File size: 4,123 Bytes
fd49381 |
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 |
#include <fstream>
#include "DsgModel.h"
#include "dsgHyp.h"
#include "moses/Util.h"
#include "util/exception.hh"
using namespace std;
using namespace lm::ngram;
namespace Moses
{
DesegModel::DesegModel(const std::string &line)
:StatefulFeatureFunction(5, line )
{
tFactor = 0;
order=5;
numFeatures = 5;
optimistic = 1;
ReadParameters();
}
DesegModel::~DesegModel()
{
delete DSGM;
}
void DesegModel :: readLanguageModel(const char *lmFile)
{
DSGM = ConstructDsgLM(m_lmPath.c_str());
State startState = DSGM->NullContextState();
desegT=new Desegmenter(m_desegPath,m_simple);// Desegmentation Table
}
void DesegModel::Load(AllOptions::ptr const& opts)
{
m_options = opts;
readLanguageModel(m_lmPath.c_str());
}
void DesegModel:: EvaluateInIsolation(const Phrase &source
, const TargetPhrase &targetPhrase
, ScoreComponentCollection &scoreBreakdown
, ScoreComponentCollection &estimatedScores) const
{
dsgHypothesis obj;
vector <string> myTargetPhrase;
vector<float> scores;
vector<string> targ_phrase; //stores the segmented tokens in the target phrase
const AlignmentInfo &align = targetPhrase.GetAlignTerm();
for (int i = 0; i < targetPhrase.GetSize(); i++) {
targ_phrase.push_back(targetPhrase.GetWord(i).GetFactor(tFactor)->GetString().as_string());
}
obj.setState(DSGM->NullContextState());
obj.setPhrases(targ_phrase);
obj.calculateDsgProbinIsol(*DSGM,*desegT,align);
obj.populateScores(scores,numFeatures);
estimatedScores.PlusEquals(this, scores);
}
FFState* DesegModel::EvaluateWhenApplied(
const Hypothesis& cur_hypo,
const FFState* prev_state,
ScoreComponentCollection* accumulator) const
{
const TargetPhrase &target = cur_hypo.GetCurrTargetPhrase();
const Range &src_rng =cur_hypo.GetCurrSourceWordsRange();
const AlignmentInfo &align = cur_hypo.GetCurrTargetPhrase().GetAlignTerm();
size_t sourceOffset = src_rng.GetStartPos();
dsgHypothesis obj;
vector<float> scores;
vector<string> targ_phrase; //stores the segmented tokens in the target phrase
bool isCompleted;
isCompleted=cur_hypo.IsSourceCompleted();
for (int i = 0; i < cur_hypo.GetCurrTargetLength(); i++) {
targ_phrase.push_back(target.GetWord(i).GetFactor(tFactor)->GetString().as_string());
}
obj.setState(prev_state);
obj.setPhrases( targ_phrase );
obj.calculateDsgProb(*DSGM,*desegT,isCompleted,align, sourceOffset, optimistic);
obj.populateScores(scores,numFeatures);
accumulator->PlusEquals(this, scores);
return obj.saveState();
}
FFState* DesegModel::EvaluateWhenApplied(
const ChartHypothesis& /* cur_hypo */,
int /* featureID - used to index the state in the previous hypotheses */,
ScoreComponentCollection* accumulator) const
{
UTIL_THROW2("Chart decoding not support by UTIL_THROW2");
}
const FFState* DesegModel::EmptyHypothesisState(const InputType &input) const
{
VERBOSE(3,"DesegModel::EmptyHypothesisState()" << endl);
State startState = DSGM->BeginSentenceState();
dsgState ss= dsgState(startState);
return new dsgState(ss);
}
std::string DesegModel::GetScoreProducerWeightShortName(unsigned idx) const
{
return "dsg";
}
void DesegModel::SetParameter(const std::string& key, const std::string& value)
{
if (key == "path") {
m_lmPath = value;
} else if (key == "contiguity-features") {
if(value == "no")
numFeatures = 1;
else
numFeatures = 5;
} else if (key == "output-factor") {
tFactor = Scan<int>(value);
} else if (key == "optimistic") {
if (value == "n")
optimistic = 0;
else
optimistic = 1;
} else if (key == "deseg-path") {
m_desegPath = Scan<int>(value);
} else if (key == "deseg-scheme") {
if(value == "s")
m_simple = 1;
else
m_simple = 0;
} else if (key == "order") {
order = Scan<int>(value);
} else {
StatefulFeatureFunction::SetParameter(key, value);
}
}
bool DesegModel::IsUseable(const FactorMask &mask) const
{
bool ret = mask[0];
return ret;
}
} // namespace
|