span length

moses/FF/SpanLength.cpp

@@ -7,6 +7,7 @@
 #include "moses/StackVec.h"
 #include "moses/TargetPhrase.h"
 #include "moses/PP/PhraseProperty.h"
+#include "moses/PP/SpanLengthPhraseProperty.h"
 
 using namespace std;
 
@@ -15,7 +16,9 @@ namespace Moses
 SpanLength::SpanLength(const std::string &line)
 :StatelessFeatureFunction(1, line)
 ,m_smoothingMethod(None)
+,m_const(0)
 {
+  ReadParameters();
 }
 
 void SpanLength::Evaluate(const Phrase &source
@@ -42,19 +45,26 @@ void SpanLength::Evaluate(const InputType &input
 	  return;
   }
 
-  string str = property.get()->GetValueString();
+  PhraseProperty *prop = property.get();
+  SpanLengthPhraseProperty *slProp = static_cast<SpanLengthPhraseProperty*>(prop);
+  assert(slProp);
+
+  string str = slProp->GetValueString();
   cerr << "str=" << str << endl;
 
   const Phrase *ruleSource = targetPhrase.GetRuleSource();
   assert(ruleSource);
   cerr << *ruleSource << endl;
 
+  float score = 0;
   for (size_t i = 0; i < stackVec->size(); ++i) {
 	  const ChartCellLabel &cell = *stackVec->at(i);
 	  const WordsRange &ntRange = cell.GetCoverage();
-
-
+	  size_t sourceWidth = ntRange.GetNumWordsCovered();
+	  float prob = slProp->GetProb(i, sourceWidth, m_const);
+	  score += TransformScore(prob);
   }
+  scoreBreakdown.PlusEquals(this, score);
 
 }
 

moses/PP/SpanLengthPhraseProperty.cpp

@@ -18,6 +18,10 @@ SpanLengthPhraseProperty::SpanLengthPhraseProperty(const std::string &value)
 	  float count = Scan<float>(toks[i + 1]);
 	  Populate(span, count);
   }
+
+  // totals
+  CalcTotals(m_source);
+  CalcTotals(m_target);
 }
 
 void SpanLengthPhraseProperty::Populate(const string &span, float count)
@@ -37,9 +41,44 @@ void SpanLengthPhraseProperty::Populate(const std::vector<size_t> &toks, float c
 	  m_source.resize(ntInd + 1);
 	  m_target.resize(ntInd + 1);
   }
-  m_source[ntInd][sourceLength] = count;
-  m_target[ntInd][targetLength] = count;
+  m_source[ntInd].first[sourceLength] = count;
+  m_target[ntInd].first[targetLength] = count;
 
 }
 
+void SpanLengthPhraseProperty::CalcTotals(Vec &vec)
+{
+	for (size_t i = 0; i < vec.size(); i = i + 2) {
+		float total = 0;
+
+		const Map &map = vec[i].first;
+		Map::const_iterator iter;
+		for (iter = map.begin(); iter != map.end(); ++iter) {
+			float count = iter->second;
+			total += count;
+		}
+
+		vec[i].second = total;
+	}
+}
+
+float SpanLengthPhraseProperty::GetProb(size_t ntInd, size_t sourceWidth, float smoothing) const
+{
+	float count;
+
+	const std::pair<Map, float> &data = m_source[ntInd];
+	const Map &map = data.first;
+	Map::const_iterator iter = map.find(sourceWidth);
+	if (iter == map.end()) {
+	  count = 0;
+	}
+	else {
+      count = iter->second;
+	}
+	count += smoothing;
+
+	float ret = count / (data.second + smoothing * map.size());
+	return ret;
+}
+
 }

moses/PP/SpanLengthPhraseProperty.h

@@ -14,14 +14,16 @@ class SpanLengthPhraseProperty : public PhraseProperty
 public:
 	SpanLengthPhraseProperty(const std::string &value);
 
+	float GetProb(size_t ntInd, size_t sourceWidth, float smoothing) const;
 protected:
 	// fractional counts
 	typedef std::map<size_t, float> Map;
-	typedef std::vector<Map> Vec;
+	typedef std::vector<std::pair<Map, float> > Vec;
 	Vec m_source, m_target;
 
 	void Populate(const std::string &span, float count);
 	void Populate(const std::vector<size_t> &toks, float count);
+	void CalcTotals(Vec &vec);
 };
 
 } // namespace Moses