moses-smt/mosesdecoder
1
1
Fork 0
mirror of https://github.com/moses-smt/mosesdecoder.git synced 2024-12-26 13:23:25 +03:00
Code Issues Packages Projects Releases Wiki Activity

segment confusion network in class TranslationOptionCollectionConfusionNet, rather than in PDAImp. Get ready to make it work with any phrase table

Browse Source
This commit is contained in:
Hieu Hoang 2013-06-14 23:42:30 +01:00
parent 3516ef42b8
commit 8c6344c4ce
4 changed files with 76 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

73
moses/TranslationOptionCollectionConfusionNet.cpp
View File

@ -60,11 +60,16 @@ TranslationOptionCollectionConfusionNet::TranslationOptionCollectionConfusionNet
// create subphrases by appending words to previous subphrases
for (size_t startPos = 0; startPos < size; ++startPos) {
for (size_t endPos = startPos + 1; endPos < size; ++endPos) {
std::vector<SourcePath> &newSubphrases = GetPhrases(startPos, endPos);
const std::vector<SourcePath> &prevSubphrases = GetPhrases(startPos, endPos - 1);
const ConfusionNet::Column &col = input.GetColumn(endPos);
CreateSubPhrases(newSubphrases, prevSubphrases, col, *inputFeature);
size_t maxSize = size - startPos;
size_t maxSizePhrase = StaticData::Instance().GetMaxPhraseLength();
maxSize = std::min(maxSize, maxSizePhrase);
size_t end = startPos + maxSize - 1;
for (size_t endPos = startPos + 1; endPos < end; ++endPos) {
std::vector<SourcePath> &newSubphrases = GetPhrases(startPos, endPos);
const std::vector<SourcePath> &prevSubphrases = GetPhrases(startPos, endPos - 1);
const ConfusionNet::Column &col = input.GetColumn(endPos);
CreateSubPhrases(newSubphrases, prevSubphrases, col, *inputFeature);
}
}
@ -85,29 +90,29 @@ TranslationOptionCollectionConfusionNet::TranslationOptionCollectionConfusionNet
}
void TranslationOptionCollectionConfusionNet::CreateSubPhrases(std::vector<SourcePath> &newSubphrases
, const std::vector<SourcePath> &prevSubphrases
, const ConfusionNet::Column &col
, const InputFeature &inputFeature)
, const std::vector<SourcePath> &prevSubphrases
, const ConfusionNet::Column &col
, const InputFeature &inputFeature)
{
std::vector<SourcePath>::const_iterator iterSourcePath;
for (iterSourcePath = prevSubphrases.begin(); iterSourcePath != prevSubphrases.end(); ++iterSourcePath) {
const SourcePath &sourcePath = *iterSourcePath;
const Phrase &prevSubPhrase = sourcePath.first;
const ScoreComponentCollection &prevScore = sourcePath.second;
std::vector<SourcePath>::const_iterator iterSourcePath;
for (iterSourcePath = prevSubphrases.begin(); iterSourcePath != prevSubphrases.end(); ++iterSourcePath) {
const SourcePath &sourcePath = *iterSourcePath;
const Phrase &prevSubPhrase = sourcePath.first;
const ScoreComponentCollection &prevScore = sourcePath.second;
ConfusionNet::Column::const_iterator iterCol;
for (iterCol = col.begin(); iterCol != col.end(); ++iterCol) {
const std::pair<Word,std::vector<float> > &node = *iterCol;
Phrase subphrase(prevSubPhrase);
subphrase.AddWord(node.first);
ConfusionNet::Column::const_iterator iterCol;
for (iterCol = col.begin(); iterCol != col.end(); ++iterCol) {
const std::pair<Word,std::vector<float> > &node = *iterCol;
Phrase subphrase(prevSubPhrase);
subphrase.AddWord(node.first);
ScoreComponentCollection score(prevScore);
score.PlusEquals(&inputFeature, node.second);
ScoreComponentCollection score(prevScore);
score.PlusEquals(&inputFeature, node.second);
SourcePath newSourcePath(subphrase, score);
newSubphrases.push_back(newSourcePath);
}
}
SourcePath newSourcePath(subphrase, score);
newSubphrases.push_back(newSourcePath);
}
}
}
/* forcibly create translation option for a particular source word.
@ -128,24 +133,16 @@ void TranslationOptionCollectionConfusionNet::ProcessUnknownWord(size_t sourcePo
const std::vector<TranslationOptionCollectionConfusionNet::SourcePath> &TranslationOptionCollectionConfusionNet::GetPhrases(size_t startPos, size_t endPos) const
{
size_t maxSize = endPos - startPos;
//size_t maxSizePhrase = StaticData::Instance().GetMaxPhraseLength();
//maxSize = std::min(maxSize, maxSizePhrase);
CHECK(maxSize < m_collection[startPos].size());
return m_collection[startPos][maxSize];
size_t offset = endPos - startPos;
CHECK(offset < m_collection[startPos].size());
return m_collection[startPos][offset];
}
std::vector<TranslationOptionCollectionConfusionNet::SourcePath> &TranslationOptionCollectionConfusionNet::GetPhrases(size_t startPos, size_t endPos)
{
size_t maxSize = endPos - startPos;
//size_t maxSizePhrase = StaticData::Instance().GetMaxPhraseLength();
//maxSize = std::min(maxSize, maxSizePhrase);
CHECK(maxSize < m_collection[startPos].size());
return m_collection[startPos][maxSize];
size_t offset = endPos - startPos;
CHECK(offset < m_collection[startPos].size());
return m_collection[startPos][offset];
}
} // namespace

6
moses/TranslationOptionCollectionConfusionNet.h
View File

@ -27,9 +27,9 @@ protected:
std::vector<std::vector<std::vector<SourcePath> > > m_collection;
void CreateSubPhrases(std::vector<SourcePath> &newSubphrases
, const std::vector<SourcePath> &prevSubphrases
, const ConfusionNet::Column &col
, const InputFeature &inputFeature);
, const std::vector<SourcePath> &prevSubphrases
, const ConfusionNet::Column &col
, const InputFeature &inputFeature);
};
}

36
moses/TranslationOptionCollectionText.cpp
View File

@ -30,8 +30,28 @@ using namespace std;
namespace Moses
{
/** constructor; just initialize the base class */
TranslationOptionCollectionText::TranslationOptionCollectionText(Sentence const &inputSentence, size_t maxNoTransOptPerCoverage, float translationOptionThreshold)
: TranslationOptionCollection(inputSentence, maxNoTransOptPerCoverage, translationOptionThreshold) {}
TranslationOptionCollectionText::TranslationOptionCollectionText(Sentence const &input
, size_t maxNoTransOptPerCoverage
, float translationOptionThreshold)
: TranslationOptionCollection(input, maxNoTransOptPerCoverage, translationOptionThreshold)
{
size_t size = input.GetSize();
m_collection.resize(size);
for (size_t startPos = 0; startPos < size; ++startPos) {
std::vector<Phrase> &vec = m_collection[startPos];
for (size_t endPos = startPos; endPos < size; ++endPos) {
Phrase subphrase(input.GetSubString(WordsRange(startPos, endPos)));
vec.push_back(subphrase);
}
}
/*
for (size_t startPos = 0; startPos < size; ++startPos) {
for (size_t endPos = startPos; endPos < size; ++endPos) {
cerr << startPos << "-" << endPos << "=" << GetPhrase(startPos, endPos) << endl;
}
}
*/
}
/* forcibly create translation option for a particular source word.
* For text, this function is easy, just call the base class' ProcessOneUnknownWord()
@ -66,10 +86,16 @@ void TranslationOptionCollectionText::CreateXmlOptionsForRange(size_t startPosit
for(size_t i=0; i<xmlOptions.size(); i++) {
Add(xmlOptions[i]);
}
};
}
const Phrase &TranslationOptionCollectionText::GetPhrase(size_t startPos, size_t endPos) const
{
size_t offset = endPos - startPos;
CHECK(offset < m_collection[startPos].size());
return m_collection[startPos][offset];
}
} // namespace

11
moses/TranslationOptionCollectionText.h
View File

@ -22,7 +22,9 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
#ifndef moses_TranslationOptionCollectionText_h
#define moses_TranslationOptionCollectionText_h
#include <vector>
#include "TranslationOptionCollection.h"
#include "Phrase.h"
namespace Moses
{
@ -35,14 +37,15 @@ class Sentence;
class TranslationOptionCollectionText : public TranslationOptionCollection
{
public:
TranslationOptionCollectionText(Sentence const& input, size_t maxNoTransOptPerCoverage, float translationOptionThreshold);
void ProcessUnknownWord(size_t sourcePos);
TranslationOptionCollectionText(Sentence const& inputSentence, size_t maxNoTransOptPerCoverage, float translationOptionThreshold);
bool HasXmlOptionsOverlappingRange(size_t startPosition, size_t endPosition) const;
void CreateXmlOptionsForRange(size_t startPosition, size_t endPosition);
const Phrase &GetPhrase(size_t startPos, size_t endPos) const;
protected:
std::vector<std::vector<Phrase> > m_collection;
};