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
|
// $Id$
/***********************************************************************
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 "TranslationOptionCollectionText.h"
#include "Sentence.h"
#include "DecodeStep.h"
#include "DecodeStepTranslation.h"
#include "FactorCollection.h"
#include "Range.h"
#include <list>
#include "TranslationTask.h"
using namespace std;
namespace Moses
{
/** constructor; just initialize the base class */
TranslationOptionCollectionText::
TranslationOptionCollectionText(ttasksptr const& ttask, Sentence const &input)
//, size_t maxNoTransOptPerCoverage, float translationOptionThreshold)
: TranslationOptionCollection(ttask,input)
// , maxNoTransOptPerCoverage, translationOptionThreshold)
{
size_t maxNoTransOptPerCoverage
= ttask->options()->search.max_trans_opt_per_cov;
float translationOptionThreshold
= ttask->options()->search.trans_opt_threshold;
size_t size = input.GetSize();
m_inputPathMatrix.resize(size);
for (size_t phaseSize = 1; phaseSize <= size; ++phaseSize) {
for (size_t startPos = 0; startPos < size - phaseSize + 1; ++startPos) {
size_t endPos = startPos + phaseSize -1;
vector<InputPath*> &vec = m_inputPathMatrix[startPos];
Range range(startPos, endPos);
Phrase subphrase(input.GetSubString(Range(startPos, endPos)));
const NonTerminalSet &labels = input.GetLabelSet(startPos, endPos);
InputPath *path;
if (range.GetNumWordsCovered() == 1) {
path = new InputPath(ttask.get(), subphrase, labels, range, NULL, NULL);
vec.push_back(path);
} else {
const InputPath &prevPath = GetInputPath(startPos, endPos - 1);
path = new InputPath(ttask.get(), subphrase, labels, range, &prevPath, NULL);
vec.push_back(path);
}
m_inputPathQueue.push_back(path);
}
}
}
/* forcibly create translation option for a particular source word.
* For text, this function is easy, just call the base class' ProcessOneUnknownWord()
*/
void TranslationOptionCollectionText::ProcessUnknownWord(size_t sourcePos)
{
const InputPath &inputPath = GetInputPath(sourcePos, sourcePos);
ProcessOneUnknownWord(inputPath,sourcePos);
}
/**
* Check the source sentence for coverage data
*/
bool TranslationOptionCollectionText::HasXmlOptionsOverlappingRange(size_t startPosition, size_t endPosition) const
{
Sentence const& source=static_cast<Sentence const&>(m_source);
return source.XmlOverlap(startPosition,endPosition);
}
/**
* Check if the given translation option violates a specified xml Option
*/
bool TranslationOptionCollectionText::ViolatesXmlOptionsConstraint(size_t startPosition, size_t endPosition, TranslationOption *transOpt) const
{
// skip if there is no overlap
Sentence const& source=static_cast<Sentence const&>(m_source);
if (!source.XmlOverlap(startPosition,endPosition)) {
return false;
}
vector <TranslationOption*> xmlOptions;
source.GetXmlTranslationOptions(xmlOptions);
for(size_t i=0; i<xmlOptions.size(); i++) {
const Range &range = xmlOptions[i]->GetSourceWordsRange();
// if transOpt is a subphrase of a xml specification, do not use it
if (range.GetStartPos() <= startPosition && range.GetEndPos() >= endPosition &&
(range.GetStartPos() < startPosition || range.GetEndPos() > endPosition)) {
return true;
}
// if transOpt is partially overlapping, do not use it
if ((range.GetStartPos() < startPosition && range.GetEndPos() >= startPosition && range.GetEndPos() < endPosition) ||
(range.GetEndPos() > endPosition && range.GetStartPos() <= endPosition && range.GetStartPos() > startPosition)) {
return true;
}
// if transOpt is match or superphrase, check
if (range.GetStartPos() >= startPosition && range.GetEndPos() <= endPosition) {
const TargetPhrase &phrase = transOpt->GetTargetPhrase();
const TargetPhrase &xmlPhrase = xmlOptions[i]->GetTargetPhrase();
// if transOpt target is shorter, do not use it
if (phrase.GetSize() < xmlPhrase.GetSize()) {
return true;
}
// match may start in middle of phrase
for(size_t offset=0; offset <= phrase.GetSize()-xmlPhrase.GetSize(); offset++) {
bool match = true;
// match every word (only surface factor)
for(size_t wordPos=0; match && wordPos < xmlPhrase.GetSize(); wordPos++) {
if (phrase.GetFactor( wordPos+offset,0 )->Compare(*(xmlPhrase.GetFactor( wordPos,0 )))) {
match = false;
}
}
if (match) {
return false; // no violation if matching xml option found
}
}
return true;
}
}
return false;
}
/**
* Create xml-based translation options for the specific input span
*/
void TranslationOptionCollectionText::CreateXmlOptionsForRange(size_t startPos, size_t endPos)
{
Sentence const& source=static_cast<Sentence const&>(m_source);
InputPath &inputPath = GetInputPath(startPos,endPos);
vector <TranslationOption*> xmlOptions;
source.GetXmlTranslationOptions(xmlOptions,startPos,endPos);
//get vector of TranslationOptions from Sentence
for(size_t i=0; i<xmlOptions.size(); i++) {
TranslationOption *transOpt = xmlOptions[i];
transOpt->SetInputPath(inputPath);
Add(transOpt);
}
};
InputPath &TranslationOptionCollectionText::GetInputPath(size_t startPos, size_t endPos)
{
size_t offset = endPos - startPos;
assert(offset < m_inputPathMatrix[startPos].size());
return *m_inputPathMatrix[startPos][offset];
}
void TranslationOptionCollectionText::CreateTranslationOptions()
{
GetTargetPhraseCollectionBatch();
TranslationOptionCollection::CreateTranslationOptions();
}
/** create translation options that exactly cover a specific input span.
* Called by CreateTranslationOptions() and ProcessUnknownWord()
* \param decodeGraph list of decoding steps
* \param factorCollection input sentence with all factors
* \param startPos first position in input sentence
* \param lastPos last position in input sentence
* \param adhereTableLimit whether phrase & generation table limits are adhered to
*/
bool
TranslationOptionCollectionText::
CreateTranslationOptionsForRange
(const DecodeGraph &decodeGraph, size_t startPos, size_t endPos,
bool adhereTableLimit, size_t graphInd)
{
InputPath &inputPath = GetInputPath(startPos, endPos);
return
TranslationOptionCollection::
CreateTranslationOptionsForRange
(decodeGraph, startPos, endPos, adhereTableLimit, graphInd, inputPath);
}
}
|
