-
Notifications
You must be signed in to change notification settings - Fork 2
/
train.go
176 lines (143 loc) · 4.42 KB
/
train.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
// Copyright 2016 The Citar Authors. All rights reserved.
// Use of this source code is governed by a BSD-style
// license that can be found in the LICENSE file.
package model
import (
"fmt"
"unicode"
"unicode/utf8"
"github.com/danieldk/conllx"
)
// A FrequencyCollector collects frequencies from the training corpus that
// are relevant to a trigram HMM tagger.
type FrequencyCollector struct {
numberer *StringNumberer
lexicon map[string]map[Tag]int
unigrams map[Unigram]int
bigrams map[Bigram]int
trigrams map[Trigram]int
}
// NewFrequencyCollector constructs a FrequencyCollector instance.
func NewFrequencyCollector() FrequencyCollector {
return FrequencyCollector{
numberer: NewStringStringNumberer(),
lexicon: make(map[string]map[Tag]int),
unigrams: make(map[Unigram]int),
bigrams: make(map[Bigram]int),
trigrams: make(map[Trigram]int),
}
}
// Model returns the collected frequencies as a model.
func (c FrequencyCollector) Model() Model {
return newModel(c.numberer, c.lexicon, c.unigrams, c.bigrams, c.trigrams,
make(ClosedClassSet))
}
// ModelWithClosedClass returns the collected frequencies as a model, the
// closed class set can be used by e.g. word handlers.
func (c FrequencyCollector) ModelWithClosedClass(closedClassTags ClosedClassSet) Model {
return newModel(c.numberer, c.lexicon, c.unigrams, c.bigrams, c.trigrams, closedClassTags)
}
// Process a sentence.
func (c FrequencyCollector) Process(sentence []conllx.Token) error {
sentence = c.addMarkers(sentence)
wordTags, err := c.sentenceToWordTags(sentence)
if err != nil {
return err
}
for i := 0; i < len(wordTags); i++ {
if err := c.addLexiconEntry(wordTags[i]); err != nil {
return err
}
c.addUnigram(wordTags[i])
if i > 0 {
c.addBigram(wordTags[i-1], wordTags[i])
}
if i > 1 {
c.addTrigram(wordTags[i-2], wordTags[i-1], wordTags[i])
}
}
return nil
}
type wordTag struct {
word string
tag uint
isUpper bool
}
func (c FrequencyCollector) sentenceToWordTags(sentence []conllx.Token) ([]wordTag, error) {
wordTags := make([]wordTag, 0, len(sentence))
for _, token := range sentence {
form, ok := token.Form()
if !ok {
return nil, fmt.Errorf("token does not contain a form: %s", token)
}
pos, ok := token.PosTag()
if !ok {
return nil, fmt.Errorf("token does not contain a part-of-speech: %s", token)
}
first, _ := utf8.DecodeRuneInString(form)
if first == utf8.RuneError {
return nil, fmt.Errorf("invalid UTF-8 character in form: %s", form)
}
wordTags = append(wordTags, wordTag{form, c.numberer.Number(pos), unicode.IsUpper(first)})
}
return wordTags, nil
}
func (c FrequencyCollector) addLexiconEntry(wordTag wordTag) error {
tagFreqs, ok := c.lexicon[wordTag.word]
if !ok {
tagFreqs = make(map[Tag]int)
c.lexicon[wordTag.word] = tagFreqs
}
tagFreqs[Tag{wordTag.tag, wordTag.isUpper}]++
return nil
}
func (c FrequencyCollector) addBigram(wordTag wordTag, wordTag2 wordTag) {
c.bigrams[Bigram{
T1: Tag{wordTag.tag, wordTag.isUpper},
T2: Tag{wordTag2.tag, wordTag2.isUpper},
}]++
}
func (c FrequencyCollector) addTrigram(wordTag wordTag, wordTag2 wordTag, wordTag3 wordTag) {
c.trigrams[Trigram{
T1: Tag{wordTag.tag, wordTag.isUpper},
T2: Tag{wordTag2.tag, wordTag2.isUpper},
T3: Tag{wordTag3.tag, wordTag3.isUpper},
}]++
}
func (c FrequencyCollector) addUnigram(wordTag wordTag) {
c.unigrams[Unigram{
T1: Tag{wordTag.tag, wordTag.isUpper},
}]++
}
func (c FrequencyCollector) addMarkers(sentence []conllx.Token) []conllx.Token {
startToken := conllx.NewToken()
startToken.SetForm(StartToken)
startToken.SetPosTag(StartToken)
endToken := conllx.NewToken()
endToken.SetForm(EndToken)
endToken.SetPosTag(EndToken)
start := []conllx.Token{*startToken, *startToken}
sentence = append(start, sentence...)
sentence = append(sentence, *endToken)
return sentence
}
func (c FrequencyCollector) addCapitalTags(sentence []conllx.Token) error {
for i := 0; i < len(sentence); i++ {
form, ok := sentence[i].Form()
pos, pok := sentence[i].PosTag()
if ok && pok {
first, _ := utf8.DecodeRuneInString(form)
if first == utf8.RuneError {
return fmt.Errorf("invalid UTF-8 character in form: %s", form)
}
if unicode.IsUpper(first) {
sentence[i].SetPosTag(fmt.Sprintf("c-%s", pos))
} else {
sentence[i].SetPosTag(fmt.Sprintf("n-%s", pos))
}
} else {
return fmt.Errorf("token does not contain form or part-of-speech: %s", sentence[i])
}
}
return nil
}