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
|
// $Id$
/***********************************************************************
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 <cassert>
#include <algorithm>
#include "TargetPhrase.h"
#include "PhraseDictionaryMemory.h"
#include "GenerationDictionary.h"
#include "LanguageModel.h"
#include "StaticData.h"
#include "ScoreIndexManager.h"
#include "LMList.h"
#include "ScoreComponentCollection.h"
#include "Util.h"
using namespace std;
TargetPhrase::TargetPhrase(FactorDirection direction)
//:Phrase(direction), m_ngramScore(0.0), m_fullScore(0.0), m_sourcePhrase(0)
:Phrase(direction),m_transScore(0.0), m_ngramScore(0.0), m_fullScore(0.0), m_sourcePhrase(0)
{
}
void TargetPhrase::SetScore()
{ // used when creating translations of unknown words:
m_transScore = m_ngramScore = 0;
//m_ngramScore = 0;
m_fullScore = - StaticData::Instance().GetWeightWordPenalty();
}
void TargetPhrase::SetScore(float score)
{
//we use an existing score producer to figure out information for score setting (number of scores and weights)
//TODO: is this a good idea?
ScoreProducer* prod = StaticData::Instance().GetPhraseDictionaries()[0];
//get the weight list
unsigned int id = prod->GetScoreBookkeepingID();
const vector<float> &allWeights = StaticData::Instance().GetAllWeights();
size_t beginIndex = StaticData::Instance().GetScoreIndexManager().GetBeginIndex(id);
size_t endIndex = StaticData::Instance().GetScoreIndexManager().GetEndIndex(id);
vector<float> weights;
std::copy(allWeights.begin() +beginIndex, allWeights.begin() + endIndex,std::back_inserter(weights));
//find out how many items are in the score vector for this producer
size_t numScores = prod->GetNumScoreComponents();
//divide up the score among all of the score vectors
vector <float> scoreVector(numScores,score/numScores);
//Now we have what we need to call the full SetScore method
SetScore(prod,scoreVector,weights,StaticData::Instance().GetWeightWordPenalty(),StaticData::Instance().GetAllLM());
}
void TargetPhrase::SetScore(const ScoreProducer* translationScoreProducer,
const vector<float> &scoreVector, const vector<float> &weightT,
float weightWP, const LMList &languageModels)
{
assert(weightT.size() == scoreVector.size());
// calc average score if non-best
m_transScore = std::inner_product(scoreVector.begin(), scoreVector.end(), weightT.begin(), 0.0f);
m_scoreBreakdown.PlusEquals(translationScoreProducer, scoreVector);
// Replicated from TranslationOptions.cpp
float totalFutureScore = 0;
float totalNgramScore = 0;
float totalFullScore = 0;
LMList::const_iterator lmIter;
for (lmIter = languageModels.begin(); lmIter != languageModels.end(); ++lmIter)
{
const LanguageModel &lm = **lmIter;
if (lm.Useable(*this))
{ // contains factors used by this LM
const float weightLM = lm.GetWeight();
float fullScore, nGramScore;
lm.CalcScore(*this, fullScore, nGramScore);
m_scoreBreakdown.Assign(&lm, nGramScore);
// total LM score so far
totalNgramScore += nGramScore * weightLM;
totalFullScore += fullScore * weightLM;
}
}
m_ngramScore = totalNgramScore;
m_fullScore = m_transScore + totalFutureScore + totalFullScore
- (this->GetSize() * weightWP); // word penalty
}
void TargetPhrase::SetWeights(const ScoreProducer* translationScoreProducer, const vector<float> &weightT)
{
// calling this function in case of confusion net input is undefined
assert(StaticData::Instance().GetInputType()==SentenceInput);
/* one way to fix this, you have to make sure the weightT contains (in
addition to the usual phrase translation scaling factors) the input
weight factor as last element
*/
m_transScore = m_scoreBreakdown.PartialInnerProduct(translationScoreProducer, weightT);
}
void TargetPhrase::ResetScore()
{
m_fullScore = m_ngramScore = 0;
m_scoreBreakdown.ZeroAll();
}
TargetPhrase *TargetPhrase::MergeNext(const TargetPhrase &inputPhrase) const
{
if (! IsCompatible(inputPhrase))
{
return NULL;
}
// ok, merge
TargetPhrase *clone = new TargetPhrase(*this);
clone->m_sourcePhrase = m_sourcePhrase;
int currWord = 0;
const size_t len = GetSize();
for (size_t currPos = 0 ; currPos < len ; currPos++)
{
const Word &inputWord = inputPhrase.GetWord(currPos);
Word &cloneWord = clone->GetWord(currPos);
cloneWord.Merge(inputWord);
currWord++;
}
return clone;
}
TO_STRING_BODY(TargetPhrase);
std::ostream& operator<<(std::ostream& os, const TargetPhrase& tp)
{
os << static_cast<const Phrase&>(tp) << ", pC=" << tp.m_transScore << ", c=" << tp.m_fullScore;
//os << static_cast<const Phrase&>(tp) << ", c=" << tp.m_fullScore;
return os;
}
|
