|
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 $dictionary = LEXICON; |
|
16
|
|
|
|
|
17
|
12 |
|
public function tag($text) { |
|
18
|
|
|
|
|
19
|
12 |
|
preg_match_all("/[\w\d\.'%@]+/", $text, $matches); |
|
20
|
|
|
|
|
21
|
12 |
|
$tags = []; |
|
22
|
12 |
|
$i = 0; |
|
23
|
|
|
|
|
24
|
12 |
|
foreach ($matches[0] as $token) { |
|
25
|
|
|
# default to a common noun |
|
26
|
12 |
|
$tags[$i] = ['token' => $token, 'tag' => 'NN']; |
|
27
|
|
|
|
|
28
|
|
|
# remove trailing full stops |
|
29
|
12 |
|
if (substr(trim($token), -1) == '.') { |
|
30
|
12 |
|
$token = preg_replace('/\.+$/', '', $token); |
|
31
|
|
|
} |
|
32
|
|
|
|
|
33
|
|
|
# get from dictionary if set |
|
34
|
12 |
|
if (isset($this->dictionary[strtolower($token)])) { |
|
35
|
12 |
|
$tags[$i]['tag'] = $this->dictionary[strtolower($token)][0]; |
|
36
|
|
|
} |
|
37
|
|
|
|
|
38
|
|
|
# tag numerals, cardinals, money (NNS) |
|
39
|
12 |
|
if (preg_match(NUMERAL, $token)) { |
|
40
|
|
|
$tags[$i]['tag'] = 'NNS'; |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
# tag years |
|
44
|
12 |
|
if (preg_match(YEAR, $token, $matches)) { |
|
45
|
|
|
$tags[$i]['tag'] = (isset($matches['nns'])) ? 'NNS' : 'CD'; |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
# tag percentages |
|
49
|
12 |
|
if (preg_match(PERCENTAGE, $token)) { |
|
50
|
1 |
|
$tags[$i]['tag'] = 'NN'; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
# Anything that ends 'ly' is an adverb |
|
54
|
12 |
|
if ($this->isAdverb($token)) { |
|
55
|
1 |
|
$tags[$i]['tag'] = 'RB'; |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
# Common noun to adj. if it ends with 'al', to gerund if 'ing', to past tense if 'ed' |
|
59
|
12 |
|
if ($this->isNoun($tags[$i]['tag'])) { |
|
60
|
|
|
|
|
61
|
11 |
|
if ($this->isAdjective($token)) { |
|
62
|
|
|
$tags[$i]['tag'] = 'JJ'; |
|
63
|
11 |
|
} elseif ($this->isGerund($token)) { |
|
64
|
1 |
|
$tags[$i]['tag'] = 'VBG'; |
|
65
|
11 |
|
} elseif ($this->isPastParticiple($token)) { |
|
66
|
|
|
$tags[$i]['tag'] = 'VBN'; |
|
67
|
11 |
|
} elseif ($token === 'I') { |
|
68
|
4 |
|
$tags[$i]['tag'] = 'PPSS'; |
|
69
|
|
|
} |
|
70
|
|
|
# Convert noun to number if . appears |
|
71
|
11 |
|
if(strpos($token, '.') !== false) { |
|
72
|
|
|
$tags[$i]['tag'] = 'CD'; |
|
73
|
|
|
} |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
# Noun to plural if it ends with an 's' |
|
77
|
12 |
|
if ($this->isPluralNoun($tags[$i]['tag'], $token)) { |
|
78
|
4 |
|
$tags[$i]['tag'] = 'NNS'; |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
12 |
|
if ($i > 0) { |
|
82
|
|
|
|
|
83
|
|
|
# Converts verbs after 'the' to nouns |
|
84
|
12 |
|
if ($tags[$i-1]['tag'] == 'DT' && $this->isVerb($tags[$i]['tag'])) { |
|
85
|
|
|
$tags[$i]['tag'] = 'NN'; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
# Noun to verb if the word before is 'would' |
|
89
|
12 |
|
if ($this->isSingularNoun($tags[$i]['tag']) && strtolower($tags[$i-1]['token']) == 'would') { |
|
90
|
|
|
$tags[$i]['tag'] = 'VB'; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
# If we get noun noun, and the 2nd can be a verb, convert to verb |
|
94
|
12 |
|
if ($this->isNoun($tags[$i]['tag']) && |
|
95
|
12 |
|
$this->isNoun($tags[$i-1]['tag']) && |
|
96
|
12 |
|
$this->tokenExists($token) |
|
97
|
|
|
) { |
|
98
|
3 |
|
if ($this->isPastTenseVerb($token)) { |
|
99
|
|
|
$tags[$i]['tag'] = 'VBN'; |
|
100
|
3 |
|
} elseif ($this->isPresentTenseVerb($token)) { |
|
101
|
3 |
|
$tags[$i]['tag'] = 'VBZ'; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
} |
|
105
|
|
|
|
|
106
|
12 |
|
$i++; |
|
107
|
|
|
} |
|
108
|
|
|
|
|
109
|
12 |
|
return $tags; |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
3 |
|
public function tokenExists($token){ |
|
113
|
3 |
|
return isset($this->dictionary[strtolower($token)]); |
|
114
|
|
|
} |
|
115
|
|
|
|
|
116
|
12 |
|
public function isNoun($tag) { |
|
117
|
12 |
|
return substr(trim($tag), 0, 1) == 'N'; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
12 |
|
public function isSingularNoun($tag){ |
|
121
|
12 |
|
return $tag == 'NN'; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
12 |
|
public function isPluralNoun($tag, $token) { |
|
125
|
12 |
|
return ($this->isNoun($tag) && substr($token, -1) == 's'); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
1 |
|
public function isVerb($tag) { |
|
129
|
1 |
|
return substr(trim($tag), 0, 2) == 'VB'; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
public function isPronoun($tag) { |
|
133
|
|
|
return substr(trim($tag), 0, 1) == 'P'; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
3 |
|
public function isPastTenseVerb($token) { |
|
137
|
3 |
|
return in_array('VBN', $this->dictionary[strtolower($token)]); |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
3 |
|
public function isPresentTenseVerb($token) { |
|
141
|
3 |
|
return in_array('VBZ', $this->dictionary[strtolower($token)]); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
# it him me us you 'em thee we'uns |
|
145
|
|
|
public function isAccusativePronoun($tag) { |
|
146
|
|
|
return $tag === 'PPO'; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
# it he she thee |
|
150
|
|
|
public function isThirdPersonPronoun($tag) { |
|
151
|
|
|
return $tag === 'PPS'; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
|
|
# they we I you ye thou you'uns |
|
155
|
|
|
public function isSingularPersonalPronoun($tag) { |
|
156
|
|
|
return $tag === 'PPSS'; |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
# itself himself myself yourself herself oneself ownself |
|
160
|
|
|
public function isSingularReflexivePronoun($tag) { |
|
161
|
|
|
return $tag === 'PPL'; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
# themselves ourselves yourselves |
|
165
|
|
|
public function isPluralReflexivePronoun($tag) { |
|
166
|
|
|
return $tag === 'PPLS'; |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
# ours mine his her/hers their/theirs our its my your/yours out thy thine |
|
170
|
|
|
public function isPossessivePronoun($tag) { |
|
171
|
|
|
return in_array($tag,['PP$$', 'PP$']); |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
11 |
|
public function isAdjective($token) { |
|
175
|
11 |
|
return substr($token, -2) == 'al'; |
|
176
|
|
|
} |
|
177
|
|
|
|
|
178
|
11 |
|
public function isGerund($token) { |
|
179
|
11 |
|
return substr($token, -3) == 'ing'; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
11 |
|
public function isPastParticiple($token) { |
|
183
|
11 |
|
return substr($token, -2) == 'ed'; |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
12 |
|
public function isAdverb($token){ |
|
187
|
12 |
|
return substr($token, -2) == 'ly'; |
|
188
|
|
|
} |
|
189
|
|
|
} |
|
190
|
|
|
|
ekinhbayar /
BrillTagger
