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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
|
// $Id$
/***********************************************************************
Moses - factored phrase-based language decoder
Copyright (C) 2006 University of Edinburgh
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
***********************************************************************/
#include <assert.h>
#include <limits>
#include <iostream>
#include <fstream>
#include "NGramNode.h"
#include "LanguageModel_Internal.h"
#include "TypeDef.h"
#include "Util.h"
#include "FactorCollection.h"
#include "Phrase.h"
#include "InputFileStream.h"
using namespace std;
// class methods
LanguageModel_Internal::LanguageModel_Internal()
{
}
void LanguageModel_Internal::Load(size_t id
, const std::string &fileName
, FactorCollection &factorCollection
, FactorType factorType
, float weight
, size_t nGramOrder)
{
m_id = id;
m_factorType = factorType;
m_weight = weight;
m_nGramOrder = nGramOrder;
// make sure start & end tags in factor collection
m_sentenceStart = factorCollection.AddFactor(Output, m_factorType, SENTENCE_START);
m_sentenceEnd = factorCollection.AddFactor(Output, m_factorType, SENTENCE_END);
// read in file
TRACE_ERR(fileName << endl);
InputFileStream inFile(fileName);
string line;
int lineNo = 0;
while( !getline(inFile, line, '\n').eof())
{
lineNo++;
if (line.size() != 0 && line.substr(0,1) != "\\")
{
vector<string> tokens = Tokenize(line, "\t");
if (tokens.size() >= 2)
{
// split unigram/bigram trigrams
vector<string> factorStr = Tokenize(tokens[1], " ");
// create / traverse down tree
NGramCollection *ngramColl = &m_map;
NGramNode *nGram;
const Factor *factor;
for (int currFactor = (int) factorStr.size() - 1 ; currFactor >= 0 ; currFactor--)
{
factor = factorCollection.AddFactor(Output, m_factorType, factorStr[currFactor]);
nGram = ngramColl->GetOrCreateNGram(factor);
ngramColl = nGram->GetNGramColl();
}
NGramNode *rootNGram = m_map.GetNGram(factor);
nGram->SetRootNGram(rootNGram);
LmId rootlmid(rootNGram);
factorCollection.SetFactorLmId(factor, rootlmid);
float score = TransformSRIScore(Scan<float>(tokens[0]));
nGram->SetScore( score );
if (tokens.size() == 3)
{
float logBackOff = TransformSRIScore(Scan<float>(tokens[2]));
nGram->SetLogBackOff( logBackOff );
}
else
{
nGram->SetLogBackOff( 0 );
}
}
}
}
}
float LanguageModel_Internal::GetValue(const std::vector<const Factor*> &contextFactor) const
{
float score;
size_t nGramOrder = contextFactor.size();
assert(nGramOrder <= 3);
if (nGramOrder == 1)
score = GetValue(contextFactor[0]);
else if (nGramOrder == 2)
score = GetValue(contextFactor[0], contextFactor[1]);
else if (nGramOrder == 3)
score = GetValue(contextFactor[0], contextFactor[1], contextFactor[2]);
return FloorSRIScore(score);
}
float LanguageModel_Internal::GetValue(const Factor *factor0) const
{
float prob;
const NGramNode *nGram = factor0->GetLmId().internal;
if (nGram == NULL || factor0->GetLmId().irst == -1)
{
prob = -numeric_limits<float>::infinity();
}
else
{
prob = nGram->GetScore();
}
return FloorSRIScore(prob);
}
float LanguageModel_Internal::GetValue(const Factor *factor0, const Factor *factor1) const
{
float score;
const NGramNode *nGram[2];
nGram[1] = factor1->GetLmId().internal;
if (nGram[1] == NULL || factor1->GetLmId().irst == -1)
{
score = -numeric_limits<float>::infinity();
}
else
{
nGram[0] = nGram[1]->GetNGram(factor0);
if (nGram[0] == NULL || factor0->GetLmId().irst == -1)
{ // something unigram
nGram[0] = factor0->GetLmId().internal;
if (nGram[0] == NULL || factor0->GetLmId().irst == -1)
{ // stops at unigram
score = nGram[1]->GetScore();
}
else
{ // unigram unigram
score = nGram[1]->GetScore() + nGram[0]->GetLogBackOff();
}
}
else
{ // bigram
score = nGram[0]->GetScore();
}
}
return FloorSRIScore(score);
}
float LanguageModel_Internal::GetValue(const Factor *factor0, const Factor *factor1, const Factor *factor2) const
{
float score;
const NGramNode *nGram[3];
nGram[2] = factor2->GetLmId().internal;
if (nGram[2] == NULL || factor2->GetLmId().irst == -1)
{
score = -numeric_limits<float>::infinity();
}
else
{
nGram[1] = nGram[2]->GetNGram(factor1);
if (nGram[1] == NULL || factor1->GetLmId().irst == -1)
{ // something unigram
nGram[1] = factor1->GetLmId().internal;
if (nGram[1] == NULL || factor1->GetLmId().irst == -1)
{ // stops at unigram
score = nGram[2]->GetScore();
}
else
{
nGram[0] = nGram[1]->GetNGram(factor0);
if (nGram[0] == NULL || factor0->GetLmId().irst == -1)
{ // unigram unigram
score = nGram[2]->GetScore() + nGram[1]->GetLogBackOff();
}
else
{ // unigram bigram
score = nGram[2]->GetScore() + nGram[1]->GetLogBackOff() + nGram[0]->GetLogBackOff();
}
}
}
else
{ // trigram, or something bigram
nGram[0] = nGram[1]->GetNGram(factor0);
if (nGram[0] != NULL)
{ // trigram
score = nGram[0]->GetScore();
}
else
{
score = nGram[1]->GetScore();
nGram[1] = nGram[1]->GetRootNGram();
nGram[0] = nGram[1]->GetNGram(factor0);
if (nGram[0] == NULL)
{ // just bigram
// do nothing
}
else
{
score += nGram[0]->GetLogBackOff();
}
}
// else do nothing. just use 1st bigram
}
}
return FloorSRIScore(score);
}
LmId LanguageModel_Internal::GetLmID( const Factor *factor ) const
{
return factor->GetLmId();
}
|
