Hiero phrase orientation

moses/FF/PhraseOrientationFeature.cpp

@@ -15,7 +15,6 @@
 #include "moses/Hypothesis.h"
 #include "moses/ChartHypothesis.h"
 #include "moses/ChartManager.h"
-#include "phrase-extract/extract-ghkm/Alignment.h"
 #include <boost/shared_ptr.hpp>
 
 
@@ -23,15 +22,59 @@ namespace Moses
 {
 size_t PhraseOrientationFeatureState::hash() const
 {
-  UTIL_THROW2("TODO:Haven't figure this out yet");
+  if (!m_distinguishStates) {
+    return 0;
+  }
+
+  size_t ret = 0;
+
+  if (m_leftBoundaryIsSet) {
+    HashCombineLeftBoundaryRecursive(ret, *this, m_useSparseNT);
+  }
+  if (m_rightBoundaryIsSet) {
+    boost::hash_combine(ret, 42);
+    HashCombineRightBoundaryRecursive(ret, *this, m_useSparseNT);
+  }
+
+  return 0;
 }
 
 bool PhraseOrientationFeatureState::operator==(const FFState& other) const
 {
-  UTIL_THROW2("TODO:Haven't figure this out yet");
+  if (!m_distinguishStates) {
+    return true;
+  }
+
+  const PhraseOrientationFeatureState &otherState = static_cast<const PhraseOrientationFeatureState&>(other);
+
+  if (!m_leftBoundaryIsSet && !otherState.m_leftBoundaryIsSet &&
+      !m_rightBoundaryIsSet && !otherState.m_rightBoundaryIsSet) {
+    return true;
+  }
+  if (m_leftBoundaryIsSet != otherState.m_leftBoundaryIsSet) {
+    return false;
+  }
+  if (m_rightBoundaryIsSet != otherState.m_rightBoundaryIsSet) {
+    return false;
+  }
+
+  if (m_leftBoundaryIsSet) {
+    int compareLeft = CompareLeftBoundaryRecursive(*this, otherState, m_useSparseNT);
+    if (compareLeft != 0) {
+      return false;
+    }
+  }
+  if (m_rightBoundaryIsSet) {
+    int compareRight = CompareRightBoundaryRecursive(*this, otherState, m_useSparseNT);
+    if (compareRight != 0) {
+      return false;
+    }
+  }
+
+  return true;
 }
 
-////////////////////////////////////////////////////////////////////////////////
+
 const std::string PhraseOrientationFeature::MORIENT("M");
 const std::string PhraseOrientationFeature::SORIENT("S");
 const std::string PhraseOrientationFeature::DORIENT("D");
@@ -143,7 +186,7 @@ void PhraseOrientationFeature::EvaluateInIsolation(const Phrase &source,
   if (targetPhrase.GetAlignNonTerm().GetSize() != 0) {
 
     // Initialize phrase orientation scoring object
-    MosesTraining::Syntax::GHKM::PhraseOrientation phraseOrientation(source.GetSize(), targetPhrase.GetSize(),
+    MosesTraining::PhraseOrientation phraseOrientation(source.GetSize(), targetPhrase.GetSize(),
         targetPhrase.GetAlignTerm(), targetPhrase.GetAlignNonTerm());
 
     PhraseOrientationFeature::ReoClassData* reoClassData = new PhraseOrientationFeature::ReoClassData();
@@ -159,7 +202,7 @@ void PhraseOrientationFeature::EvaluateInIsolation(const Phrase &source,
 
       // LEFT-TO-RIGHT DIRECTION
 
-      MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS l2rOrientation = phraseOrientation.GetOrientationInfo(sourceIndex,sourceIndex,MosesTraining::Syntax::GHKM::PhraseOrientation::REO_DIR_L2R);
+      MosesTraining::PhraseOrientation::REO_CLASS l2rOrientation = phraseOrientation.GetOrientationInfo(sourceIndex,sourceIndex,MosesTraining::PhraseOrientation::REO_DIR_L2R);
 
       if ( ((targetIndex == 0) || !phraseOrientation.TargetSpanIsAligned(0,targetIndex)) // boundary non-terminal in rule-initial position (left boundary)
            && (targetPhraseLHS != m_glueTargetLHS) ) { // and not glue rule
@@ -179,7 +222,7 @@ void PhraseOrientationFeature::EvaluateInIsolation(const Phrase &source,
         if (reoClassData->firstNonTerminalPreviousSourceSpanIsAligned &&
             reoClassData->firstNonTerminalFollowingSourceSpanIsAligned) {
           // discontinuous
-          l2rOrientation = MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT;
+          l2rOrientation = MosesTraining::PhraseOrientation::REO_CLASS_DLEFT;
         } else {
           reoClassData->firstNonTerminalIsBoundary = true;
         }
@@ -189,7 +232,7 @@ void PhraseOrientationFeature::EvaluateInIsolation(const Phrase &source,
 
       // RIGHT-TO-LEFT DIRECTION
 
-      MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS r2lOrientation = phraseOrientation.GetOrientationInfo(sourceIndex,sourceIndex,MosesTraining::Syntax::GHKM::PhraseOrientation::REO_DIR_R2L);
+      MosesTraining::PhraseOrientation::REO_CLASS r2lOrientation = phraseOrientation.GetOrientationInfo(sourceIndex,sourceIndex,MosesTraining::PhraseOrientation::REO_DIR_R2L);
 
       if ( ((targetIndex == targetPhrase.GetSize()-1) || !phraseOrientation.TargetSpanIsAligned(targetIndex,targetPhrase.GetSize()-1)) // boundary non-terminal in rule-final position (right boundary)
            && (targetPhraseLHS != m_glueTargetLHS) ) { // and not glue rule
@@ -209,7 +252,7 @@ void PhraseOrientationFeature::EvaluateInIsolation(const Phrase &source,
         if (reoClassData->lastNonTerminalPreviousSourceSpanIsAligned &&
             reoClassData->lastNonTerminalFollowingSourceSpanIsAligned) {
           // discontinuous
-          r2lOrientation = MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT;
+          r2lOrientation = MosesTraining::PhraseOrientation::REO_CLASS_DLEFT;
         } else {
           reoClassData->lastNonTerminalIsBoundary = true;
         }
@@ -344,25 +387,25 @@ FFState* PhraseOrientationFeature::EvaluateWhenApplied(
 
       // LEFT-TO-RIGHT DIRECTION
 
-      MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS l2rOrientation = reoClassData->nonTerminalReoClassL2R[nNT];
+      MosesTraining::PhraseOrientation::REO_CLASS l2rOrientation = reoClassData->nonTerminalReoClassL2R[nNT];
 
       IFFEATUREVERBOSE(2) {
         FEATUREVERBOSE(2, "l2rOrientation ");
         switch (l2rOrientation) {
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_LEFT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_LEFT:
           FEATUREVERBOSE2(2, "mono" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_RIGHT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_RIGHT:
           FEATUREVERBOSE2(2, "swap" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_DLEFT:
           FEATUREVERBOSE2(2, "dleft" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DRIGHT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_DRIGHT:
           FEATUREVERBOSE2(2, "dright" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN:
+        case MosesTraining::PhraseOrientation::REO_CLASS_UNKNOWN:
-          // modelType == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_MSLR
+          // modelType == MosesTraining::PhraseOrientation::REO_MSLR
           FEATUREVERBOSE2(2, "unknown->dleft" << std::endl);
           break;
         default:
@@ -405,23 +448,23 @@ FFState* PhraseOrientationFeature::EvaluateWhenApplied(
 
       } else {
 
-        if ( l2rOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_LEFT ) {
+        if ( l2rOrientation == MosesTraining::PhraseOrientation::REO_CLASS_LEFT ) {
 
           newScores[0] += TransformScore(orientationPhraseProperty->GetLeftToRightProbabilityMono());
           // if sub-derivation has left-boundary non-terminal:
           // add recursive actual score of boundary non-terminal from subderivation
           LeftBoundaryL2RScoreRecursive(featureID, prevState, 0x1, newScores, accumulator);
 
-        } else if ( l2rOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_RIGHT ) {
+        } else if ( l2rOrientation == MosesTraining::PhraseOrientation::REO_CLASS_RIGHT ) {
 
           newScores[1] += TransformScore(orientationPhraseProperty->GetLeftToRightProbabilitySwap());
           // if sub-derivation has left-boundary non-terminal:
           // add recursive actual score of boundary non-terminal from subderivation
           LeftBoundaryL2RScoreRecursive(featureID, prevState, 0x2, newScores, accumulator);
 
-        } else if ( ( l2rOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT ) ||
+        } else if ( ( l2rOrientation == MosesTraining::PhraseOrientation::REO_CLASS_DLEFT ) ||
-                    ( l2rOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DRIGHT ) ||
+                    ( l2rOrientation == MosesTraining::PhraseOrientation::REO_CLASS_DRIGHT ) ||
-                    ( l2rOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
+                    ( l2rOrientation == MosesTraining::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
 
           newScores[2] += TransformScore(orientationPhraseProperty->GetLeftToRightProbabilityDiscontinuous());
           // if sub-derivation has left-boundary non-terminal:
@@ -446,25 +489,25 @@ FFState* PhraseOrientationFeature::EvaluateWhenApplied(
 
       // RIGHT-TO-LEFT DIRECTION
 
-      MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS r2lOrientation = reoClassData->nonTerminalReoClassR2L[nNT];
+      MosesTraining::PhraseOrientation::REO_CLASS r2lOrientation = reoClassData->nonTerminalReoClassR2L[nNT];
 
       IFFEATUREVERBOSE(2) {
         FEATUREVERBOSE(2, "r2lOrientation ");
         switch (r2lOrientation) {
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_LEFT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_LEFT:
           FEATUREVERBOSE2(2, "mono" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_RIGHT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_RIGHT:
           FEATUREVERBOSE2(2, "swap" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_DLEFT:
           FEATUREVERBOSE2(2, "dleft" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DRIGHT:
+        case MosesTraining::PhraseOrientation::REO_CLASS_DRIGHT:
           FEATUREVERBOSE2(2, "dright" << std::endl);
           break;
-        case MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN:
+        case MosesTraining::PhraseOrientation::REO_CLASS_UNKNOWN:
-          // modelType == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_MSLR
+          // modelType == MosesTraining::PhraseOrientation::REO_MSLR
           FEATUREVERBOSE2(2, "unknown->dleft" << std::endl);
           break;
         default:
@@ -507,23 +550,23 @@ FFState* PhraseOrientationFeature::EvaluateWhenApplied(
 
       } else {
 
-        if ( r2lOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_LEFT ) {
+        if ( r2lOrientation == MosesTraining::PhraseOrientation::REO_CLASS_LEFT ) {
 
           newScores[m_offsetR2LScores+0] += TransformScore(orientationPhraseProperty->GetRightToLeftProbabilityMono());
           // if sub-derivation has right-boundary non-terminal:
           // add recursive actual score of boundary non-terminal from subderivation
           RightBoundaryR2LScoreRecursive(featureID, prevState, 0x1, newScores, accumulator);
 
-        } else if ( r2lOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_RIGHT ) {
+        } else if ( r2lOrientation == MosesTraining::PhraseOrientation::REO_CLASS_RIGHT ) {
 
           newScores[m_offsetR2LScores+1] += TransformScore(orientationPhraseProperty->GetRightToLeftProbabilitySwap());
           // if sub-derivation has right-boundary non-terminal:
           // add recursive actual score of boundary non-terminal from subderivation
           RightBoundaryR2LScoreRecursive(featureID, prevState, 0x2, newScores, accumulator);
 
-        } else if ( ( r2lOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT ) ||
+        } else if ( ( r2lOrientation == MosesTraining::PhraseOrientation::REO_CLASS_DLEFT ) ||
-                    ( r2lOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DRIGHT ) ||
+                    ( r2lOrientation == MosesTraining::PhraseOrientation::REO_CLASS_DRIGHT ) ||
-                    ( r2lOrientation == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
+                    ( r2lOrientation == MosesTraining::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
 
           newScores[m_offsetR2LScores+2] += TransformScore(orientationPhraseProperty->GetRightToLeftProbabilityDiscontinuous());
           // if sub-derivation has right-boundary non-terminal:
@@ -871,17 +914,17 @@ void PhraseOrientationFeature::SparseNonTerminalR2LScore(const Factor* nonTermin
 }
 
 
-const std::string* PhraseOrientationFeature::ToString(const MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS o) const
+const std::string* PhraseOrientationFeature::ToString(const MosesTraining::PhraseOrientation::REO_CLASS o) const
 {
-  if ( o == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_LEFT ) {
+  if ( o == MosesTraining::PhraseOrientation::REO_CLASS_LEFT ) {
     return &MORIENT;
 
-  } else if ( o == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_RIGHT ) {
+  } else if ( o == MosesTraining::PhraseOrientation::REO_CLASS_RIGHT ) {
     return &SORIENT;
 
-  } else if ( ( o == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DLEFT ) ||
+  } else if ( ( o == MosesTraining::PhraseOrientation::REO_CLASS_DLEFT ) ||
-              ( o == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_DRIGHT ) ||
+              ( o == MosesTraining::PhraseOrientation::REO_CLASS_DRIGHT ) ||
-              ( o == MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
+              ( o == MosesTraining::PhraseOrientation::REO_CLASS_UNKNOWN ) ) {
     return &DORIENT;
 
   } else {

moses/FF/PhraseOrientationFeature.h

@@ -16,7 +16,7 @@
 #include "StatefulFeatureFunction.h"
 #include "FFState.h"
 #include "moses/Factor.h"
-#include "phrase-extract/extract-ghkm/PhraseOrientation.h"
+#include "phrase-extract/PhraseOrientation.h"
 #include "moses/PP/OrientationPhraseProperty.h"
 #include <boost/unordered_set.hpp>
 
@@ -226,6 +226,54 @@ protected:
     return CompareRightBoundaryRecursive(*prevState, *otherPrevState, useSparseNT);
   };
 
+
+  static void HashCombineLeftBoundaryRecursive(size_t &hash, const PhraseOrientationFeatureState& state, bool useSparseNT) {
+    if (useSparseNT) {
+      boost::hash_combine(hash, state.m_leftBoundaryNonTerminalSymbol);
+    }
+    // boost::hash_combine(hash, state.m_leftBoundaryNonTerminalL2RHeuristicScoreIndex);
+    // boost::hash_combine(hash, state.m_leftBoundaryNonTerminalL2RPossibleFutureOrientations);
+
+    for (size_t i=0; i<state.m_leftBoundaryNonTerminalL2RScores.size(); ++i) {
+      if (state.m_leftBoundaryNonTerminalL2RPossibleFutureOrientations[i]) {
+        boost::hash_combine(hash, state.m_leftBoundaryNonTerminalL2RScores[i]);
+      } else {
+        boost::hash_combine(hash, 0);
+      }
+    }
+
+    if (!state.m_leftBoundaryRecursionGuard) {
+      const PhraseOrientationFeatureState *prevState = state.m_leftBoundaryPrevState;
+      if (prevState->m_leftBoundaryIsSet) {
+        HashCombineLeftBoundaryRecursive(hash, *prevState, useSparseNT);
+      }
+    }
+  };
+
+  static void HashCombineRightBoundaryRecursive(size_t &hash, const PhraseOrientationFeatureState& state, bool useSparseNT) {
+    if (useSparseNT) {
+      boost::hash_combine(hash, state.m_rightBoundaryNonTerminalSymbol);
+    }
+    // boost::hash_combine(hash, state.m_leftBoundaryNonTerminalL2RHeuristicScoreIndex);
+    // boost::hash_combine(hash, state.m_leftBoundaryNonTerminalL2RPossibleFutureOrientations);
+
+    for (size_t i=0; i<state.m_rightBoundaryNonTerminalR2LScores.size(); ++i) {
+      if (state.m_rightBoundaryNonTerminalR2LPossibleFutureOrientations[i]) {
+        boost::hash_combine(hash, state.m_rightBoundaryNonTerminalR2LScores[i]);
+      } else {
+        boost::hash_combine(hash, 0);
+      }
+    }
+
+    if (!state.m_rightBoundaryRecursionGuard) {
+      const PhraseOrientationFeatureState *prevState = state.m_rightBoundaryPrevState;
+      if (prevState->m_rightBoundaryIsSet) {
+        HashCombineRightBoundaryRecursive(hash, *prevState, useSparseNT);
+      }
+    }
+  };
+
+
   template<std::size_t N> static bool Smaller(const std::bitset<N>& x, const std::bitset<N>& y) {
     for (size_t i=0; i<N; ++i) {
       if (x[i] ^ y[i])
@@ -264,8 +312,8 @@ public:
 
   struct ReoClassData {
   public:
-    std::vector<MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS> nonTerminalReoClassL2R;
+    std::vector<MosesTraining::PhraseOrientation::REO_CLASS> nonTerminalReoClassL2R;
-    std::vector<MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS> nonTerminalReoClassR2L;
+    std::vector<MosesTraining::PhraseOrientation::REO_CLASS> nonTerminalReoClassR2L;
     bool firstNonTerminalIsBoundary;
     bool firstNonTerminalPreviousSourceSpanIsAligned;
     bool firstNonTerminalFollowingSourceSpanIsAligned;
@@ -351,7 +399,7 @@ protected:
                                  ScoreComponentCollection* scoreBreakdown,
                                  const std::string* o) const;
 
-  const std::string* ToString(const MosesTraining::Syntax::GHKM::PhraseOrientation::REO_CLASS o) const;
+  const std::string* ToString(const MosesTraining::PhraseOrientation::REO_CLASS o) const;
 
   static const std::string MORIENT;
   static const std::string SORIENT;

moses/Jamfile

@@ -105,7 +105,7 @@ lib moses :
   TranslationModel/RuleTable/*.cpp
   TranslationModel/Scope3Parser/*.cpp
   TranslationModel/CYKPlusParser/*.cpp
-  ../phrase-extract/extract-ghkm/PhraseOrientation.cpp
+  ../phrase-extract/PhraseOrientation.cpp
   FF/*.cpp
   FF/bilingual-lm/*.cpp
   FF/OSM-Feature/*.cpp

phrase-extract/Alignment.cpp

@@ -27,10 +27,6 @@
 
 namespace MosesTraining
 {
-namespace Syntax
-{
-namespace GHKM
-{
 
 void ReadAlignment(const std::string &s, Alignment &a)
 {
@@ -46,7 +42,7 @@ void ReadAlignment(const std::string &s, Alignment &a)
     }
     int src = std::atoi(s.substr(begin, end-begin).c_str());
     if (end+1 == s.size()) {
-      throw Exception("Target index missing");
+      throw Syntax::Exception("Target index missing");
     }
 
     begin = end+1;
@@ -71,6 +67,4 @@ void FlipAlignment(Alignment &a)
   }
 }
 
-}  // namespace GHKM
-}  // namespace Syntax
 }  // namespace MosesTraining

phrase-extract/Alignment.h

@@ -25,10 +25,6 @@
 
 namespace MosesTraining
 {
-namespace Syntax
-{
-namespace GHKM
-{
 
 typedef std::vector<std::pair<int, int> > Alignment;
 
@@ -36,6 +32,4 @@ void ReadAlignment(const std::string &, Alignment &);
 
 void FlipAlignment(Alignment &);
 
-}  // namespace GHKM
-}  // namespace Syntax
 }  // namespace MosesTraining

phrase-extract/ExtractedRule.h

@@ -26,6 +26,8 @@
 #include <sstream>
 #include <map>
 
+#include "extract-ghkm/PhraseOrientation.h"
+
 namespace MosesTraining
 {
 
@@ -37,8 +39,6 @@ public:
   std::string target;
   std::string alignment;
   std::string alignmentInv;
-  std::string orientation;
-  std::string orientationForward;
   std::string sourceContextLeft;
   std::string sourceContextRight;
   std::string targetContextLeft;
@@ -51,14 +51,14 @@ public:
   int endS;
   float count;
   double pcfgScore;
+  Syntax::GHKM::PhraseOrientation::REO_CLASS l2rOrientation;
+  Syntax::GHKM::PhraseOrientation::REO_CLASS r2lOrientation;
 
   ExtractedRule(int sT, int eT, int sS, int eS)
     : source()
     , target()
     , alignment()
     , alignmentInv()
-    , orientation()
-    , orientationForward()
     , sourceContextLeft()
     , sourceContextRight()
     , targetContextLeft()
@@ -70,8 +70,10 @@ public:
     , startS(sS)
     , endS(eS)
     , count(0)
-    , pcfgScore(0.0) {
+    , pcfgScore(0.0) 
-  }
+    , l2rOrientation(Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN)
+    , r2lOrientation(Syntax::GHKM::PhraseOrientation::REO_CLASS_UNKNOWN)
+  { }
 };
 
 }

phrase-extract/PhraseOrientation.cpp

@@ -28,10 +28,6 @@
 
 namespace MosesTraining
 {
-namespace Syntax
-{
-namespace GHKM
-{
 
 std::vector<float> PhraseOrientation::m_l2rOrientationPriorCounts = boost::assign::list_of(0)(0)(0)(0)(0);
 std::vector<float> PhraseOrientation::m_r2lOrientationPriorCounts = boost::assign::list_of(0)(0)(0)(0)(0);
@@ -100,6 +96,18 @@ PhraseOrientation::PhraseOrientation(int sourceSize,
   Init(sourceSize, targetSize, m_alignedToT, alignedToS, alignedCountS);
 }
 
+PhraseOrientation::PhraseOrientation(int sourceSize,
+                                     int targetSize,
+                                     const std::vector<std::vector<int> > &alignedToT,
+                                     const std::vector<std::vector<int> > &alignedToS,
+                                     const std::vector<int> &alignedCountS)
+  : m_countF(sourceSize)
+  , m_countE(targetSize)
+  , m_alignedToT(alignedToT)
+{
+  Init(sourceSize, targetSize, m_alignedToT, alignedToS, alignedCountS);
+}
+
 
 void PhraseOrientation::Init(int sourceSize,
                              int targetSize,
@@ -470,6 +478,4 @@ void PhraseOrientation::WritePriorCounts(std::ostream& out, const REO_MODEL_TYPE
   }
 }
 
-}  // namespace GHKM
-}  // namespace Syntax
 }  // namespace MosesTraining

phrase-extract/PhraseOrientation.h

@@ -32,10 +32,6 @@
 
 namespace MosesTraining
 {
-namespace Syntax
-{
-namespace GHKM
-{
 
 // The key of the map is the English index and the value is a set of the source ones
 typedef std::map <int, std::set<int> > HSentenceVertices;
@@ -49,6 +45,7 @@ public:
   enum REO_CLASS {REO_CLASS_LEFT, REO_CLASS_RIGHT, REO_CLASS_DLEFT, REO_CLASS_DRIGHT, REO_CLASS_UNKNOWN};
   enum REO_DIR {REO_DIR_L2R, REO_DIR_R2L, REO_DIR_BIDIR};
 
+  PhraseOrientation() {};
 
   PhraseOrientation(int sourceSize,
                     int targetSize,
@@ -59,6 +56,12 @@ public:
                     const Moses::AlignmentInfo &alignTerm,
                     const Moses::AlignmentInfo &alignNonTerm);
 
+  PhraseOrientation(int sourceSize,
+                    int targetSize,
+                    const std::vector<std::vector<int> > &alignedToT,
+                    const std::vector<std::vector<int> > &alignedToS,
+                    const std::vector<int> &alignedCountS);
+
   REO_CLASS GetOrientationInfo(int startF, int endF, REO_DIR direction) const;
   REO_CLASS GetOrientationInfo(int startF, int startE, int endF, int endE, REO_DIR direction) const;
   const std::string GetOrientationInfoString(int startF, int endF, REO_DIR direction=REO_DIR_BIDIR) const;
@@ -104,8 +107,8 @@ private:
     return first < second;
   };
 
-  const int m_countF;
+  int m_countF;
-  const int m_countE;
+  int m_countE;
 
   std::vector<std::vector<int> > m_alignedToT;
 
@@ -121,6 +124,4 @@ private:
   static std::vector<float> m_r2lOrientationPriorCounts;
 };
 
-}  // namespace GHKM
-}  // namespace Syntax
 }  // namespace MosesTraining

phrase-extract/RuleExtractionOptions.h

@@ -54,6 +54,7 @@ public:
   bool conditionOnTargetLhs;
   bool boundaryRules;
   bool flexScoreFlag;
+  bool phraseOrientation;
 
   RuleExtractionOptions()
     : maxSpan(10)
@@ -86,7 +87,8 @@ public:
     , unpairedExtractFormat(false)
     , conditionOnTargetLhs(false)
     , boundaryRules(false)
-    , flexScoreFlag(false) {}
+    , flexScoreFlag(false)
+    , phraseOrientation(false) {}
 };
 
 }

phrase-extract/extract-rules-main.cpp

@@ -46,6 +46,7 @@
 #include "XmlTree.h"
 #include "InputFileStream.h"
 #include "OutputFileStream.h"
+#include "extract-ghkm/PhraseOrientation.h"
 
 using namespace std;
 using namespace MosesTraining;
@@ -62,6 +63,7 @@ private:
   Moses::OutputFileStream& m_extractFileInv;
   Moses::OutputFileStream& m_extractFileContext;
   Moses::OutputFileStream& m_extractFileContextInv;
+  Syntax::GHKM::PhraseOrientation m_phraseOrientation;
 
   vector< ExtractedRule > m_extractedRules;
 
@@ -109,6 +111,7 @@ public:
 void collectWordLabelCounts(SentenceAlignmentWithSyntax &sentence );
 void writeGlueGrammar(const string &, RuleExtractionOptions &options, set< string > &targetLabelCollection, map< string, int > &targetTopLabelCollection);
 void writeUnknownWordLabel(const string &);
+void writePhraseOrientationPriors(const string &);
 
 double getPcfgScore(const SyntaxNode &);
 
@@ -142,7 +145,8 @@ int main(int argc, char* argv[])
          << " | --UnpairedExtractFormat"
          << " | --ConditionOnTargetLHS ]"
          << " | --BoundaryRules[" << options.boundaryRules << "]"
-         << " | --FlexibilityScore\n";
+         << " | --FlexibilityScore"
+         << " | --PhraseOrientation\n";
 
     exit(1);
   }
@@ -267,6 +271,8 @@ int main(int argc, char* argv[])
       options.conditionOnTargetLhs = true;
     } else if (strcmp(argv[i],"--FlexibilityScore") == 0) {
       options.flexScoreFlag = true;
+    } else if (strcmp(argv[i],"--PhraseOrientation") == 0) {
+      options.phraseOrientation = true;
     } else if (strcmp(argv[i],"-threads") == 0 ||
                strcmp(argv[i],"--threads") == 0 ||
                strcmp(argv[i],"--Threads") == 0) {
@@ -377,6 +383,11 @@ int main(int argc, char* argv[])
 
   if (options.unknownWordLabelFlag)
     writeUnknownWordLabel(fileNameUnknownWordLabel);
+
+  if (options.phraseOrientation) {
+    std::string fileNamePhraseOrientationPriors = fileNameExtract + string(".phraseOrientationPriors");
+    writePhraseOrientationPriors(fileNamePhraseOrientationPriors);
+  }
 }
 
 void ExtractTask::Run()
@@ -392,6 +403,12 @@ void ExtractTask::extractRules()
   int countT = m_sentence.target.size();
   int countS = m_sentence.source.size();
 
+  // initialize phrase orientation scoring object (for lexicalized reordering model)
+  if (m_options.phraseOrientation) {
+    m_sentence.invertAlignment(); // fill m_sentence.alignedToS
+    m_phraseOrientation = Syntax::GHKM::PhraseOrientation(countS, countT, m_sentence.alignedToT, m_sentence.alignedToS, m_sentence.alignedCountS);
+  }
+
   // phrase repository for creating hiero phrases
   RuleExist ruleExist(countT);
 
@@ -990,6 +1007,10 @@ void ExtractTask::addRule( int startT, int endT, int startS, int endS, int count
     }
   }
 
+  rule.alignment.erase(rule.alignment.size()-1);
+  if (!m_options.onlyDirectFlag)
+    rule.alignmentInv.erase(rule.alignmentInv.size()-1);
+
   // context (words to left and right)
   if (m_options.flexScoreFlag) {
     rule.sourceContextLeft = startS == 0 ? "<s>" : m_sentence.source[startS-1];
@@ -998,9 +1019,14 @@ void ExtractTask::addRule( int startT, int endT, int startS, int endS, int count
     rule.targetContextRight = endT+1 == m_sentence.target.size() ? "<s>" : m_sentence.target[endT+1];
   }
 
-  rule.alignment.erase(rule.alignment.size()-1);
+  // phrase orientation (lexicalized reordering model)
-  if (!m_options.onlyDirectFlag)
+  if (m_options.phraseOrientation) {
-    rule.alignmentInv.erase(rule.alignmentInv.size()-1);
+    rule.l2rOrientation = m_phraseOrientation.GetOrientationInfo(startS,endS,Syntax::GHKM::PhraseOrientation::REO_DIR_L2R);
+    rule.r2lOrientation = m_phraseOrientation.GetOrientationInfo(startS,endS,Syntax::GHKM::PhraseOrientation::REO_DIR_R2L);
+    // std::cerr << "span " << startS << " " << endS << std::endl;
+    // std::cerr << "phraseOrientationL2R " << m_phraseOrientation.GetOrientationInfo(startS,endS,Syntax::GHKM::PhraseOrientation::REO_DIR_L2R) << std::endl;
+    // std::cerr << "phraseOrientationR2L " << m_phraseOrientation.GetOrientationInfo(startS,endS,Syntax::GHKM::PhraseOrientation::REO_DIR_R2L) << std::endl;
+  }
 
   addRuleToCollection( rule );
 }
@@ -1070,6 +1096,15 @@ void ExtractTask::writeRulesToFile()
     if (m_options.pcfgScore) {
       out << " ||| " << rule->pcfgScore;
     }
+    if (m_options.phraseOrientation) {
+      out << " {{Orientation ";
+      m_phraseOrientation.WriteOrientation(out,rule->l2rOrientation);
+      out << " ";
+      m_phraseOrientation.WriteOrientation(out,rule->r2lOrientation);
+      m_phraseOrientation.IncrementPriorCount(Syntax::GHKM::PhraseOrientation::REO_DIR_L2R,rule->l2rOrientation,1);
+      m_phraseOrientation.IncrementPriorCount(Syntax::GHKM::PhraseOrientation::REO_DIR_R2L,rule->r2lOrientation,1);
+      out << "}}";
+    }
     out << "\n";
 
     if (!m_options.onlyDirectFlag) {
@@ -1171,8 +1206,7 @@ void collectWordLabelCounts( SentenceAlignmentWithSyntax &sentence )
   }
 }
 
-void writeUnknownWordLabel(const string & fileName)
+void writeUnknownWordLabel(const string & fileName) {
-{
   ofstream outFile;
   outFile.open(fileName.c_str());
   typedef map<string,int>::const_iterator I;
@@ -1196,6 +1230,14 @@ void writeUnknownWordLabel(const string & fileName)
   outFile.close();
 }
 
+void writePhraseOrientationPriors(const string &fileName) 
+{
+  ofstream outFile;
+  outFile.open(fileName.c_str());
+  Syntax::GHKM::PhraseOrientation::WritePriorCounts(outFile);
+  outFile.close();
+}
+
 double getPcfgScore(const SyntaxNode &node)
 {
   double score = 0.0f;

scripts/ems/experiment.perl

@@ -2364,6 +2364,12 @@ sub define_training_extract_phrases {
           $cmd .= "-unknown-word-soft-matches $unknown_word_soft_matches ";
       }
 
+      if (&get("TRAINING:phrase-orientation")) {
+        $cmd .= "-phrase-orientation ";
+        my $phrase_orientation_priors_file = &versionize(&long_file_name("phrase-orientation-priors","model",""));
+        $cmd .= "-phrase-orientation-priors-file $phrase_orientation_priors_file ";
+      }
+
       if (&get("TRAINING:use-ghkm")) {
         $cmd .= "-ghkm ";
       }
@@ -2372,12 +2378,6 @@ sub define_training_extract_phrases {
         $cmd .= "-ghkm-tree-fragments ";
       }
 
-      if (&get("TRAINING:ghkm-phrase-orientation")) {
-        $cmd .= "-ghkm-phrase-orientation ";
-        my $phrase_orientation_priors_file = &versionize(&long_file_name("phrase-orientation-priors","model",""));
-        $cmd .= "-phrase-orientation-priors-file $phrase_orientation_priors_file ";
-      }
-
       if (&get("TRAINING:ghkm-source-labels")) {
         $cmd .= "-ghkm-source-labels ";
         my $source_labels_file = &versionize(&long_file_name("source-labels","model",""));
@@ -2427,16 +2427,16 @@ sub define_training_build_ttable {
 
     if (&get("TRAINING:hierarchical-rule-set")) {
 
-      if (&get("TRAINING:ghkm-tree-fragments")) {
+      if (&get("TRAINING:phrase-orientation")) {
-        $cmd .= "-ghkm-tree-fragments ";
+        $cmd .= "-phrase-orientation ";
-      }
-
-      if (&get("TRAINING:ghkm-phrase-orientation")) {
-        $cmd .= "-ghkm-phrase-orientation ";
         my $phrase_orientation_priors_file = &versionize(&long_file_name("phrase-orientation-priors","model",""));
         $cmd .= "-phrase-orientation-priors-file $phrase_orientation_priors_file ";
       }
 
+      if (&get("TRAINING:ghkm-tree-fragments")) {
+        $cmd .= "-ghkm-tree-fragments ";
+      }
+
       if (&get("TRAINING:ghkm-source-labels")) {
         $cmd .= "-ghkm-source-labels ";
         my $source_labels_file = &versionize(&long_file_name("source-labels","model",""));
@@ -2640,6 +2640,10 @@ sub define_training_create_config {
       }
     }
 
+    if (&get("TRAINING:phrase-orientation")) {
+      $cmd .= "-phrase-orientation ";
+    }
+
     if (&get("TRAINING:ghkm-source-labels")) {
       $cmd .= "-ghkm-source-labels ";
       my $source_labels_file = &versionize(&long_file_name("source-labels","model",""));

scripts/training/train-model.perl

@@ -93,10 +93,10 @@ my($_EXTERNAL_BINDIR,
    	$_GLUE_GRAMMAR_FILE,
    	$_DONT_TUNE_GLUE_GRAMMAR,
    	$_UNKNOWN_WORD_LABEL_FILE,
+   	$_PHRASE_ORIENTATION,
+   	$_PHRASE_ORIENTATION_PRIORS_FILE,
    	$_GHKM,
    	$_GHKM_TREE_FRAGMENTS,
-   	$_GHKM_PHRASE_ORIENTATION,
-   	$_PHRASE_ORIENTATION_PRIORS_FILE,
    	$_GHKM_SOURCE_LABELS,
    	$_GHKM_SOURCE_LABELS_FILE,
    	$_GHKM_PARTS_OF_SPEECH,
@@ -210,10 +210,10 @@ $_HELP = 1
 		       'dont-tune-glue-grammar' => \$_DONT_TUNE_GLUE_GRAMMAR,
 		       'unknown-word-label-file=s' => \$_UNKNOWN_WORD_LABEL_FILE,
 		       'unknown-word-soft-matches-file=s' => \$_UNKNOWN_WORD_SOFT_MATCHES_FILE, # give dummy label to unknown word, and allow soft matches to all other labels (with cost determined by sparse features)
+		       'phrase-orientation' => \$_PHRASE_ORIENTATION,
+		       'phrase-orientation-priors-file=s' => \$_PHRASE_ORIENTATION_PRIORS_FILE, # currently relevant for Hiero and GHKM extraction only; phrase orientation for PBT has different implementation
 		       'ghkm' => \$_GHKM,
 		       'ghkm-tree-fragments' => \$_GHKM_TREE_FRAGMENTS,
-		       'ghkm-phrase-orientation' => \$_GHKM_PHRASE_ORIENTATION,
-		       'phrase-orientation-priors-file=s' => \$_PHRASE_ORIENTATION_PRIORS_FILE, # currently relevant for GHKM extraction only; phrase orientation for PBT has different implementation
                'ghkm-source-labels' => \$_GHKM_SOURCE_LABELS,
                'ghkm-source-labels-file=s' => \$_GHKM_SOURCE_LABELS_FILE,
                'ghkm-parts-of-speech' => \$_GHKM_PARTS_OF_SPEECH,
@@ -1561,11 +1561,11 @@ sub extract_phrase {
         $cmd .= " --PCFG" if $_PCFG;
         $cmd .= " --UnpairedExtractFormat" if $_ALT_DIRECT_RULE_SCORE_1 || $_ALT_DIRECT_RULE_SCORE_2;
         $cmd .= " --ConditionOnTargetLHS" if $_ALT_DIRECT_RULE_SCORE_1;
+        $cmd .= " --PhraseOrientation" if $_PHRASE_ORIENTATION;
+        $cmd .= " --PhraseOrientationPriors $_PHRASE_ORIENTATION_PRIORS_FILE" if defined($_PHRASE_ORIENTATION_PRIORS_FILE);
         if (defined($_GHKM))
         {
           $cmd .= " --TreeFragments" if $_GHKM_TREE_FRAGMENTS;
-          $cmd .= " --PhraseOrientation" if $_GHKM_PHRASE_ORIENTATION;
-          $cmd .= " --PhraseOrientationPriors $_PHRASE_ORIENTATION_PRIORS_FILE" if defined($_PHRASE_ORIENTATION_PRIORS_FILE);
           $cmd .= " --SourceLabels" if $_GHKM_SOURCE_LABELS;
           $cmd .= " --PartsOfSpeech" if $_GHKM_PARTS_OF_SPEECH;
           $cmd .= " --PartsOfSpeechFactor" if $_GHKM_PARTS_OF_SPEECH_FACTOR;
@@ -1701,6 +1701,7 @@ sub score_phrase_phrase_extract {
     my $LOG_PROB = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /LogProb/);
     my $NEG_LOG_PROB = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /NegLogProb/);
     my $NO_LEX = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /NoLex/);
+    my $MIN_COUNT = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /MinCount ([\d\.]+)/) ? $1 : undef;
     my $MIN_COUNT_HIERARCHICAL = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /MinCountHierarchical ([\d\.]+)/) ? $1 : undef;
     my $SOURCE_LABELS = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /SourceLabels/);
     my $SOURCE_LABEL_COUNTS_LHS = (defined($_SCORE_OPTIONS) && $_SCORE_OPTIONS =~ /SourceLabelCountsLHS/);
@@ -1748,13 +1749,14 @@ sub score_phrase_phrase_extract {
         $cmd .= " --SpanLength" if $SPAN_LENGTH && $inverse eq "";
         $cmd .= " --UnalignedPenalty" if $UNALIGNED_COUNT;
         $cmd .= " --UnalignedFunctionWordPenalty ".($inverse ? $UNALIGNED_FW_F : $UNALIGNED_FW_E) if $UNALIGNED_FW_COUNT;
+        $cmd .= " --MinCount $MIN_COUNT" if $MIN_COUNT;
         $cmd .= " --MinCountHierarchical $MIN_COUNT_HIERARCHICAL" if $MIN_COUNT_HIERARCHICAL;
         $cmd .= " --PCFG" if $_PCFG;
         $cmd .= " --UnpairedExtractFormat" if $_ALT_DIRECT_RULE_SCORE_1 || $_ALT_DIRECT_RULE_SCORE_2;
         $cmd .= " --ConditionOnTargetLHS" if $_ALT_DIRECT_RULE_SCORE_1;
         $cmd .= " --TreeFragments" if $_GHKM_TREE_FRAGMENTS;
-        $cmd .= " --PhraseOrientation" if $_GHKM_PHRASE_ORIENTATION;
+        $cmd .= " --PhraseOrientation" if $_PHRASE_ORIENTATION;
-        $cmd .= " --PhraseOrientationPriors $_PHRASE_ORIENTATION_PRIORS_FILE" if $_GHKM_PHRASE_ORIENTATION && defined($_PHRASE_ORIENTATION_PRIORS_FILE);
+        $cmd .= " --PhraseOrientationPriors $_PHRASE_ORIENTATION_PRIORS_FILE" if $_PHRASE_ORIENTATION && defined($_PHRASE_ORIENTATION_PRIORS_FILE);
         $cmd .= " --SourceLabels $_GHKM_SOURCE_LABELS_FILE" if $_GHKM_SOURCE_LABELS && defined($_GHKM_SOURCE_LABELS_FILE);
         $cmd .= " --PartsOfSpeech $_GHKM_PARTS_OF_SPEECH_FILE" if $_GHKM_PARTS_OF_SPEECH && defined($_GHKM_PARTS_OF_SPEECH_FILE);
         $cmd .= " $DOMAIN" if $DOMAIN;
@@ -2365,6 +2367,7 @@ sub create_ini {
   print INI "PhrasePenalty\n";
   print INI "SoftMatchingFeature name=SM0 path=$_UNKNOWN_WORD_SOFT_MATCHES_FILE\n" if $_TARGET_SYNTAX && defined($_UNKNOWN_WORD_SOFT_MATCHES_FILE);
   print INI "SoftSourceSyntacticConstraintsFeature sourceLabelSetFile=$_GHKM_SOURCE_LABELS_FILE\n" if $_GHKM_SOURCE_LABELS && defined($_GHKM_SOURCE_LABELS_FILE);
+  print INI "PhraseOrientationFeature\n" if $_PHRASE_ORIENTATION;
   print INI $feature_spec;
 
   print INI "\n# dense weights for feature functions\n";
@@ -2375,6 +2378,7 @@ sub create_ini {
   print INI "WordPenalty0= -1\n";
   print INI "PhrasePenalty0= 0.2\n";
   print INI "SoftSourceSyntacticConstraintsFeature0= -0.2 -0.2 -0.2 0.1 0.1 0.1\n" if $_GHKM_SOURCE_LABELS && defined($_GHKM_SOURCE_LABELS_FILE);
+  print INI "PhraseOrientationFeature0= 0.05 0.05 0.05 0.05 0.05 0.05\n" if $_PHRASE_ORIENTATION;
   print INI $weight_spec;
   close(INI);
 }