From dfd23deb734266fed5096043f5411c03feb2c356 Mon Sep 17 00:00:00 2001
From: Patrick Simianer <simianer@cl.uni-heidelberg.de>
Date: Sun, 24 Jun 2012 16:42:56 +0200
Subject: RuleTargetBigramFeatures, parallelize.rb

---
 decoder/cdec_ff.cc  |  3 ++-
 decoder/ff_rules.cc | 48 +++++++++++++++++++++++++++++++++++++++++++-----
 decoder/ff_rules.h  | 19 +++++++++++++++++--
 3 files changed, 62 insertions(+), 8 deletions(-)

(limited to 'decoder')

diff --git a/decoder/cdec_ff.cc b/decoder/cdec_ff.cc
index b516c386..d64bdada 100644
--- a/decoder/cdec_ff.cc
+++ b/decoder/cdec_ff.cc
@@ -47,8 +47,9 @@ void register_feature_functions() {
   ff_registry.Register("RuleIdentityFeatures", new FFFactory<RuleIdentityFeatures>());
   ff_registry.Register("SourceSyntaxFeatures", new FFFactory<SourceSyntaxFeatures>);
   ff_registry.Register("SourceSpanSizeFeatures", new FFFactory<SourceSpanSizeFeatures>);
-  ff_registry.Register("RuleNgramFeatures", new FFFactory<RuleNgramFeatures>());
   ff_registry.Register("CMR2008ReorderingFeatures", new FFFactory<CMR2008ReorderingFeatures>());
+  ff_registry.Register("RuleSourceBigramFeatures", new FFFactory<RuleSourceBigramFeatures>());
+  ff_registry.Register("RuleTargetBigramFeatures", new FFFactory<RuleTargetBigramFeatures>());
   ff_registry.Register("KLanguageModel", new KLanguageModelFactory());
   ff_registry.Register("NonLatinCount", new FFFactory<NonLatinCount>);
   ff_registry.Register("RuleShape", new FFFactory<RuleShapeFeatures>);
diff --git a/decoder/ff_rules.cc b/decoder/ff_rules.cc
index bd4c4cc0..3d0e514a 100644
--- a/decoder/ff_rules.cc
+++ b/decoder/ff_rules.cc
@@ -66,15 +66,15 @@ void RuleIdentityFeatures::TraversalFeaturesImpl(const SentenceMetadata& smeta,
   features->add_value(it->second, 1);
 }
 
-RuleNgramFeatures::RuleNgramFeatures(const std::string& param) {
+RuleSourceBigramFeatures::RuleSourceBigramFeatures(const std::string& param) {
 }
 
-void RuleNgramFeatures::PrepareForInput(const SentenceMetadata& smeta) {
+void RuleSourceBigramFeatures::PrepareForInput(const SentenceMetadata& smeta) {
 //  std::map<const TRule*, SparseVector<double> >
   rule2_feats_.clear();
 }
 
-void RuleNgramFeatures::TraversalFeaturesImpl(const SentenceMetadata& smeta,
+void RuleSourceBigramFeatures::TraversalFeaturesImpl(const SentenceMetadata& smeta,
                                          const Hypergraph::Edge& edge,
                                          const vector<const void*>& ant_contexts,
                                          SparseVector<double>* features,
@@ -92,14 +92,52 @@ void RuleNgramFeatures::TraversalFeaturesImpl(const SentenceMetadata& smeta,
       assert(w > 0);
       const string& cur = TD::Convert(w);
       ostringstream os;
-      os << "RB:" << prev << '_' << cur;
+      os << "RBS:" << prev << '_' << cur;
       const int fid = FD::Convert(Escape(os.str()));
       if (fid <= 0) return;
       f.add_value(fid, 1.0);
       prev = cur;
     }
     ostringstream os;
-    os << "RB:" << prev << '_' << "</r>";
+    os << "RBS:" << prev << '_' << "</r>";
+    f.set_value(FD::Convert(Escape(os.str())), 1.0);
+  }
+  (*features) += it->second;
+}
+
+RuleTargetBigramFeatures::RuleTargetBigramFeatures(const std::string& param) {
+}
+
+void RuleTargetBigramFeatures::PrepareForInput(const SentenceMetadata& smeta) {
+  rule2_feats_.clear();
+}
+
+void RuleTargetBigramFeatures::TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                         const Hypergraph::Edge& edge,
+                                         const vector<const void*>& ant_contexts,
+                                         SparseVector<double>* features,
+                                         SparseVector<double>* estimated_features,
+                                         void* context) const {
+  map<const TRule*, SparseVector<double> >::iterator it = rule2_feats_.find(edge.rule_.get());
+  if (it == rule2_feats_.end()) {
+    const TRule& rule = *edge.rule_;
+    it = rule2_feats_.insert(make_pair(&rule, SparseVector<double>())).first;
+    SparseVector<double>& f = it->second;
+    string prev = "<r>";
+    for (int i = 0; i < rule.e_.size(); ++i) {
+      WordID w = rule.e_[i];
+      if (w < 0) w = -w;
+      if (w == 0) return;
+      const string& cur = TD::Convert(w);
+      ostringstream os;
+      os << "RBT:" << prev << '_' << cur;
+      const int fid = FD::Convert(Escape(os.str()));
+      if (fid <= 0) return;
+      f.add_value(fid, 1.0);
+      prev = cur;
+    }
+    ostringstream os;
+    os << "RBT:" << prev << '_' << "</r>";
     f.set_value(FD::Convert(Escape(os.str())), 1.0);
   }
   (*features) += it->second;
diff --git a/decoder/ff_rules.h b/decoder/ff_rules.h
index 48d8bd05..08b168b0 100644
--- a/decoder/ff_rules.h
+++ b/decoder/ff_rules.h
@@ -22,9 +22,24 @@ class RuleIdentityFeatures : public FeatureFunction {
   mutable std::map<const TRule*, int> rule2_fid_;
 };
 
-class RuleNgramFeatures : public FeatureFunction {
+class RuleSourceBigramFeatures : public FeatureFunction {
  public:
-  RuleNgramFeatures(const std::string& param);
+  RuleSourceBigramFeatures(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* context) const;
+  virtual void PrepareForInput(const SentenceMetadata& smeta);
+ private:
+  mutable std::map<const TRule*, SparseVector<double> > rule2_feats_;
+};
+
+class RuleTargetBigramFeatures : public FeatureFunction {
+ public:
+  RuleTargetBigramFeatures(const std::string& param);
  protected:
   virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
                                      const Hypergraph::Edge& edge,
-- 
cgit v1.2.3