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
|
// $Id$
/***********************************************************************
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
***********************************************************************/
#ifndef moses_TargetPhrase_h
#define moses_TargetPhrase_h
#include <vector>
#include "TypeDef.h"
#include "Phrase.h"
#include "ScoreComponentCollection.h"
#include "AlignmentInfo.h"
#include "util/string_piece.hh"
#if HAVE_CONFIG_H
#include "config.h"
#endif
#ifdef HAVE_PROTOBUF
#include "rule.pb.h"
#endif
namespace Moses
{
class LMList;
class ScoreProducer;
class TranslationSystem;
class WordPenaltyProducer;
/** represents an entry on the target side of a phrase table (scores, translation, alignment)
*/
class TargetPhrase: public Phrase
{
friend std::ostream& operator<<(std::ostream&, const TargetPhrase&);
protected:
float m_transScore;
float m_fullScore;
ScoreComponentCollection m_scoreBreakdown;
// in case of confusion net, ptr to source phrase
Phrase m_sourcePhrase;
const AlignmentInfo* m_alignmentInfo;
Word m_lhsTarget;
size_t m_ruleCount;
public:
TargetPhrase();
TargetPhrase(std::string out_string);
TargetPhrase(const Phrase &targetPhrase);
~TargetPhrase();
//! used by the unknown word handler- these targets
//! don't have a translation score, so wp is the only thing used
void SetScore(const TranslationSystem* system);
//!Set score for Sentence XML target options
void SetScore(float score);
//! Set score for unknown words with input weights
void SetScore(const TranslationSystem* system, const Scores &scoreVector);
/*** Called immediately after creation to initialize scores.
*
* @param translationScoreProducer The PhraseDictionaryMemory that this TargetPhrase is contained by.
* Used to identify where the scores for this phrase belong in the list of all scores.
* @param scoreVector the vector of scores (log probs) associated with this translation
* @param weighT the weights for the individual scores (t-weights in the .ini file)
* @param languageModels all the LanguageModels that should be used to compute the LM scores
* @param weightWP the weight of the word penalty
*
* @TODO should this be part of the constructor? If not, add explanation why not.
*/
void SetScore(const ScoreProducer* translationScoreProducer,
const Scores &scoreVector,
const ScoreComponentCollection &sparseScoreVector,
const std::vector<float> &weightT,
float weightWP,
const LMList &languageModels);
void SetScoreChart(const ScoreProducer* translationScoreProducer
,const Scores &scoreVector
,const std::vector<float> &weightT
,const LMList &languageModels
,const WordPenaltyProducer* wpProducer);
// used by for unknown word proc in chart decoding
void SetScore(const ScoreProducer* producer, const Scores &scoreVector);
// used when creating translations of unknown words:
void ResetScore();
void SetWeights(const ScoreProducer*, const std::vector<float> &weightT);
TargetPhrase *MergeNext(const TargetPhrase &targetPhrase) const;
// used for translation step
#ifdef HAVE_PROTOBUF
void WriteToRulePB(hgmert::Rule* pb) const;
#endif
/* inline float GetTranslationScore() const
{
return m_transScore;
}*/
/***
* return the estimated score resulting from our being added to a sentence
* (it's an estimate because we don't have full n-gram info for the language model
* without using the (unknown) full sentence)
*
*/
inline float GetFutureScore() const {
return m_fullScore;
}
inline const ScoreComponentCollection &GetScoreBreakdown() const
{
return m_scoreBreakdown;
}
//! TODO - why is this needed and is it set correctly by every phrase dictionary class ? should be set in constructor
// NOTE: this is not set correctly for unbinarized phrase tables
void SetSourcePhrase(const Phrase& p)
{
m_sourcePhrase=p;
}
const Phrase& GetSourcePhrase() const
{
return m_sourcePhrase;
}
void SetTargetLHS(const Word &lhs)
{ m_lhsTarget = lhs; }
const Word &GetTargetLHS() const
{ return m_lhsTarget; }
void SetAlignmentInfo(const StringPiece &alignString);
void SetAlignmentInfo(const StringPiece &alignString, Phrase &sourcePhrase);
void SetAlignmentInfo(const std::set<std::pair<size_t,size_t> > &alignmentInfo);
void SetAlignmentInfo(const std::set<std::pair<size_t,size_t> > &alignmentInfo, int* indicator);
void SetAlignmentInfo(const AlignmentInfo *alignmentInfo) {
m_alignmentInfo = alignmentInfo;
}
const AlignmentInfo &GetAlignmentInfo() const
{ return *m_alignmentInfo; }
void SetRuleCount(const StringPiece &ruleCountString, std::vector<float> &scoreVector);
size_t GetRuleCount() const { return m_ruleCount; }
TO_STRING();
};
std::ostream& operator<<(std::ostream&, const TargetPhrase&);
}
#endif
|
