mirror of
https://github.com/moses-smt/mosesdecoder.git
synced 2024-12-27 05:55:02 +03:00
e4f5c69109
Some FF (Mmsapt, LexicalReordering, Many single-value FF) provide this number during "registration"; when missing, a default weight vector of uniform 1.0 is automatically generated. This eliminates the need for the user to figure out what the exact number of features is for each FF, which can get complicated, e.g. in the case of Mmsapt/PhraseDictionaryBitextSampling.
111 lines
3.1 KiB
C++
111 lines
3.1 KiB
C++
#include "moses/TranslationModel/PhraseDictionaryDynSuffixArray.h"
|
|
#include "moses/FactorCollection.h"
|
|
#include "moses/StaticData.h"
|
|
#include "moses/TargetPhrase.h"
|
|
#include <iomanip>
|
|
#include <boost/foreach.hpp>
|
|
using namespace std;
|
|
|
|
namespace Moses
|
|
{
|
|
PhraseDictionaryDynSuffixArray::
|
|
PhraseDictionaryDynSuffixArray(const std::string &line)
|
|
: PhraseDictionary(line, true)
|
|
,m_biSA(new BilingualDynSuffixArray())
|
|
{
|
|
ReadParameters();
|
|
}
|
|
|
|
|
|
void
|
|
PhraseDictionaryDynSuffixArray::
|
|
Load()
|
|
{
|
|
SetFeaturesToApply();
|
|
|
|
vector<float> weight = StaticData::Instance().GetWeights(this);
|
|
m_biSA->Load(m_input, m_output, m_source, m_target, m_alignments, weight);
|
|
}
|
|
|
|
PhraseDictionaryDynSuffixArray::
|
|
~PhraseDictionaryDynSuffixArray()
|
|
{
|
|
delete m_biSA;
|
|
}
|
|
|
|
void
|
|
PhraseDictionaryDynSuffixArray::
|
|
SetParameter(const std::string& key, const std::string& value)
|
|
{
|
|
if (key == "source") {
|
|
m_source = value;
|
|
} else if (key == "target") {
|
|
m_target = value;
|
|
} else if (key == "alignment") {
|
|
m_alignments = value;
|
|
} else {
|
|
PhraseDictionary::SetParameter(key, value);
|
|
}
|
|
}
|
|
|
|
const TargetPhraseCollection*
|
|
PhraseDictionaryDynSuffixArray::
|
|
GetTargetPhraseCollectionLEGACY(const Phrase& src) const
|
|
{
|
|
typedef map<SAPhrase, vector<float> >::value_type pstat_entry;
|
|
map<SAPhrase, vector<float> > pstats; // phrase (pair) statistics
|
|
m_biSA->GatherCands(src,pstats);
|
|
|
|
TargetPhraseCollection *ret = new TargetPhraseCollection();
|
|
BOOST_FOREACH(pstat_entry & e, pstats) {
|
|
TargetPhrase* tp = m_biSA->GetMosesFactorIDs(e.first, src, this);
|
|
tp->GetScoreBreakdown().Assign(this,e.second);
|
|
tp->EvaluateInIsolation(src);
|
|
ret->Add(tp);
|
|
}
|
|
// return ret;
|
|
// TargetPhraseCollection *ret = new TargetPhraseCollection();
|
|
// std::vector< std::pair< Scores, TargetPhrase*> > trg;
|
|
//
|
|
// // extract target phrases and their scores from suffix array
|
|
// m_biSA->GetTargetPhrasesByLexicalWeight(src, trg);
|
|
//
|
|
// std::vector< std::pair< Scores, TargetPhrase*> >::iterator itr;
|
|
// for(itr = trg.begin(); itr != trg.end(); ++itr) {
|
|
// Scores scoreVector = itr->first;
|
|
// TargetPhrase *targetPhrase = itr->second;
|
|
// std::transform(scoreVector.begin(),scoreVector.end(),
|
|
// scoreVector.begin(),FloorScore);
|
|
// targetPhrase->GetScoreBreakdown().Assign(this, scoreVector);
|
|
// targetPhrase->Evaluate();
|
|
// ret->Add(targetPhrase);
|
|
// }
|
|
ret->NthElement(m_tableLimit); // sort the phrases for the decoder
|
|
return ret;
|
|
}
|
|
|
|
void
|
|
PhraseDictionaryDynSuffixArray::
|
|
insertSnt(string& source, string& target, string& alignment)
|
|
{
|
|
m_biSA->addSntPair(source, target, alignment); // insert sentence pair into suffix arrays
|
|
//StaticData::Instance().ClearTransOptionCache(); // clear translation option cache
|
|
}
|
|
|
|
void
|
|
PhraseDictionaryDynSuffixArray::
|
|
deleteSnt(unsigned /* idx */, unsigned /* num2Del */)
|
|
{
|
|
// need to implement --
|
|
}
|
|
|
|
ChartRuleLookupManager*
|
|
PhraseDictionaryDynSuffixArray::
|
|
CreateRuleLookupManager(const ChartParser &, const ChartCellCollectionBase&, std::size_t)
|
|
{
|
|
UTIL_THROW(util::Exception, "SCFG decoding not supported with dynamic suffix array");
|
|
return 0;
|
|
}
|
|
|
|
}// end namepsace
|