| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
| #include <boost/algorithm/string/predicate.hpp> |
| #include "ScoreFeature.h" |
| #include "DomainFeature.h" |
| #include "InternalStructFeature.h" |
|
|
| using namespace std; |
| using namespace boost::algorithm; |
|
|
| namespace MosesTraining |
| { |
|
|
|
|
| const string& ScoreFeatureManager::usage() const |
| { |
| const static string& usage = "[--[Sparse]Domain[Indicator|Ratio|Subset|Bin] domain-file [bins]]" ; |
| return usage; |
| } |
|
|
| void ScoreFeatureManager::configure(const std::vector<std::string> args) |
| { |
| bool domainAdded = false; |
| bool sparseDomainAdded = false; |
|
|
| for (size_t i = 0; i < args.size(); ++i) { |
| if (args[i] == "--IgnoreSentenceId") { |
| m_includeSentenceId = true; |
| } else if (starts_with(args[i], "--Domain")) { |
| string type = args[i].substr(8); |
| ++i; |
| UTIL_THROW_IF(i == args.size(), ScoreFeatureArgumentException, "Missing domain file"); |
| string domainFile = args[i]; |
| UTIL_THROW_IF(domainAdded, ScoreFeatureArgumentException, |
| "Only allowed one domain feature"); |
| if (type == "Subset") { |
| m_features.push_back(ScoreFeaturePtr(new SubsetDomainFeature(domainFile))); |
| } else if (type == "Ratio") { |
| m_features.push_back(ScoreFeaturePtr(new RatioDomainFeature(domainFile))); |
| } else if (type == "Indicator") { |
| m_features.push_back(ScoreFeaturePtr(new IndicatorDomainFeature(domainFile))); |
| } else { |
| UTIL_THROW(ScoreFeatureArgumentException, "Unknown domain feature type " << type); |
| } |
| domainAdded = true; |
| m_includeSentenceId = true; |
| } else if (starts_with(args[i], "--SparseDomain")) { |
| string type = args[i].substr(14); |
| ++i; |
| UTIL_THROW_IF(i == args.size(), ScoreFeatureArgumentException, "Missing domain file"); |
| string domainFile = args[i]; |
| UTIL_THROW_IF(sparseDomainAdded, ScoreFeatureArgumentException, |
| "Only allowed one sparse domain feature"); |
| if (type == "Subset") { |
| m_features.push_back(ScoreFeaturePtr(new SparseSubsetDomainFeature(domainFile))); |
| } else if (type == "Ratio") { |
| m_features.push_back(ScoreFeaturePtr(new SparseRatioDomainFeature(domainFile))); |
| } else if (type == "Indicator") { |
| m_features.push_back(ScoreFeaturePtr(new SparseIndicatorDomainFeature(domainFile))); |
| } else { |
| UTIL_THROW(ScoreFeatureArgumentException, "Unknown domain feature type " << type); |
| } |
| sparseDomainAdded = true; |
| m_includeSentenceId = true; |
| } else if(args[i] == "--TreeFeatureSparse") { |
| |
| m_features.push_back(ScoreFeaturePtr(new InternalStructFeatureSparse())); |
| } else if(args[i] == "--TreeFeatureDense") { |
| |
| m_features.push_back(ScoreFeaturePtr(new InternalStructFeatureDense())); |
| } else { |
| UTIL_THROW(ScoreFeatureArgumentException,"Unknown score argument " << args[i]); |
| } |
|
|
| } |
|
|
| } |
|
|
| void ScoreFeatureManager::addPropertiesToPhrasePair(ExtractionPhrasePair &phrasePair, |
| float count, |
| int sentenceId) const |
| { |
| for (size_t i = 0; i < m_features.size(); ++i) { |
| m_features[i]->addPropertiesToPhrasePair(phrasePair, count, sentenceId); |
| } |
| } |
|
|
| void ScoreFeatureManager::addFeatures(const ScoreFeatureContext& context, |
| std::vector<float>& denseValues, |
| std::map<std::string,float>& sparseValues) const |
| { |
| for (size_t i = 0; i < m_features.size(); ++i) { |
| m_features[i]->add(context, denseValues, sparseValues); |
| } |
| } |
| } |
|
|
|
|
