added Dense and Sparse feature to scorer

4 files changed, 38 insertions, 17 deletions

diff --git a/phrase-extract/InternalStructFeature.cpp b/phrase-extract/InternalStructFeature.cpp
index 160a0baa2..5f558f4d7 100644
--- a/phrase-extract/InternalStructFeature.cpp
+++ b/phrase-extract/InternalStructFeature.cpp
@@ -7,7 +7,7 @@ namespace MosesTraining
 
 InternalStructFeature::InternalStructFeature()
 	:m_type(0){
-	cout<<"InternalStructFeature: Construct "<<m_type<<"\n";
+	//cout<<"InternalStructFeature: Construct "<<m_type<<"\n";
 }
 
 bool InternalStructFeature::equals(const PhraseAlignment& lhs, const PhraseAlignment& rhs) const{
@@ -25,21 +25,37 @@ bool InternalStructFeature::equals(const PhraseAlignment& lhs, const PhraseAlign
 void InternalStructFeature::add(const ScoreFeatureContext& context,
 	                   std::vector<float>& denseValues,
 	                   std::map<std::string,float>& sparseValues) const{
-	std::string *internalStruct=new string("(NP((DT)(NN)))");
-	add(internalStruct, denseValues, sparseValues);
+	for(size_t i=0; i<context.phrasePair.size(); i++) {
+		add(&context.phrasePair[i]->ghkmParse, denseValues, sparseValues);
+	}
+
 }
 
 void InternalStructFeatureDense::add(std::string *internalStruct,
 	                   std::vector<float>& denseValues,
 	                   std::map<std::string,float>& sparseValues) const{
-	cout<<internalStruct<<endl;
+	//cout<<"Dense: "<<*internalStruct<<endl;
+	size_t start=0;
+	int countNP=0;
+	while((start = internalStruct->find("NP", start)) != string::npos) {
+		countNP++;
+		start+=2; //length of "NP"
+	}
+	//should add e^countNP so in the decoder I get log(e^countNP)=countNP -> but is log or ln?
+	//should use this but don't know what it does? -> maybeLog( (bitmap == i) ? 2.718 : 1 )
+	denseValues.push_back(exp(countNP));
 
 }
 
 void InternalStructFeatureSparse::add(std::string *internalStruct,
 	                   std::vector<float>& denseValues,
 	                   std::map<std::string,float>& sparseValues) const{
-	cout<<internalStruct<<endl;
+	//cout<<"Sparse: "<<*internalStruct<<endl;
+	if(internalStruct->find("VBZ")!=std::string::npos)
+		sparseValues["NT_VBZ"] = 1;
+	if(internalStruct->find("VBD")!=std::string::npos)
+			sparseValues["NT_VBD"] = 1;
+
 }
 
 
diff --git a/phrase-extract/InternalStructFeature.h b/phrase-extract/InternalStructFeature.h
index 353808ef2..bd513a715 100644
--- a/phrase-extract/InternalStructFeature.h
+++ b/phrase-extract/InternalStructFeature.h
@@ -36,7 +36,7 @@ protected:
 	/** Overriden in subclass */
 	 virtual void add(std::string *internalStruct,
 	                   std::vector<float>& denseValues,
-	                   std::map<std::string,float>& sparseValues) const;
+	                   std::map<std::string,float>& sparseValues) const = 0;
 	int m_type;
 
 };
@@ -45,7 +45,7 @@ class InternalStructFeatureDense : public InternalStructFeature
 {
 public:
 	InternalStructFeatureDense()
-		:InternalStructFeature(){m_type=1; std::cout<<"InternalStructFeatureDense: Construct "<<m_type<<"\n";}
+		:InternalStructFeature(){m_type=1;} //std::cout<<"InternalStructFeatureDense: Construct "<<m_type<<"\n";}
 protected:
 	virtual void add(std::string *internalStruct,
 		             std::vector<float>& denseValues,
@@ -56,7 +56,7 @@ class InternalStructFeatureSparse : public InternalStructFeature
 {
 public:
 	InternalStructFeatureSparse()
-		:InternalStructFeature(){m_type=2; std::cout<<"InternalStructFeatureSparse: Construct "<<m_type<<"\n";}
+		:InternalStructFeature(){m_type=2;}// std::cout<<"InternalStructFeatureSparse: Construct "<<m_type<<"\n";}
 protected:
 	virtual void add(std::string *internalStruct,
 		             std::vector<float>& denseValues,
diff --git a/phrase-extract/ScoreFeature.cpp b/phrase-extract/ScoreFeature.cpp
index e967900f9..f98759755 100644
--- a/phrase-extract/ScoreFeature.cpp
+++ b/phrase-extract/ScoreFeature.cpp
@@ -19,7 +19,7 @@
 
 #include "ScoreFeature.h"
 #include "domain.h"
-//#include "InternalStructFeature.h"
+#include "InternalStructFeature.h"
 
 using namespace std;
 
@@ -37,8 +37,9 @@ 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") {
+  	if (args[i] == "--IgnoreSentenceId") {
       m_includeSentenceId = true;
     } else if (args[i].substr(0,8) == "--Domain") {
       string type = args[i].substr(8);
@@ -76,15 +77,17 @@ void ScoreFeatureManager::configure(const std::vector<std::string> args)
       }
       sparseDomainAdded = true;
       m_includeSentenceId = true;
+    } else if(args[i] == "--GHKMFeatureSparse"){
+    	//MARIA
+    	m_features.push_back(ScoreFeaturePtr(new InternalStructFeatureSparse()));
+    } else if(args[i] == "--GHKMFeatureDense"){
+    	//MARIA
+    	m_features.push_back(ScoreFeaturePtr(new InternalStructFeatureDense()));
     } else {
       UTIL_THROW(ScoreFeatureArgumentException,"Unknown score argument " << args[i]);
-    }
-    //MARIA
-    //m_features.push_back(ScoreFeaturePtr(new InternalStructSparseFeature()));
+    	}
+
   }
-  //MARIA
-  //m_features.push_back(ScoreFeaturePtr(new InternalStructFeatureSparse()));
-  //InternalStructFeatureSparse *test=new InternalStructFeatureSparse();
 
 }
 
diff --git a/phrase-extract/ScoreFeatureTest.cpp b/phrase-extract/ScoreFeatureTest.cpp
index f4570fe30..65f1f4437 100644
--- a/phrase-extract/ScoreFeatureTest.cpp
+++ b/phrase-extract/ScoreFeatureTest.cpp
@@ -69,7 +69,9 @@ static void checkDomainConfigured(
   ScoreFeatureManager manager;
   manager.configure(args);
   const std::vector<ScoreFeaturePtr>& features  = manager.getFeatures();
-  BOOST_REQUIRE_EQUAL(features.size(), 1);
+  //BOOST_REQUIRE_EQUAL(features.size(), 2);
+  //if I add to features this check will fail?
+  BOOST_REQUIRE_EQUAL(features.size(), 1); //MARIA -> what is this check and why does it fail when I add my feature?
   Expected* feature = dynamic_cast<Expected*>(features[0].get());
   BOOST_REQUIRE(feature);
   BOOST_CHECK(manager.includeSentenceId());