1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Part Of Speech Tagging |
4
|
|
|
* Brill Tagger |
5
|
|
|
* |
6
|
|
|
* @category BrillTagger |
7
|
|
|
* @author Ekin H. Bayar <[email protected]> |
8
|
|
|
* @version 0.1.0 |
9
|
|
|
*/ |
10
|
|
|
|
11
|
|
|
namespace BrillTagger; |
12
|
|
|
|
13
|
|
|
class BrillTagger |
14
|
|
|
{ |
15
|
|
|
private $dictionary = LEXICON; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* @param $text |
19
|
|
|
* @return array |
20
|
|
|
*/ |
21
|
22 |
|
public function tag($text) |
22
|
|
|
{ |
23
|
|
|
|
24
|
22 |
|
preg_match_all("/[\w\d\.'%@]+/", $text, $matches); |
25
|
|
|
|
26
|
22 |
|
$tags = []; |
27
|
22 |
|
$i = 0; |
28
|
|
|
|
29
|
22 |
|
foreach ($matches[0] as $token) { |
30
|
|
|
# default to a common noun |
31
|
22 |
|
$tags[$i] = ['token' => $token, 'tag' => 'NN']; |
32
|
|
|
|
33
|
|
|
# remove trailing full stops |
34
|
22 |
|
if (substr(trim($token), -1) == '.') { |
35
|
13 |
|
$token = preg_replace('/\.+$/', '', $token); |
36
|
|
|
} |
37
|
|
|
|
38
|
|
|
# get from dictionary if set |
39
|
22 |
|
if ($this->tokenExists($token)) { |
40
|
22 |
|
$tags[$i]['tag'] = $this->dictionary[$token][0]; |
41
|
|
|
} |
42
|
|
|
|
43
|
22 |
|
$tags[$i]['tag'] = $this->transformNumerics($tags[$i]['tag'], $token); |
44
|
|
|
|
45
|
|
|
# Anything that ends 'ly' is an adverb |
46
|
22 |
|
if ($this->isAdverb($token)) { |
47
|
1 |
|
$tags[$i]['tag'] = 'RB'; |
48
|
|
|
} |
49
|
|
|
|
50
|
22 |
|
if ($this->isNoun($tags[$i]['tag']) && !$this->isProperNoun($tags[$i]['tag'])) { |
51
|
16 |
|
$tags[$i]['tag'] = $this->transformNoun($tags[$i]['tag'], $token); |
52
|
|
|
} |
53
|
|
|
|
54
|
22 |
|
if ($i > 0) { |
55
|
13 |
|
$tags[$i]['tag'] = $this->transformNounToVerb($tags, $i, $token); |
56
|
13 |
|
$tags[$i]['tag'] = $this->transformVerbToNoun($tags, $i); |
57
|
|
|
} |
58
|
|
|
|
59
|
22 |
|
$i++; |
60
|
|
|
} |
61
|
|
|
|
62
|
22 |
|
return $tags; |
63
|
|
|
} |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* @param string $token |
67
|
|
|
* @return bool |
68
|
|
|
*/ |
69
|
23 |
|
public function tokenExists($token) |
70
|
|
|
{ |
71
|
23 |
|
return isset($this->dictionary[$token]); |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
/** |
75
|
|
|
* @param string $tag |
76
|
|
|
* @return bool |
77
|
|
|
*/ |
78
|
22 |
|
public function isNoun($tag) |
79
|
|
|
{ |
80
|
22 |
|
return substr(trim($tag), 0, 1) == 'N'; |
81
|
|
|
} |
82
|
|
|
|
83
|
|
|
/** |
84
|
|
|
* @param string $tag |
85
|
|
|
* @return bool |
86
|
|
|
*/ |
87
|
16 |
|
public function isProperNoun($tag) |
88
|
|
|
{ |
89
|
16 |
|
return substr(trim($tag), 0, 2) == 'NP'; |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* @param string $tag |
94
|
|
|
* @return bool |
95
|
|
|
*/ |
96
|
14 |
|
public function isSingularNoun($tag) |
97
|
|
|
{ |
98
|
14 |
|
return $tag == 'NN'; |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @param string $tag |
103
|
|
|
* @param string $token |
104
|
|
|
* @return bool |
105
|
|
|
*/ |
106
|
16 |
|
public function isPluralNoun($tag, $token) |
107
|
|
|
{ |
108
|
16 |
|
return ($this->isNoun($tag) && substr($token, -1) == 's'); |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* @param string $tag |
113
|
|
|
* @return bool |
114
|
|
|
*/ |
115
|
2 |
|
public function isVerb($tag) |
116
|
|
|
{ |
117
|
2 |
|
return substr(trim($tag), 0, 2) == 'VB'; |
118
|
|
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* @param string $tag |
122
|
|
|
* @return bool |
123
|
|
|
*/ |
124
|
1 |
|
public function isPronoun($tag) |
125
|
|
|
{ |
126
|
1 |
|
return substr(trim($tag), 0, 1) == 'P'; |
127
|
|
|
} |
128
|
|
|
|
129
|
|
|
/** |
130
|
|
|
* @param string $token |
131
|
|
|
* @return bool |
132
|
|
|
*/ |
133
|
4 |
|
public function isPastTenseVerb($token) |
134
|
|
|
{ |
135
|
4 |
|
return in_array('VBN', $this->dictionary[$token]); |
136
|
|
|
} |
137
|
|
|
|
138
|
|
|
/** |
139
|
|
|
* @param string $token |
140
|
|
|
* @return bool |
141
|
|
|
*/ |
142
|
4 |
|
public function isPresentTenseVerb($token) |
143
|
|
|
{ |
144
|
4 |
|
return in_array('VBZ', $this->dictionary[$token]); |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
/** it him me us you 'em thee we'uns |
148
|
|
|
* |
149
|
|
|
* @param string $tag |
150
|
|
|
* @return bool |
151
|
|
|
*/ |
152
|
1 |
|
public function isAccusativePronoun($tag) |
153
|
|
|
{ |
154
|
1 |
|
return substr(trim($tag), 0, 3) === 'PPO'; |
155
|
|
|
} |
156
|
|
|
|
157
|
|
|
/** it he she thee |
158
|
|
|
* |
159
|
|
|
* @param string $tag |
160
|
|
|
* @return bool |
161
|
|
|
*/ |
162
|
1 |
|
public function isThirdPersonPronoun($tag) |
163
|
|
|
{ |
164
|
1 |
|
return substr(trim($tag), 0, 3) === 'PPS'; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
/** they we I you ye thou you'uns |
168
|
|
|
* |
169
|
|
|
* @param string $tag |
170
|
|
|
* @return bool |
171
|
|
|
*/ |
172
|
1 |
|
public function isSingularPersonalPronoun($tag) |
173
|
|
|
{ |
174
|
1 |
|
return substr(trim($tag), 0, 4) === 'PPSS'; |
175
|
|
|
} |
176
|
|
|
|
177
|
|
|
/** itself himself myself yourself herself oneself ownself |
178
|
|
|
* |
179
|
|
|
* @param string $tag |
180
|
|
|
* @return bool |
181
|
|
|
*/ |
182
|
1 |
|
public function isSingularReflexivePronoun($tag) |
183
|
|
|
{ |
184
|
1 |
|
return substr(trim($tag), 0, 3) === 'PPL'; |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
/** themselves ourselves yourselves |
188
|
|
|
* |
189
|
|
|
* @param string $tag |
190
|
|
|
* @return bool |
191
|
|
|
*/ |
192
|
1 |
|
public function isPluralReflexivePronoun($tag) |
193
|
|
|
{ |
194
|
1 |
|
return substr(trim($tag), 0, 4) === 'PPLS'; |
195
|
|
|
} |
196
|
|
|
|
197
|
|
|
/** ours mine his her/hers their/theirs our its my your/yours out thy thine |
198
|
|
|
* |
199
|
|
|
* @param string $tag |
200
|
|
|
* @return bool |
201
|
|
|
*/ |
202
|
1 |
|
public function isPossessivePronoun($tag) |
203
|
|
|
{ |
204
|
1 |
|
return in_array($tag, ['PP$$', 'PP$']); |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
/** |
208
|
|
|
* @param string $token |
209
|
|
|
* @return bool |
210
|
|
|
*/ |
211
|
17 |
|
public function isAdjective($token) |
212
|
|
|
{ |
213
|
17 |
|
return (substr($token, -2) == 'al' || in_array('JJ', $this->dictionary[$token])); |
214
|
|
|
} |
215
|
|
|
|
216
|
|
|
/** |
217
|
|
|
* @param string $token |
218
|
|
|
* @return bool |
219
|
|
|
*/ |
220
|
17 |
|
public function isGerund($token) |
221
|
|
|
{ |
222
|
17 |
|
return substr($token, -3) == 'ing'; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* @param string $token |
227
|
|
|
* @return bool |
228
|
|
|
*/ |
229
|
17 |
|
public function isPastParticiple($token) |
230
|
|
|
{ |
231
|
17 |
|
return substr($token, -2) == 'ed'; |
232
|
|
|
} |
233
|
|
|
|
234
|
|
|
/** |
235
|
|
|
* @param string $token |
236
|
|
|
* @return bool |
237
|
|
|
*/ |
238
|
23 |
|
public function isAdverb($token) |
239
|
|
|
{ |
240
|
23 |
|
return substr($token, -2) == 'ly'; |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
/** Common noun to adj. if it ends with 'al', |
244
|
|
|
* to gerund if 'ing', to past tense if 'ed' |
245
|
|
|
* |
246
|
|
|
* @param string $tag |
247
|
|
|
* @param string $token |
248
|
|
|
* @return string |
249
|
|
|
*/ |
250
|
16 |
|
public function transformNoun($tag, $token) |
251
|
|
|
{ |
252
|
|
|
|
253
|
16 |
|
if ($this->isAdjective($token)) { |
254
|
|
|
$tag = 'JJ'; |
255
|
16 |
|
} elseif ($this->isGerund($token)) { |
256
|
1 |
|
$tag = 'VBG'; |
257
|
16 |
|
} elseif ($this->isPastParticiple($token)) { |
258
|
|
|
$tag = 'VBN'; |
259
|
16 |
|
} elseif ($token === 'I') { |
260
|
|
|
$tag = 'PPSS'; |
261
|
16 |
|
} elseif ($this->isPluralNoun($tag, $token)) { |
262
|
8 |
|
$tag = 'NNS'; |
263
|
|
|
} |
264
|
|
|
|
265
|
|
|
# Convert noun to number if . appears |
266
|
16 |
|
if (strpos($token, '.') !== false) { |
267
|
|
|
$tag = 'CD'; |
268
|
|
|
} |
269
|
|
|
|
270
|
16 |
|
return $tag; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
/** |
274
|
|
|
* @param array $tags |
275
|
|
|
* @param int $i |
276
|
|
|
* @param string $token |
277
|
|
|
* @return mixed |
278
|
|
|
*/ |
279
|
13 |
|
public function transformNounToVerb($tags, $i, $token) |
280
|
|
|
{ |
281
|
|
|
# Noun to verb if the word before is 'would' |
282
|
13 |
|
if ($this->isSingularNoun($tags[$i]['tag']) && $tags[$i - 1]['token'] == 'would') { |
283
|
|
|
$tags[$i]['tag'] = 'VB'; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
# If we get noun noun, and the 2nd can be a verb, convert to verb |
287
|
13 |
|
if ($this->isNoun($tags[$i]['tag']) && |
288
|
13 |
|
$this->isNoun($tags[$i - 1]['tag']) && |
289
|
13 |
|
$this->tokenExists($token) |
290
|
|
|
) { |
291
|
3 |
|
if ($this->isPastTenseVerb($token)) { |
292
|
|
|
$tags[$i]['tag'] = 'VBN'; |
293
|
3 |
|
} elseif ($this->isPresentTenseVerb($token)) { |
294
|
3 |
|
$tags[$i]['tag'] = 'VBZ'; |
295
|
|
|
} |
296
|
|
|
} |
297
|
|
|
|
298
|
13 |
|
return $tags[$i]['tag']; |
299
|
|
|
} |
300
|
|
|
|
301
|
|
|
/** |
302
|
|
|
* @param array $tags |
303
|
|
|
* @param int $i |
304
|
|
|
* @return mixed |
305
|
|
|
*/ |
306
|
13 |
|
public function transformVerbToNoun($tags, $i) |
307
|
|
|
{ |
308
|
|
|
# Converts verbs after 'the' to nouns |
309
|
13 |
|
if ($tags[$i - 1]['tag'] == 'DT' && $this->isVerb($tags[$i]['tag'])) { |
310
|
|
|
$tags[$i]['tag'] = 'NN'; |
311
|
|
|
} |
312
|
|
|
|
313
|
13 |
|
return $tags[$i]['tag']; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* @param string $tag |
318
|
|
|
* @param string $token |
319
|
|
|
* @return string |
320
|
|
|
*/ |
321
|
22 |
|
public function transformNumerics($tag, $token) |
322
|
|
|
{ |
323
|
|
|
# tag numerals, cardinals, money (NNS) |
324
|
22 |
|
if (preg_match(NUMERAL, $token)) { |
325
|
1 |
|
$tag = 'NNS'; |
326
|
|
|
} |
327
|
|
|
|
328
|
|
|
# tag years |
329
|
22 |
|
if (preg_match(YEAR, $token, $matches)) { |
330
|
1 |
|
$tag = (isset($matches['nns'])) ? 'NNS' : 'CD'; |
331
|
|
|
} |
332
|
|
|
|
333
|
|
|
# tag percentages |
334
|
22 |
|
if (preg_match(PERCENTAGE, $token)) { |
335
|
1 |
|
$tag = 'NN'; |
336
|
|
|
} |
337
|
|
|
|
338
|
22 |
|
return $tag; |
339
|
|
|
} |
340
|
|
|
} |
341
|
|
|
|