fix bug in lattices with OOVs

5 files changed, 94 insertions, 0 deletions

diff --git a/rst_parser/Makefile.am b/rst_parser/Makefile.am
new file mode 100644
index 00000000..fef1c1a2
--- /dev/null
+++ b/rst_parser/Makefile.am
@@ -0,0 +1,16 @@
+bin_PROGRAMS = \
+  mst_train
+
+noinst_PROGRAMS = \
+  rst_test
+
+TESTS = rst_test
+
+noinst_LIBRARIES = librst.a
+
+librst_a_SOURCES = rst.cc
+
+mst_train_SOURCES = mst_train.cc
+mst_train_LDADD = librst.a $(top_srcdir)/decoder/libcdec.a $(top_srcdir)/mteval/libmteval.a $(top_srcdir)/utils/libutils.a ../klm/lm/libklm.a ../klm/util/libklm_util.a -lz
+
+AM_CPPFLAGS = -W -Wall -Wno-sign-compare $(GTEST_CPPFLAGS) -I$(top_srcdir)/decoder -I$(top_srcdir)/utils -I$(top_srcdir)/mteval -I../klm
diff --git a/rst_parser/arc_factored.h b/rst_parser/arc_factored.h
new file mode 100644
index 00000000..312d7d67
--- /dev/null
+++ b/rst_parser/arc_factored.h
@@ -0,0 +1,58 @@
+#ifndef _ARC_FACTORED_H_
+#define _ARC_FACTORED_H_
+
+#include <vector>
+#include <cassert>
+#include "array2d.h"
+#include "sparse_vector.h"
+
+class ArcFactoredForest {
+ public:
+  explicit ArcFactoredForest(short num_words) :
+      num_words_(num_words),
+      root_edges_(num_words),
+      edges_(num_words, num_words) {}
+
+  struct Edge {
+    Edge() : features(), edge_prob(prob_t::Zero()) {}
+    SparseVector<weight_t> features;
+    prob_t edge_prob;
+  };
+
+  template <class V>
+  void Reweight(const V& weights) {
+    for (int m = 0; m < num_words_; ++m) {
+      for (int h = 0; h < num_words_; ++h) {
+        if (h != m) {
+          Edge& e = edges_(h, m);
+          e.edge_prob.logeq(e.features.dot(weights));
+        }
+      }
+      if (m) {
+        Edge& e = root_edges_[m];
+        e.edge_prob.logeq(e.features.dot(weights));
+      }
+    }
+  }
+
+  const Edge& operator()(short h, short m) const {
+    assert(m > 0);
+    assert(m <= num_words_);
+    assert(h >= 0);
+    assert(h <= num_words_);
+    return h ? edges_(h - 1, m - 1) : root_edges[m - 1];
+  }
+  Edge& operator()(short h, short m) {
+    assert(m > 0);
+    assert(m <= num_words_);
+    assert(h >= 0);
+    assert(h <= num_words_);
+    return h ? edges_(h - 1, m - 1) : root_edges[m - 1];
+  }
+ private:
+  unsigned num_words_;
+  std::vector<Edge> root_edges_;
+  Array2D<Edge> edges_;
+};
+
+#endif
diff --git a/rst_parser/mst_train.cc b/rst_parser/mst_train.cc
new file mode 100644
index 00000000..1bceaff5
--- /dev/null
+++ b/rst_parser/mst_train.cc
@@ -0,0 +1,11 @@
+#include "arc_factored.h"
+
+#include <iostream>
+
+using namespace std;
+
+int main(int argc, char** argv) {
+  ArcFactoredForest af(5);
+  return 0;
+}
+
diff --git a/rst_parser/rst.cc b/rst_parser/rst.cc
new file mode 100644
index 00000000..0ab3e296
--- /dev/null
+++ b/rst_parser/rst.cc
@@ -0,0 +1,2 @@
+#include "rst.h"
+
diff --git a/rst_parser/rst.h b/rst_parser/rst.h
new file mode 100644
index 00000000..30a1f8a4
--- /dev/null
+++ b/rst_parser/rst.h
@@ -0,0 +1,7 @@
+#ifndef _RST_H_
+#define _RST_H_
+
+struct RandomSpanningTree {
+};
+
+#endif