregation' => $self->aggregation, |
258
|
|
|
|
|
|
|
'semanticCorrespondance' => $self->semanticCorrespondance, |
259
|
|
|
|
|
|
|
'semFeaturesIndex' => $self->semFeaturesIndex, |
260
|
|
|
|
|
|
|
'sortedSemanticUnits' => $self->_sortedSemanticUnits, |
261
|
|
|
|
|
|
|
'config' => $self->config, |
262
|
|
|
|
|
|
|
) |
263
|
|
|
|
|
|
|
); |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
# warn $self->query->semanticCorrespondance; |
266
|
9
|
|
|
|
|
45
|
$self->questionAbstraction; |
267
|
9
|
|
|
|
|
74
|
$self->query->queryConstruction($self->questionTopic); |
268
|
9
|
|
|
|
|
59
|
$self->query->queryGeneration; |
269
|
|
|
|
|
|
|
# $$outStr .= $self->query->queryString; |
270
|
9
|
|
|
|
|
37
|
return(1); |
271
|
|
|
|
|
|
|
} |
272
|
|
|
|
|
|
|
|
273
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
# DOC |
275
|
|
|
|
|
|
|
sub questionAbstraction { |
276
|
9
|
|
|
9
|
1
|
19
|
my $self = shift; |
277
|
|
|
|
|
|
|
|
278
|
|
|
|
|
|
|
# my %aggregation = ('TERM' => { |
279
|
|
|
|
|
|
|
# 'count' => {}, |
280
|
|
|
|
|
|
|
# 'max' => {}, |
281
|
|
|
|
|
|
|
# 'min' => {}, |
282
|
|
|
|
|
|
|
# 'distinct' => {}, |
283
|
|
|
|
|
|
|
# 'per' => {}, |
284
|
|
|
|
|
|
|
# }, |
285
|
|
|
|
|
|
|
# 'QT' => {}, |
286
|
|
|
|
|
|
|
# 'QTVAR' => {}, |
287
|
|
|
|
|
|
|
# 'PREDICATE' => {}, |
288
|
|
|
|
|
|
|
# 'ASK' => 0, |
289
|
|
|
|
|
|
|
# ); |
290
|
|
|
|
|
|
|
# my @union = (); |
291
|
|
|
|
|
|
|
# my $conjunction = 0; |
292
|
9
|
|
|
|
|
20
|
my $conjunction2 = 0; |
293
|
|
|
|
|
|
|
# my %negation = (); |
294
|
|
|
|
|
|
|
|
295
|
|
|
|
|
|
|
# my $questionTopic; |
296
|
|
|
|
|
|
|
|
297
|
|
|
|
|
|
|
# my @wherePart; |
298
|
|
|
|
|
|
|
# my @selectPart; |
299
|
9
|
|
|
|
|
18
|
my %viewedPredicates; |
300
|
|
|
|
|
|
|
my %term2semFeatures; |
301
|
|
|
|
|
|
|
# my %semFeaturesIndex; |
302
|
0
|
|
|
|
|
0
|
my %variableSet; |
303
|
|
|
|
|
|
|
|
304
|
0
|
|
|
|
|
0
|
my %lastsemf; |
305
|
0
|
|
|
|
|
0
|
my $semCat; |
306
|
0
|
|
|
|
|
0
|
my $indexCat; |
307
|
0
|
|
|
|
|
0
|
my $term; |
308
|
0
|
|
|
|
|
0
|
my $root; |
309
|
0
|
|
|
|
|
0
|
my $semf; |
310
|
0
|
|
|
|
|
0
|
my $i; |
311
|
0
|
|
|
|
|
0
|
my $aggregOp; |
312
|
|
|
|
|
|
|
|
313
|
0
|
|
|
|
|
0
|
my $line; |
314
|
|
|
|
|
|
|
|
315
|
|
|
|
|
|
|
# my @sortedSemanticUnits = (); |
316
|
9
|
|
|
|
|
39
|
$self->_removeLargerExtractedTerms; |
317
|
9
|
|
|
|
|
37
|
$self->_contextualRewriting; |
318
|
|
|
|
|
|
|
# $self->_getSortedSemanticUnits(\@sortedSemanticUnits, \$conjunction, \%negation); |
319
|
9
|
|
|
|
|
35
|
$self->_getSortedSemanticUnits; # (\$conjunction, \%negation); |
320
|
|
|
|
|
|
|
|
321
|
9
|
|
|
|
|
41
|
$self->_detecteNegation; # (\%negation); |
322
|
|
|
|
|
|
|
|
323
|
|
|
|
|
|
|
# AGGREGATION/OPERATION FUNCTION? |
324
|
9
|
|
|
|
|
37
|
$self->_identifyAggregationOperator;#(\%aggregation); |
325
|
|
|
|
|
|
|
|
326
|
|
|
|
|
|
|
# QUESTION TOPIC ? |
327
|
|
|
|
|
|
|
# $questionTopic = |
328
|
9
|
|
|
|
|
34
|
$self->_getQuestionTopic;#(\%aggregation); |
329
|
|
|
|
|
|
|
|
330
|
9
|
50
|
|
|
|
84
|
if (defined $self->questionTopic) { |
331
|
|
|
|
|
|
|
# $questionTopic = $self->semanticCorrespondance->{$lang}->{'VARIABLE'}->{$questionTopicCat}; |
332
|
9
|
|
|
|
|
54
|
$self->_printVerbose("Question Topic: " . $self->questionTopic . "\n"); |
333
|
|
|
|
|
|
|
|
334
|
|
|
|
|
|
|
# @wherePart = @{$self->query->wherePart}; |
335
|
9
|
|
|
|
|
65
|
%viewedPredicates = (); |
336
|
|
|
|
|
|
|
# @selectPart = (); |
337
|
9
|
|
|
|
|
21
|
%term2semFeatures = (); |
338
|
|
|
|
|
|
|
# %semFeaturesIndex = (); |
339
|
9
|
|
|
|
|
129
|
%variableSet = (); |
340
|
|
|
|
|
|
|
|
341
|
|
|
|
|
|
|
# Identify role of the semantic elements |
342
|
9
|
|
|
|
|
98
|
$self->_printVerbose("\n[LOG] recording semanticTypes (Property vs Predicate)\n"); |
343
|
|
|
|
|
|
|
|
344
|
9
|
|
|
|
|
16
|
for($i=0; $i < scalar(@{$self->_sortedSemanticUnits});$i++) { |
|
38
|
|
|
|
|
234
|
|
345
|
29
|
|
|
|
|
67
|
$term = $self->_sortedSemanticUnits->[$i]; |
346
|
29
|
|
|
|
|
157
|
$self->_printVerbose(">> " . $term->{'semanticUnit'} . " ($conjunction2)\n",2); |
347
|
29
|
|
|
|
|
51
|
foreach $semf (keys %{$term->{'semanticTypes'}}) { |
|
29
|
|
|
|
|
86
|
|
348
|
31
|
|
|
|
|
114
|
$self->_printVerbose(" " . $semf . "\n",2); |
349
|
31
|
100
|
|
|
|
110
|
if (!exists $term2semFeatures{$term->{'semanticUnit'}}) { |
350
|
29
|
|
|
|
|
97
|
$term2semFeatures{$term->{'semanticUnit'}} = []; |
351
|
|
|
|
|
|
|
} |
352
|
31
|
|
|
|
|
44
|
push @{$term2semFeatures{$term->{'semanticUnit'}}}, $semf; |
|
31
|
|
|
|
|
91
|
|
353
|
|
|
|
|
|
|
|
354
|
31
|
|
|
|
|
101
|
$semCat = $term->{'semanticTypes'}->{$semf}; |
355
|
|
|
|
|
|
|
# foreach $semCat (@{$semf->semantic_category}) { |
356
|
|
|
|
|
|
|
# warn join('/', @$semCat) . " ($conjunction2)\n"; |
357
|
31
|
100
|
66
|
|
|
36
|
if (scalar(@{$semCat}) > 1) { |
|
31
|
100
|
66
|
|
|
85
|
|
|
22
|
100
|
66
|
|
|
227
|
|
|
|
100
|
33
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
358
|
9
|
|
|
|
|
67
|
$self->_printVerbose(join('/', @$semCat) . " ($conjunction2)\n",2); |
359
|
9
|
100
|
66
|
|
|
200
|
if ((exists $self->semanticCorrespondance->{$self->language}->{'RESOURCE'}->{$semCat->[0]."/".$semCat->[1]}->{'CORRESP'}) && |
360
|
|
|
|
|
|
|
(exists $self->semanticCorrespondance->{$self->language}->{'RESOURCE'}->{$semCat->[0]."/".$semCat->[1]}->{'CORRESP'}->{$semCat->[$#$semCat-1]})) { |
361
|
8
|
|
|
|
|
73
|
$indexCat=$self->semanticCorrespondance->{$self->language}->{'RESOURCE'}->{$semCat->[0]."/".$semCat->[1]}->{'CORRESP'}->{$semCat->[$#$semCat-1]}; |
362
|
8
|
|
|
|
|
33
|
$root=$semCat->[0]."/".$semCat->[1]; |
363
|
|
|
|
|
|
|
} else { |
364
|
1
|
|
|
|
|
5
|
my @tmp = @$semCat; |
365
|
1
|
|
|
|
|
2
|
pop @tmp; |
366
|
1
|
|
|
|
|
4
|
$indexCat=join('/', @tmp); |
367
|
1
|
|
|
|
|
9
|
$root=$self->semanticCorrespondance->{$self->language}->{'DEFAULT_ROOT'}; |
368
|
|
|
|
|
|
|
} |
369
|
9
|
|
|
|
|
53
|
$self->_printVerbose("indexCat: $indexCat ($root)\n",2); |
370
|
9
|
50
|
|
|
|
49
|
if (!exists($self->semFeaturesIndex->{$indexCat})) { |
371
|
9
|
|
|
|
|
65
|
$self->semFeaturesIndex->{$indexCat}->{'ROOT'} = $root; |
372
|
9
|
|
|
|
|
45
|
$self->semFeaturesIndex->{$indexCat}->{'CAT'} = []; |
373
|
9
|
|
|
|
|
44
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'} = {}; |
374
|
9
|
|
|
|
|
42
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_S'} = -1; |
375
|
9
|
|
|
|
|
36
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_O'} = -1; |
376
|
|
|
|
|
|
|
} |
377
|
|
|
|
|
|
|
# warn ">> $indexCat : " . join("/", @$semCat) . "\n"; |
378
|
|
|
|
|
|
|
|
379
|
|
|
|
|
|
|
# @lastsemf=@{$document->getAnnotations->getSemanticFeaturesLevel->getElementFromIndex("refid_semantic_unit", $term->{'id'})}; |
380
|
|
|
|
|
|
|
# @lastsemf = keys %{$term->{'semanticTypes'}; |
381
|
9
|
|
|
|
|
18
|
%lastsemf = %{$term->{'semanticTypes'}}; |
|
9
|
|
|
|
|
48
|
|
382
|
9
|
|
|
|
|
19
|
push @{$self->semFeaturesIndex->{$indexCat}->{'CAT'}}, $semCat; |
|
9
|
|
|
|
|
44
|
|
383
|
9
|
|
|
|
|
52
|
$self->semFeaturesIndex->{$indexCat}->{'TERM'}->{$term->{'id'}} = $term; |
384
|
|
|
|
|
|
|
# warn " " . $semf . "\n"; |
385
|
9
|
|
|
|
|
72
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'}->{join("/", @$semCat)} = $semCat; |
386
|
|
|
|
|
|
|
#warn "=====> " . $indexCat . "\n"; |
387
|
|
|
|
|
|
|
} elsif ((scalar(@{$semCat}) == 1) && |
388
|
21
|
|
|
|
|
131
|
(exists $self->semanticCorrespondance->{$self->language}->{'CONST'}->{$semCat->[0]})) { |
389
|
1
|
|
|
|
|
5
|
$self->_printVerbose("\t\tCONST\n",2); |
390
|
1
|
|
|
|
|
3
|
$indexCat=$semCat->[0]; |
391
|
1
|
50
|
|
|
|
7
|
if (!exists($self->semFeaturesIndex->{$indexCat})) { |
392
|
1
|
|
|
|
|
6
|
$self->semFeaturesIndex->{$indexCat}->{'ROOT'} = "const"; |
393
|
1
|
|
|
|
|
6
|
$self->semFeaturesIndex->{$indexCat}->{'CAT'} = []; |
394
|
1
|
|
|
|
|
4
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'} = {}; |
395
|
1
|
|
|
|
|
3
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_S'} = -1; |
396
|
1
|
|
|
|
|
4
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_O'} = -1; |
397
|
|
|
|
|
|
|
} |
398
|
|
|
|
|
|
|
# warn ">> $indexCat : " . join("/", @$semCat) . "\n"; |
399
|
1
|
|
|
|
|
2
|
push @{$self->semFeaturesIndex->{$indexCat}->{'CAT'}}, [$term->{'semanticUnit'}]; |
|
1
|
|
|
|
|
6
|
|
400
|
1
|
|
|
|
|
6
|
$self->semFeaturesIndex->{$indexCat}->{'TERM'}->{$term->{'id'}} = $term; |
401
|
|
|
|
|
|
|
# warn " " . $semf . "\n"; |
402
|
1
|
|
|
|
|
10
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'}->{join("/", @$semCat)} = [$term->{'semanticUnit'}]; |
403
|
|
|
|
|
|
|
|
404
|
|
|
|
|
|
|
} elsif ((scalar(@{$semCat}) == 1) && |
405
|
20
|
|
|
|
|
125
|
($semCat->[0] eq "STRING")) { |
406
|
1
|
|
|
|
|
5
|
$self->_printVerbose("\t\tSTRING\n",2); |
407
|
1
|
|
|
|
|
2
|
$indexCat=$semCat->[0]; |
408
|
1
|
50
|
|
|
|
5
|
if (!exists($self->semFeaturesIndex->{$indexCat})) { |
409
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'ROOT'} = "STRING"; |
410
|
|
|
|
|
|
|
# $self->semFeaturesIndex->{$indexCat}->{'TERM'} = []; |
411
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'CAT'} = []; |
412
|
1
|
|
|
|
|
6
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'} = {}; |
413
|
1
|
|
|
|
|
4
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_S'} = -1; |
414
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_O'} = -1; |
415
|
|
|
|
|
|
|
} |
416
|
|
|
|
|
|
|
# warn ">> $indexCat : " . join("/", @$semCat) . "\n"; |
417
|
1
|
|
|
|
|
1
|
push @{$self->semFeaturesIndex->{$indexCat}->{'CAT'}}, [$term->{'canonical_form'}]; |
|
1
|
|
|
|
|
5
|
|
418
|
1
|
|
|
|
|
6
|
$self->semFeaturesIndex->{$indexCat}->{'TERM'}->{$term->{'id'}} = $term; |
419
|
|
|
|
|
|
|
# warn " " . $semf . "\n"; |
420
|
1
|
|
|
|
|
9
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'}->{join("/", @$semCat)} = [$term->{'canonical_form'}]; |
421
|
|
|
|
|
|
|
|
422
|
|
|
|
|
|
|
} elsif ((scalar(@{$semCat}) == 1) && |
423
|
19
|
|
|
|
|
177
|
($semCat->[0] eq "REGEX")) { |
424
|
1
|
|
|
|
|
3
|
$self->_printVerbose("\t\tREGEX\n",2); |
425
|
1
|
50
|
|
|
|
4
|
if ($conjunction2 eq 1) { |
426
|
0
|
|
|
|
|
0
|
my $semCat2; |
427
|
|
|
|
|
|
|
my $semf2; |
428
|
0
|
|
|
|
|
0
|
foreach $semf2 (keys %lastsemf) { |
429
|
0
|
|
|
|
|
0
|
$self->_printVerbose("\t\t\t -> " . $semf2 . "\n",2); |
430
|
0
|
|
|
|
|
0
|
$semCat2 = $lastsemf{$semf2}; # $semCat = $term->{'semanticTypes'}->{$semf}; |
431
|
|
|
|
|
|
|
# foreach $semCat2 (@{$semf2->semantic_category}) { |
432
|
0
|
|
|
|
|
0
|
$self->_printVerbose("\t\t\t---> " . join("/", @$semCat2) . "\n",2); |
433
|
0
|
|
|
|
|
0
|
my $indexCat2=$self->semanticCorrespondance->{$self->language}->{'RESOURCE'}->{$semCat2->[0]."/".$semCat2->[1]}->{'CORRESP'}->{$semCat2->[$#$semCat2-1]}; |
434
|
0
|
|
|
|
|
0
|
my @newSemCat = @$semCat2; |
435
|
0
|
0
|
|
|
|
0
|
if ($self->_regexForm == 1) { |
436
|
0
|
|
|
|
|
0
|
$newSemCat[$#newSemCat]=":NODEREGEX:".$term->{'semanticUnit'}; |
437
|
|
|
|
|
|
|
} else { |
438
|
0
|
|
|
|
|
0
|
$newSemCat[$#newSemCat]=":NODEREGEX:".$term->{'canonical_form'}; |
439
|
|
|
|
|
|
|
} |
440
|
0
|
0
|
|
|
|
0
|
if (!exists($self->semFeaturesIndex->{$indexCat2})) { |
441
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'ROOT'} = "REGEX"; |
442
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'CAT'} = []; |
443
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'CAT2'} = {}; |
444
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'SEEN_S'} = -1; |
445
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'SEEN_O'} = -1; |
446
|
|
|
|
|
|
|
} |
447
|
|
|
|
|
|
|
|
448
|
|
|
|
|
|
|
# warn ">> indexCat2: $indexCat2 : " . join("/", @newSemCat) . "\n"; |
449
|
0
|
|
|
|
|
0
|
push @{$self->semFeaturesIndex->{$indexCat2}->{'CAT'}}, [@newSemCat]; |
|
0
|
|
|
|
|
0
|
|
450
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'TERM'}->{$term->{'id'}} = $term; |
451
|
|
|
|
|
|
|
# warn " " . $lastsemf{$semf2} . "\n"; |
452
|
0
|
|
|
|
|
0
|
$self->semFeaturesIndex->{$indexCat2}->{'CAT2'}->{join("/", @newSemCat)} = [@newSemCat]; |
453
|
|
|
|
|
|
|
# } |
454
|
|
|
|
|
|
|
} |
455
|
|
|
|
|
|
|
} else { |
456
|
1
|
|
|
|
|
2
|
$indexCat="STRINGREGEX"; |
457
|
1
|
|
|
|
|
2
|
my @newSemCat; |
458
|
1
|
50
|
|
|
|
4
|
if ($self->_regexForm == 1) { |
459
|
1
|
|
|
|
|
5
|
@newSemCat = (":LABELREGEX:".$term->{'semanticUnit'}); |
460
|
|
|
|
|
|
|
} else { |
461
|
0
|
|
|
|
|
0
|
@newSemCat = (":LABELREGEX:".$term->{'canonical_form'}); |
462
|
|
|
|
|
|
|
} |
463
|
1
|
50
|
|
|
|
7
|
if (!exists($self->semFeaturesIndex->{$indexCat})) { |
464
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'ROOT'} = "REGEX"; |
465
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'CAT'} = []; |
466
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'} = {}; |
467
|
1
|
|
|
|
|
4
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_S'} = -1; |
468
|
1
|
|
|
|
|
4
|
$self->semFeaturesIndex->{$indexCat}->{'SEEN_O'} = -1; |
469
|
|
|
|
|
|
|
} |
470
|
1
|
|
|
|
|
2
|
push @{$self->semFeaturesIndex->{$indexCat}->{'CAT'}}, [@newSemCat]; |
|
1
|
|
|
|
|
6
|
|
471
|
1
|
|
|
|
|
5
|
$self->semFeaturesIndex->{$indexCat}->{'TERM'}->{$term->{'id'}} = $term; |
472
|
1
|
|
|
|
|
8
|
$self->semFeaturesIndex->{$indexCat}->{'CAT2'}->{join("/", @newSemCat)} = [@newSemCat]; |
473
|
|
|
|
|
|
|
} |
474
|
1
|
|
|
|
|
3
|
$conjunction2 = 0; |
475
|
|
|
|
|
|
|
|
476
|
|
|
|
|
|
|
} elsif ((scalar(@{$semCat}) == 1) && |
477
|
19
|
|
|
|
|
62
|
($semCat->[0] eq "conjunction")) { |
478
|
0
|
|
|
|
|
0
|
$conjunction2 = 1; |
479
|
|
|
|
|
|
|
} elsif (scalar(@{$semCat}) == 1) { |
480
|
|
|
|
|
|
|
# $self->_addPredicate('last', $self->semanticCorrespondance->{$self->language}->{'PREDICATE'}->{$indexCat}->{'NAME'}); |
481
|
|
|
|
|
|
|
# warn "Predicate: " . $semCat->[0] . "\n"; |
482
|
19
|
100
|
66
|
|
|
210
|
if ((!exists $viewedPredicates{$semCat->[0]}) && |
483
|
|
|
|
|
|
|
(exists $self->semanticCorrespondance->{$self->language}->{'PREDICATE'}->{$semCat->[0]})) { |
484
|
|
|
|
|
|
|
# warn "Predicate: " . $semCat->[0] . "\n"; |
485
|
11
|
|
|
|
|
34
|
$viewedPredicates{$semCat->[0]}++; |
486
|
11
|
|
|
|
|
83
|
$self->query->_newWherePartLine; |
487
|
11
|
|
|
|
|
88
|
$self->query->_addPredicate('last', $semCat->[0]); |
488
|
11
|
50
|
|
|
|
53
|
if (exists $self->negation->{$term->{'id'}}) { |
489
|
0
|
|
|
|
|
0
|
$self->negation->{$term->{'id'}} = scalar(@{$self->query->wherePart}) - 1; |
|
0
|
|
|
|
|
0
|
|
490
|
|
|
|
|
|
|
} |
491
|
11
|
|
|
|
|
19
|
foreach $aggregOp (keys %{$self->aggregation->{'TERM'}}) { |
|
11
|
|
|
|
|
56
|
|
492
|
55
|
50
|
|
|
|
410
|
if (exists $self->aggregation->{'TERM'}->{$aggregOp}->{$term->{'id'}}) { |
493
|
0
|
|
|
|
|
0
|
$self->aggregation->{'TERM'}->{$aggregOp}->{$term->{'id'}} = $semCat->[0]; |
494
|
0
|
|
|
|
|
0
|
$self->aggregation->{'PREDICATE'}->{$semCat->[0]} = $aggregOp; # = $term->{'id'}; |
495
|
0
|
|
|
|
|
0
|
$self->_printVerbose("aggregation ($aggregOp) for " . $term->{'semanticUnit'} . " at " . $self->aggregation->{'TERM'}->{$aggregOp}->{$term->{'id'}} . " (" . $semCat->[0] . ")\n",2); |
496
|
|
|
|
|
|
|
} |
497
|
|
|
|
|
|
|
} |
498
|
|
|
|
|
|
|
} |
499
|
|
|
|
|
|
|
# } |
500
|
|
|
|
|
|
|
} |
501
|
|
|
|
|
|
|
} |
502
|
|
|
|
|
|
|
} |
503
|
|
|
|
|
|
|
|
504
|
9
|
|
|
|
|
45
|
$self->_printVerbose("[LOG] Remove Property from semFeaturesIndex\n"); |
505
|
|
|
|
|
|
|
# remove Property from semFeaturesIndex |
506
|
9
|
|
|
|
|
13
|
foreach $line (@{$self->query->wherePart}) { |
|
9
|
|
|
|
|
92
|
|
507
|
11
|
|
|
|
|
54
|
delete $self->semFeaturesIndex->{$line->{'PREDICATE'}->[0]}; |
508
|
|
|
|
|
|
|
} |
509
|
9
|
|
|
|
|
61
|
$self->query->_internalPrintOfWherePart; # (\@wherePart); |
510
|
|
|
|
|
|
|
|
511
|
|
|
|
|
|
|
} else { |
512
|
0
|
|
|
|
|
0
|
return(-1); |
513
|
|
|
|
|
|
|
} |
514
|
9
|
|
|
|
|
46
|
return(0); |
515
|
|
|
|
|
|
|
} |
516
|
|
|
|
|
|
|
|
517
|
|
|
|
|
|
|
|
518
|
|
|
|
|
|
|
sub _removeLargerExtractedTerms { |
519
|
9
|
|
|
9
|
|
22
|
my ($self) = @_; |
520
|
|
|
|
|
|
|
|
521
|
9
|
|
|
|
|
17
|
my $term; |
522
|
|
|
|
|
|
|
my $subterm; |
523
|
0
|
|
|
|
|
0
|
my $offset; |
524
|
0
|
|
|
|
|
0
|
my $delete; |
525
|
0
|
|
|
|
|
0
|
my @termstodelete; |
526
|
|
|
|
|
|
|
|
527
|
9
|
|
|
|
|
43
|
$self->_printVerbose("[LOG] remove larger extracted terms\n"); |
528
|
|
|
|
|
|
|
|
529
|
9
|
|
|
|
|
47
|
for $term (@{$self->semanticUnits}) { |
|
9
|
|
|
|
|
40
|
|
530
|
|
|
|
|
|
|
# warn $term->{'semanticUnit'} . "\n"; |
531
|
32
|
|
|
|
|
41
|
$delete = 0; |
532
|
32
|
100
|
|
|
|
37
|
if (scalar(keys %{$term->{'semanticTypes'}}) == 0) { |
|
32
|
|
|
|
|
142
|
|
533
|
1
|
|
|
|
|
3
|
$offset = $term->{'start_offset'}; |
534
|
1
|
|
66
|
|
|
2
|
do { |
535
|
|
|
|
|
|
|
# warn "offset: $offset\n"; |
536
|
|
|
|
|
|
|
# warn "=>" . ref($self->_getTerms($offset)) . "\n"; |
537
|
6
|
50
|
|
|
|
14
|
if ($self->_getTerms($offset)) { |
538
|
6
|
|
|
|
|
8
|
foreach $subterm (@{$self->_getTerms($offset)}) { |
|
6
|
|
|
|
|
9
|
|
539
|
6
|
50
|
|
|
|
23
|
if ($subterm->{'id'} != $term->{'id'}) { |
540
|
|
|
|
|
|
|
# warn "\t" . $subterm->{'semanticUnit'} . "\n"; |
541
|
0
|
0
|
|
|
|
0
|
if (scalar(keys %{$subterm->{'semanticTypes'}}) > 0) { |
|
0
|
|
|
|
|
0
|
|
542
|
0
|
|
|
|
|
0
|
$delete = 1; |
543
|
0
|
|
|
|
|
0
|
push @termstodelete, $term; |
544
|
|
|
|
|
|
|
# warn "\t\t\t " . $term->{'semanticUnit'} . " to delete\n"; |
545
|
0
|
|
|
|
|
0
|
last; |
546
|
|
|
|
|
|
|
} |
547
|
|
|
|
|
|
|
} |
548
|
|
|
|
|
|
|
} |
549
|
|
|
|
|
|
|
} |
550
|
6
|
|
|
|
|
152
|
$offset++; |
551
|
|
|
|
|
|
|
} while(($delete == 0) && ($offset <= $term->{'end_offset'})); |
552
|
|
|
|
|
|
|
} |
553
|
|
|
|
|
|
|
|
554
|
|
|
|
|
|
|
} |
555
|
9
|
|
|
|
|
27
|
foreach $term (@termstodelete) { |
556
|
0
|
|
|
|
|
0
|
$self->_delSemanticUnit($term); |
557
|
0
|
|
|
|
|
0
|
$self->_printVerbose("remove " . $term->{'semanticUnit'} . "\n", 2); |
558
|
|
|
|
|
|
|
} |
559
|
9
|
|
|
|
|
33
|
$self->_printVerbose("done\n"); |
560
|
9
|
|
|
|
|
16
|
return(scalar(@termstodelete)); |
561
|
|
|
|
|
|
|
} |
562
|
|
|
|
|
|
|
|
563
|
|
|
|
|
|
|
sub _contextualRewriting { |
564
|
9
|
|
|
9
|
|
21
|
my ($self) = @_; |
565
|
|
|
|
|
|
|
|
566
|
9
|
|
|
|
|
20
|
my $term; |
567
|
|
|
|
|
|
|
my $semf; |
568
|
0
|
|
|
|
|
0
|
my $newsemf; |
569
|
0
|
|
|
|
|
0
|
my $semf2; |
570
|
0
|
|
|
|
|
0
|
my $newsemf2; |
571
|
0
|
|
|
|
|
0
|
my @tmpnewsemf2; |
572
|
0
|
|
|
|
|
0
|
my @tmpnewsemf; |
573
|
0
|
|
|
|
|
0
|
my $i; |
574
|
0
|
|
|
|
|
0
|
my $j; |
575
|
0
|
|
|
|
|
0
|
my $k; |
576
|
|
|
|
|
|
|
# my $lang = $self->language; |
577
|
0
|
|
|
|
|
0
|
my $rule; |
578
|
0
|
|
|
|
|
0
|
my $rules; |
579
|
0
|
|
|
|
|
0
|
my $semCat; |
580
|
|
|
|
|
|
|
|
581
|
|
|
|
|
|
|
# my @sortedSemanticUnits = sort {$a->start_token->getFrom <=> $b->start_token->getFrom} @{$document->getAnnotations->getSemanticUnitLevel->getElements}; |
582
|
|
|
|
|
|
|
|
583
|
|
|
|
|
|
|
# foreach $term (@{$self->sortedSemanticUnits}) { |
584
|
|
|
|
|
|
|
# warn "=> " . $term->{'semanticUnit'} . "\n"; |
585
|
|
|
|
|
|
|
|
586
|
|
|
|
|
|
|
# } |
587
|
|
|
|
|
|
|
|
588
|
9
|
|
|
|
|
26
|
$self->_printVerbose("[LOG] contextual rewriting\n"); |
589
|
9
|
|
|
|
|
20
|
for($i=0; $i < scalar(@{$self->_sortedSemanticUnits}); $i++) { |
|
41
|
|
|
|
|
92
|
|
590
|
|
|
|
|
|
|
# warn "\n>>> " . $self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . "\n"; |
591
|
32
|
|
|
|
|
42
|
foreach $semf (keys %{$self->_sortedSemanticUnits->[$i]->{'semanticTypes'}}) { |
|
32
|
|
|
|
|
64
|
|
592
|
|
|
|
|
|
|
# warn $semf . "\n"; |
593
|
32
|
100
|
|
|
|
204
|
if (exists $self->semanticCorrespondance->{$self->language}->{'CTXTL_REWRITE'}->{'RULE'}->{$semf}) { |
594
|
4
|
100
|
|
|
|
33
|
if (ref($self->semanticCorrespondance->{$self->language}->{'CTXTL_REWRITE'}->{'RULE'}->{$semf}) eq "HASH") { |
595
|
2
|
|
|
|
|
11
|
$rules = [$self->semanticCorrespondance->{$self->language}->{'CTXTL_REWRITE'}->{'RULE'}->{$semf}]; |
596
|
|
|
|
|
|
|
} else { |
597
|
2
|
|
|
|
|
15
|
$rules = $self->semanticCorrespondance->{$self->language}->{'CTXTL_REWRITE'}->{'RULE'}->{$semf}; |
598
|
|
|
|
|
|
|
} |
599
|
|
|
|
|
|
|
} else { |
600
|
28
|
|
|
|
|
51
|
$rules = []; |
601
|
|
|
|
|
|
|
} |
602
|
|
|
|
|
|
|
# warn "rules: $rules\n"; |
603
|
32
|
|
|
|
|
142
|
for($k=0;$k < scalar(@$rules);$k++) { |
604
|
|
|
|
|
|
|
# foreach $rule (@$rules) { |
605
|
6
|
|
|
|
|
11
|
$rule = $rules->[$k]; |
606
|
6
|
|
|
|
|
44
|
for($j=$i+1;$j < scalar(@{$self->_sortedSemanticUnits}); $j++) { |
|
21
|
|
|
|
|
43
|
|
607
|
15
|
|
|
|
|
18
|
foreach $semf2 (keys %{$self->_sortedSemanticUnits->[$j]->{'semanticTypes'}}) { |
|
15
|
|
|
|
|
24
|
|
608
|
|
|
|
|
|
|
# warn "\t" . $semf2 . " (" . $rule->{'CTXT'} . ")\n"; |
609
|
|
|
|
|
|
|
# TEST TO REWRITE with a HASH TABLE |
610
|
16
|
50
|
66
|
|
|
191
|
if (($semf2 eq $rule->{'CTXT'}) || (index($semf2, $rule->{'CTXT'} . ":") == 0) || |
|
|
|
66
|
|
|
|
|
|
|
|
33
|
|
|
|
|
611
|
|
|
|
|
|
|
(index($semf2, $rule->{'CTXT'} . "/") == 0) || |
612
|
|
|
|
|
|
|
(index($semf2, ":" . $rule->{'CTXT'}) > 0)) { |
613
|
2
|
50
|
|
|
|
8
|
if ($rule->{'NEWCTXT'} eq "") { |
|
|
50
|
|
|
|
|
|
614
|
0
|
|
|
|
|
0
|
$self->_printVerbose($self->_sortedSemanticUnits->[$j]->{'semanticUnit'} . ": del semf " . $semf2 . "\n",2); |
615
|
0
|
|
|
|
|
0
|
$self->_delSemanticType($self->_sortedSemanticUnits->[$j],$semf2); |
616
|
|
|
|
|
|
|
} elsif ($rule->{'NEWCTXT'} ne "-") { |
617
|
2
|
|
|
|
|
5
|
$self->_printVerbose($self->_sortedSemanticUnits->[$j]->{'semanticUnit'} . ": rewrite ctxt " . $semf2 . " into " . $rule->{'NEWCTXT'} . "\n",2); |
618
|
2
|
|
|
|
|
6
|
$self->_modifySemanticType($self->_sortedSemanticUnits->[$j], $semf2, $rule->{'NEWCTXT'}); |
619
|
|
|
|
|
|
|
# @tmpnewsemf2 = (); |
620
|
|
|
|
|
|
|
# foreach $newsemf2 (split /;/, $rule->{'NEWCTXT'}) { |
621
|
|
|
|
|
|
|
# push @tmpnewsemf2, [split /\//, $newsemf2]; |
622
|
|
|
|
|
|
|
# } |
623
|
|
|
|
|
|
|
# $semf2->semantic_category('list_refid_ontology_node', [@tmpnewsemf2]); |
624
|
|
|
|
|
|
|
} else { |
625
|
0
|
|
|
|
|
0
|
$self->_printVerbose($self->_sortedSemanticUnits->[$j]->{'semanticUnit'} . ": keep same ctxt (" . $semf2 . ")\n",2); |
626
|
|
|
|
|
|
|
} |
627
|
2
|
|
|
|
|
5
|
$newsemf = $rule->{'NEWSF'}; |
628
|
2
|
50
|
|
|
|
9
|
if ($newsemf eq "") { |
|
|
100
|
|
|
|
|
|
629
|
0
|
|
|
|
|
0
|
$self->_printVerbose($self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . ": del semf " . $semf . "\n",2); |
630
|
0
|
|
|
|
|
0
|
$self->_delSemanticType($self->_sortedSemanticUnits->[$i],$semf); |
631
|
|
|
|
|
|
|
} elsif ($rule->{'NEWSF'} ne "-") { |
632
|
1
|
|
|
|
|
25
|
$self->_printVerbose($self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . ": rewrite " . $semf . " into " . $rule->{'NEWSF'} . "\n\n",2); |
633
|
1
|
|
|
|
|
2
|
$self->_modifySemanticType($self->_sortedSemanticUnits->[$i], $semf, $rule->{'NEWSF'}); |
634
|
|
|
|
|
|
|
# @tmpnewsemf = (); |
635
|
|
|
|
|
|
|
|
636
|
|
|
|
|
|
|
# foreach $newsemf (split /;/, $rule->{'NEWSF'}) { |
637
|
|
|
|
|
|
|
# push @tmpnewsemf, [split /\//, $newsemf]; |
638
|
|
|
|
|
|
|
# } |
639
|
|
|
|
|
|
|
# $semf->semantic_category('list_refid_ontology_node', [@tmpnewsemf]); |
640
|
1
|
|
|
|
|
2
|
$j=scalar(@{$self->_sortedSemanticUnits}); |
|
1
|
|
|
|
|
2
|
|
641
|
1
|
|
|
|
|
1
|
$k=scalar(@$rules); |
642
|
1
|
|
|
|
|
3
|
last; |
643
|
|
|
|
|
|
|
} else { |
644
|
1
|
|
|
|
|
3
|
$self->_printVerbose($self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . ": keep same semf (" . $semf . ")\n\n",2); |
645
|
|
|
|
|
|
|
} |
646
|
|
|
|
|
|
|
} |
647
|
|
|
|
|
|
|
} |
648
|
|
|
|
|
|
|
} |
649
|
|
|
|
|
|
|
} |
650
|
|
|
|
|
|
|
} |
651
|
|
|
|
|
|
|
} |
652
|
9
|
|
|
|
|
29
|
$self->_printVerbose("done\n\n"); |
653
|
|
|
|
|
|
|
} |
654
|
|
|
|
|
|
|
|
655
|
|
|
|
|
|
|
sub _getSortedSemanticUnits { |
656
|
9
|
|
|
9
|
|
19
|
my ($self) = @_; |
657
|
|
|
|
|
|
|
|
658
|
9
|
|
|
|
|
17
|
my $semf; |
659
|
|
|
|
|
|
|
my $term; |
660
|
0
|
|
|
|
|
0
|
my @sortedSemanticUnits1; |
661
|
0
|
|
|
|
|
0
|
my @sortedSemanticUnits2; |
662
|
0
|
|
|
|
|
0
|
my @terms; |
663
|
|
|
|
|
|
|
|
664
|
9
|
|
|
|
|
27
|
$self->_printVerbose("[LOG] getSortedSemanticUnit\n"); |
665
|
|
|
|
|
|
|
|
666
|
9
|
|
|
|
|
36
|
$self->_reset_sortedSemanticUnits; |
667
|
|
|
|
|
|
|
|
668
|
|
|
|
|
|
|
# my @sortedTermList = sort {$a->start_token->getFrom <=> $b->start_token->getFrom} @{$document->getAnnotations->getSemanticUnitLevel->getElements}; |
669
|
|
|
|
|
|
|
|
670
|
|
|
|
|
|
|
|
671
|
|
|
|
|
|
|
# $document->getAnnotations->getSemanticFeaturesLevel->printIndex("refid_semantic_unit"); |
672
|
9
|
|
|
|
|
15
|
foreach $term (@{$self->_sortedSemanticUnits}) { |
|
9
|
|
|
|
|
27
|
|
673
|
32
|
|
|
|
|
179
|
$self->_printVerbose("-> " . $term->{'semanticUnit'} . " : " . $term->{'id'} . "\n", 2); |
674
|
32
|
100
|
|
|
|
62
|
if (scalar(keys %{$term->{'semanticTypes'}}) > 0) { |
|
32
|
|
|
|
|
99
|
|
675
|
|
|
|
|
|
|
# warn "ok\n"; |
676
|
31
|
|
|
|
|
52
|
push @sortedSemanticUnits1, $term; |
677
|
31
|
|
|
|
|
39
|
foreach $semf (keys %{$term->{'semanticTypes'}}) { |
|
31
|
|
|
|
|
87
|
|
678
|
|
|
|
|
|
|
# warn "==> " . $semf . " (" . $term->{'semanticUnit'} . ")\n"; |
679
|
33
|
50
|
|
|
|
146
|
if ($semf eq "conjunction") { |
680
|
0
|
|
|
|
|
0
|
$self->query->conjunction(1); |
681
|
|
|
|
|
|
|
} |
682
|
|
|
|
|
|
|
} |
683
|
|
|
|
|
|
|
} else { |
684
|
1
|
|
|
|
|
3
|
push @sortedSemanticUnits1, $term; |
685
|
|
|
|
|
|
|
# REGEX |
686
|
|
|
|
|
|
|
# warn "-> " . $term->{'semanticUnit'} . " -> REGEX\n"; |
687
|
1
|
|
|
|
|
3
|
$self->_addSemanticType($term, "REGEX"); |
688
|
|
|
|
|
|
|
# my $semFeatures = $self->_createSemanticFeaturesFromString("REGEX", $term->{'id'}); |
689
|
|
|
|
|
|
|
# if (defined $semFeatures) { |
690
|
|
|
|
|
|
|
# $document->getAnnotations->addSemanticFeatures($semFeatures); |
691
|
|
|
|
|
|
|
# } |
692
|
|
|
|
|
|
|
} |
693
|
|
|
|
|
|
|
} |
694
|
|
|
|
|
|
|
# push @$sortedSemanticUnits, @sortedSemanticUnits1; |
695
|
9
|
|
|
|
|
23
|
foreach $term (@sortedSemanticUnits1) { |
696
|
|
|
|
|
|
|
# warn "=> ". $term->{'semanticUnit'} . "\n"; |
697
|
32
|
100
|
|
|
|
74
|
if (scalar(@terms) == 0) { |
698
|
9
|
|
|
|
|
28
|
@terms = ($term); |
699
|
|
|
|
|
|
|
} else { |
700
|
23
|
100
|
|
|
|
76
|
if ($term->{'start_offset'} == $terms[$#terms]->{'start_offset'}) { |
701
|
1
|
|
|
|
|
3
|
push @terms, $term; |
702
|
|
|
|
|
|
|
} else { |
703
|
22
|
100
|
|
|
|
53
|
if (scalar(@terms) > 1) { |
704
|
|
|
|
|
|
|
# foreach my $term2 (@terms) { |
705
|
|
|
|
|
|
|
# # warn "\t" . $term2->{'semanticUnit'}. "\n"; |
706
|
|
|
|
|
|
|
# } |
707
|
1
|
|
|
|
|
4
|
push @sortedSemanticUnits2, $self->_getLargerTerm(\@terms); |
708
|
|
|
|
|
|
|
# warn "keep: " . $sortedSemanticUnits->[$#$sortedSemanticUnits]->{'semanticUnit'} . "\n"; |
709
|
|
|
|
|
|
|
} else { |
710
|
21
|
|
|
|
|
36
|
push @sortedSemanticUnits2, @terms; |
711
|
|
|
|
|
|
|
} |
712
|
22
|
|
|
|
|
55
|
@terms = ($term); |
713
|
|
|
|
|
|
|
} |
714
|
|
|
|
|
|
|
} |
715
|
|
|
|
|
|
|
} |
716
|
9
|
50
|
|
|
|
34
|
if (scalar(@terms) > 1) { |
717
|
|
|
|
|
|
|
# foreach my $term2 (@terms) { |
718
|
|
|
|
|
|
|
# warn "\t" . $term2->{'semanticUnit'}. "\n"; |
719
|
|
|
|
|
|
|
# } |
720
|
0
|
|
|
|
|
0
|
push @sortedSemanticUnits2, $self->_getLargerTerm(\@terms); |
721
|
|
|
|
|
|
|
# warn "keep: " . $sortedSemanticUnits->[$#$sortedSemanticUnits]->{'semanticUnit'} . "\n"; |
722
|
|
|
|
|
|
|
} else { |
723
|
9
|
|
|
|
|
20
|
push @sortedSemanticUnits2, @terms; |
724
|
|
|
|
|
|
|
} |
725
|
|
|
|
|
|
|
|
726
|
9
|
|
|
|
|
18
|
@terms = (); |
727
|
9
|
|
|
|
|
19
|
my @sortedSemanticUnits = (); |
728
|
9
|
|
|
|
|
30
|
foreach $term (sort {$a->{'start_offset'} <=> $b->{'start_offset'}} @sortedSemanticUnits2) { |
|
34
|
|
|
|
|
66
|
|
729
|
31
|
100
|
|
|
|
72
|
if (scalar(@terms) == 0) { |
730
|
9
|
|
|
|
|
20
|
@terms = ($term); |
731
|
|
|
|
|
|
|
} else { |
732
|
22
|
100
|
|
|
|
70
|
if ($term->{'end_offset'} <= $terms[$#terms]->{'end_offset'}) { |
733
|
2
|
|
|
|
|
9
|
push @terms, $term; |
734
|
|
|
|
|
|
|
} else { |
735
|
20
|
|
|
|
|
66
|
push @sortedSemanticUnits, $self->_getLargerTerm(\@terms); |
736
|
20
|
|
|
|
|
55
|
@terms = ($term); |
737
|
|
|
|
|
|
|
} |
738
|
|
|
|
|
|
|
} |
739
|
|
|
|
|
|
|
|
740
|
|
|
|
|
|
|
} |
741
|
9
|
50
|
|
|
|
37
|
if (scalar(@terms) > 1) { |
742
|
|
|
|
|
|
|
# foreach my $term2 (@terms) { |
743
|
|
|
|
|
|
|
# warn "\t" . $term2->{'semanticUnit'}. "\n"; |
744
|
|
|
|
|
|
|
# } |
745
|
0
|
|
|
|
|
0
|
push @sortedSemanticUnits, $self->_getLargerTerm(\@terms); |
746
|
|
|
|
|
|
|
# $self->sortedSemanticUnits([$self->_getLargerTerm(\@terms)]); |
747
|
|
|
|
|
|
|
# warn "keep: " . $sortedSemanticUnits->[$#$sortedSemanticUnits]->{'semanticUnit'} . "\n"; |
748
|
|
|
|
|
|
|
} else { |
749
|
|
|
|
|
|
|
# $self->sortedSemanticUnits([@terms]); |
750
|
9
|
|
|
|
|
20
|
push @sortedSemanticUnits, @terms; |
751
|
|
|
|
|
|
|
} |
752
|
|
|
|
|
|
|
|
753
|
9
|
|
|
|
|
58
|
$self->_sortedSemanticUnits([@sortedSemanticUnits]); |
754
|
|
|
|
|
|
|
|
755
|
|
|
|
|
|
|
# warn "---\n"; |
756
|
|
|
|
|
|
|
# foreach $term (@{$self->_sortedSemanticUnits}) { |
757
|
|
|
|
|
|
|
# warn "\t" . $term->{'semanticUnit'}. "\n"; |
758
|
|
|
|
|
|
|
# } |
759
|
|
|
|
|
|
|
|
760
|
9
|
|
|
|
|
28
|
$self->_printVerbose("done\n\n"); |
761
|
|
|
|
|
|
|
} |
762
|
|
|
|
|
|
|
|
763
|
|
|
|
|
|
|
|
764
|
|
|
|
|
|
|
sub _getLargerTerm { |
765
|
21
|
|
|
21
|
|
31
|
my ($self, $terms) = @_; |
766
|
|
|
|
|
|
|
|
767
|
21
|
|
|
|
|
29
|
my $largerTerm; |
768
|
|
|
|
|
|
|
my $tmpTerm; |
769
|
|
|
|
|
|
|
|
770
|
21
|
|
|
|
|
32
|
$largerTerm = $terms->[0]; |
771
|
|
|
|
|
|
|
|
772
|
21
|
|
|
|
|
39
|
foreach $tmpTerm (@$terms) { |
773
|
24
|
100
|
|
|
|
91
|
if ($tmpTerm->{'id'} != $largerTerm->{'id'}) { |
774
|
3
|
50
|
33
|
|
|
32
|
if (($tmpTerm->{'start_offset'} < $largerTerm->{'start_offset'} ) || |
775
|
|
|
|
|
|
|
($largerTerm->{'end_offset'} < $tmpTerm->{'end_offset'})) { |
776
|
0
|
|
|
|
|
0
|
$largerTerm = $tmpTerm |
777
|
|
|
|
|
|
|
} |
778
|
|
|
|
|
|
|
} |
779
|
|
|
|
|
|
|
} |
780
|
21
|
|
|
|
|
47
|
return($largerTerm); |
781
|
|
|
|
|
|
|
} |
782
|
|
|
|
|
|
|
|
783
|
|
|
|
|
|
|
sub _detecteNegation { |
784
|
9
|
|
|
9
|
|
17
|
my ($self) = @_; |
785
|
|
|
|
|
|
|
|
786
|
9
|
|
|
|
|
16
|
my $semf; |
787
|
|
|
|
|
|
|
my $term; |
788
|
0
|
|
|
|
|
0
|
my @terms; |
789
|
0
|
|
|
|
|
0
|
my $neg; |
790
|
|
|
|
|
|
|
|
791
|
9
|
|
|
|
|
28
|
$self->_printVerbose("[LOG] DetecteNegation\n"); |
792
|
|
|
|
|
|
|
# # warn "semf#: " . scalar (@{$document->getAnnotations->getSemanticFeaturesLevel->getElements}) . "\n"; |
793
|
|
|
|
|
|
|
# my @sortedTermList = sort {$a->start_token->getFrom <=> $b->start_token->getFrom} @{$document->getAnnotations->getSemanticUnitLevel->getElements}; |
794
|
|
|
|
|
|
|
|
795
|
9
|
|
|
|
|
16
|
$neg=0; |
796
|
9
|
|
|
|
|
18
|
foreach $term (@{$self->_sortedSemanticUnits}) { |
|
9
|
|
|
|
|
25
|
|
797
|
|
|
|
|
|
|
# warn "-> " . $term->{'semanticUnit'} . "\n"; |
798
|
29
|
50
|
|
|
|
39
|
if (scalar(keys %{$term->{'semanticTypes'}}) > 0) { |
|
29
|
|
|
|
|
95
|
|
799
|
29
|
|
|
|
|
38
|
foreach $semf (keys %{$term->{'semanticTypes'}}) { |
|
29
|
|
|
|
|
80
|
|
800
|
|
|
|
|
|
|
# warn "==> " . $semf . " (" . $term->{'semanticUnit'} . ")\n"; |
801
|
|
|
|
|
|
|
# foreach $semCat (@{$term->{'semanticTypes'}-{$semf}}) { |
802
|
|
|
|
|
|
|
# warn "\t" . join('/', @$semCat) . "\n"; |
803
|
31
|
100
|
|
|
|
184
|
if (exists $self->semanticCorrespondance->{$self->language}->{'PREDICATE'}->{$semf}) { |
804
|
11
|
50
|
|
|
|
89
|
if ($neg == 1) { |
805
|
0
|
|
|
|
|
0
|
$self->negation->{$term->{'id'}}=""; |
806
|
0
|
|
|
|
|
0
|
$self->_printVerbose($term->{'semanticUnit'} . " is negated\n",2); |
807
|
0
|
|
|
|
|
0
|
$neg=0; |
808
|
|
|
|
|
|
|
} |
809
|
|
|
|
|
|
|
} |
810
|
31
|
50
|
|
|
|
140
|
if ($semf eq "negation") { |
811
|
0
|
|
|
|
|
0
|
$self->_printVerbose("found negation\n",2); |
812
|
0
|
|
|
|
|
0
|
$neg=1; |
813
|
|
|
|
|
|
|
} |
814
|
|
|
|
|
|
|
# } |
815
|
|
|
|
|
|
|
} |
816
|
|
|
|
|
|
|
} |
817
|
|
|
|
|
|
|
} |
818
|
|
|
|
|
|
|
} |
819
|
|
|
|
|
|
|
|
820
|
|
|
|
|
|
|
sub _identifyAggregationOperator { |
821
|
9
|
|
|
9
|
|
21
|
my ($self) = @_; |
822
|
|
|
|
|
|
|
|
823
|
9
|
|
|
|
|
17
|
my $semf; |
824
|
|
|
|
|
|
|
my $term; |
825
|
0
|
|
|
|
|
0
|
my @sortedSemanticUnits1; |
826
|
0
|
|
|
|
|
0
|
my @terms; |
827
|
9
|
|
|
|
|
17
|
my @aggregOp = (); |
828
|
9
|
|
|
|
|
16
|
my $op; |
829
|
|
|
|
|
|
|
|
830
|
9
|
|
|
|
|
25
|
$self->_printVerbose("[LOG] _identifyAggregationOperator\n"); |
831
|
9
|
|
|
|
|
17
|
foreach $term (@{$self->_sortedSemanticUnits}) { |
|
9
|
|
|
|
|
25
|
|
832
|
29
|
50
|
|
|
|
33
|
if (scalar(keys %{$term->{'semanticTypes'}}) > 0) { |
|
29
|
|
|
|
|
94
|
|
833
|
29
|
|
|
|
|
96
|
foreach $semf (keys %{$term->{'semanticTypes'}}) { |
|
29
|
|
|
|
|
71
|
|
834
|
31
|
|
|
|
|
133
|
$self->_printVerbose("==> " . $semf . " (" . $term->{'semanticUnit'} . ")\n",2); |
835
|
|
|
|
|
|
|
# warn "$semf: " . $term->{'semanticTypes'}->{$semf} . "\n"; |
836
|
|
|
|
|
|
|
# foreach $semCat (@{$term->{'semanticTypes'}->{$semf}}) { |
837
|
31
|
50
|
|
|
|
137
|
if (exists $self->aggregation->{'TERM'}->{$semf}) { |
838
|
0
|
|
|
|
|
0
|
push @aggregOp, $semf; |
839
|
|
|
|
|
|
|
} |
840
|
|
|
|
|
|
|
|
841
|
31
|
100
|
|
|
|
214
|
if (exists $self->semanticCorrespondance->{$self->language}->{'VARIABLE'}->{$semf}) { |
842
|
19
|
50
|
|
|
|
58
|
if (scalar(@aggregOp) != 0) { |
843
|
0
|
|
|
|
|
0
|
foreach $op (@aggregOp) { |
844
|
0
|
|
|
|
|
0
|
$self->aggregation->{'TERM'}->{$op}->{$term->{'id'}}=""; |
845
|
0
|
|
|
|
|
0
|
$self->_printVerbose($term->{'semanticUnit'} . " is aggregated ($op)\n",2); |
846
|
|
|
|
|
|
|
} |
847
|
0
|
|
|
|
|
0
|
@aggregOp = (); |
848
|
|
|
|
|
|
|
} |
849
|
|
|
|
|
|
|
} |
850
|
31
|
50
|
|
|
|
130
|
if ($semf eq "exists") { |
851
|
0
|
|
|
|
|
0
|
$self->aggregation->{'ASK'} = 1; |
852
|
|
|
|
|
|
|
} |
853
|
|
|
|
|
|
|
# } |
854
|
|
|
|
|
|
|
} |
855
|
|
|
|
|
|
|
} |
856
|
|
|
|
|
|
|
} |
857
|
|
|
|
|
|
|
} |
858
|
|
|
|
|
|
|
|
859
|
|
|
|
|
|
|
sub _getQuestionTopic { |
860
|
9
|
|
|
9
|
|
17
|
my ($self) = @_; |
861
|
|
|
|
|
|
|
|
862
|
9
|
|
|
|
|
18
|
my $semf; |
863
|
9
|
|
|
|
|
15
|
my $found = 0; |
864
|
|
|
|
|
|
|
# my $questionTopic; |
865
|
9
|
|
|
|
|
16
|
my $questionTopicCat; |
866
|
|
|
|
|
|
|
my $i; |
867
|
0
|
|
|
|
|
0
|
my $aggregOp; |
868
|
|
|
|
|
|
|
|
869
|
|
|
|
|
|
|
# warn scalar(@{$self->_sortedSemanticUnits}) . "\n"; |
870
|
9
|
|
|
|
|
19
|
$i = 0; |
871
|
9
|
|
|
|
|
29
|
do { |
872
|
9
|
50
|
|
|
|
16
|
if ($i < scalar(@{$self->_sortedSemanticUnits})) { |
|
9
|
|
|
|
|
22
|
|
873
|
9
|
|
|
|
|
28
|
$self->_printVerbose("QT? " . $self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . "\n",2); |
874
|
9
|
|
|
|
|
19
|
foreach $semf (keys %{$self->_sortedSemanticUnits->[$i]->{'semanticTypes'}}) { |
|
9
|
|
|
|
|
55
|
|
875
|
10
|
|
|
|
|
43
|
$self->_printVerbose($semf . "\n",2); |
876
|
|
|
|
|
|
|
# foreach $semCat (@{$self->_sortedSemanticUnits->[$i]->{'semanticTypes'}->{$semf}}) { |
877
|
|
|
|
|
|
|
# $self->_printVerbose($self->language . "\n",2); |
878
|
|
|
|
|
|
|
# warn "\t" . join('/', @$semCat) . "\n"; |
879
|
10
|
|
|
|
|
19
|
$questionTopicCat = join('/', @{$self->_sortedSemanticUnits->[$i]->{'semanticTypes'}->{$semf}}); |
|
10
|
|
|
|
|
29
|
|
880
|
10
|
|
|
|
|
52
|
$self->_printVerbose("questionTopicCat $questionTopicCat\n", 2); |
881
|
10
|
100
|
|
|
|
85
|
if (exists $self->semanticCorrespondance->{$self->language}->{'VARIABLE'}->{$questionTopicCat}) { |
882
|
9
|
|
|
|
|
152
|
$self->questionTopic($self->semanticCorrespondance->{$self->language}->{'VARIABLE'}->{$questionTopicCat}); |
883
|
9
|
|
|
|
|
67
|
$self->_printVerbose("Question Topic: " . $questionTopicCat . " : " . $self->questionTopic . "\n",2); |
884
|
9
|
|
|
|
|
19
|
$found = 1; |
885
|
|
|
|
|
|
|
|
886
|
9
|
|
|
|
|
19
|
foreach $aggregOp (keys %{$self->aggregation->{'TERM'}}) { |
|
9
|
|
|
|
|
57
|
|
887
|
45
|
50
|
|
|
|
160
|
if (exists $self->aggregation->{'TERM'}->{$aggregOp}->{$self->_sortedSemanticUnits->[$i]->{'id'}}) { |
888
|
0
|
|
|
|
|
0
|
$self->aggregation->{'TERM'}->{$aggregOp}->{$self->_sortedSemanticUnits->[$i]->{'id'}} = $self->questionTopic; |
889
|
0
|
|
|
|
|
0
|
$self->aggregation->{'QT'}->{$aggregOp}->{$self->questionTopic} = $self->_sortedSemanticUnits->[$i]->{'id'}; |
890
|
0
|
|
|
|
|
0
|
$self->_printVerbose("aggregation ($aggregOp) for " . $self->_sortedSemanticUnits->[$i]->{'semanticUnit'} . " at " . $self->aggregation->{'TERM'}->{$aggregOp}->{$self->_sortedSemanticUnits->[$i]->{'id'}} . " (" . $self->questionTopic . ")\n",2); |
891
|
|
|
|
|
|
|
} |
892
|
|
|
|
|
|
|
} |
893
|
|
|
|
|
|
|
# } |
894
|
|
|
|
|
|
|
} else { |
895
|
|
|
|
|
|
|
# warn "no\n"; |
896
|
|
|
|
|
|
|
} |
897
|
|
|
|
|
|
|
} |
898
|
9
|
|
|
|
|
32
|
$i++; |
899
|
|
|
|
|
|
|
} |
900
|
9
|
|
33
|
|
|
17
|
} while(($i < scalar(@{$self->_sortedSemanticUnits})) && (!$found)); |
901
|
|
|
|
|
|
|
|
902
|
9
|
|
|
|
|
56
|
$self->_printVerbose("questionTopic: " . $self->questionTopic . "\n", 2); |
903
|
|
|
|
|
|
|
|
904
|
9
|
|
|
|
|
1018
|
return($self->questionTopic); |
905
|
|
|
|
|
|
|
} |
906
|
|
|
|
|
|
|
|
907
|
|
|
|
|
|
|
sub _printVerbose { |
908
|
294
|
|
|
294
|
|
457
|
my($self, $msg, $level) = @_; |
909
|
|
|
|
|
|
|
|
910
|
294
|
100
|
|
|
|
586
|
if (!defined $level) { |
911
|
99
|
|
|
|
|
123
|
$level = 1; |
912
|
|
|
|
|
|
|
} |
913
|
|
|
|
|
|
|
|
914
|
294
|
50
|
33
|
|
|
1354
|
if (($self->verbose > 0) && ($self->verbose >= $level)) { |
915
|
0
|
|
|
|
|
|
warn "$msg"; |
916
|
|
|
|
|
|
|
} |
917
|
|
|
|
|
|
|
|
918
|
|
|
|
|
|
|
} |
919
|
|
|
|
|
|
|
|
920
|
|
|
|
|
|
|
1; |
921
|
|
|
|
|
|
|
|
922
|
|
|
|
|
|
|
__END__ |