Merge remote-tracking branch 'upstream/master'

17 files changed, 116 insertions, 47 deletions

diff --git a/mteval/Makefile.am b/mteval/Makefile.am
index c833eb01..aac3e6b5 100644
--- a/mteval/Makefile.am
+++ b/mteval/Makefile.am
@@ -14,6 +14,7 @@ libmteval_a_SOURCES = \
   aer_scorer.h \
   comb_scorer.h \
   external_scorer.h \
+  levenshtein.h \
   ns.h \
   ns_cer.h \
   ns_comb.h \
@@ -21,6 +22,7 @@ libmteval_a_SOURCES = \
   ns_ext.h \
   ns_ssk.h \
   ns_ter.h \
+  ns_wer.h \
   scorer.h \
   ter.h \
   aer_scorer.cc \
@@ -34,6 +36,7 @@ libmteval_a_SOURCES = \
   ns_ext.cc \
   ns_ssk.cc \
   ns_ter.cc \
+  ns_wer.cc \
   scorer.cc \
   ter.cc
 
diff --git a/mteval/aer_scorer.h b/mteval/aer_scorer.h
index 6d53d359..cd1238f3 100644
--- a/mteval/aer_scorer.h
+++ b/mteval/aer_scorer.h
@@ -1,5 +1,5 @@
-#ifndef _AER_SCORER_
-#define _AER_SCORER_
+#ifndef AER_SCORER_
+#define AER_SCORER_
 
 #include <boost/shared_ptr.hpp>
 
diff --git a/mteval/comb_scorer.h b/mteval/comb_scorer.h
index 346be576..d17d089d 100644
--- a/mteval/comb_scorer.h
+++ b/mteval/comb_scorer.h
@@ -1,5 +1,5 @@
-#ifndef _COMB_SCORER_
-#define _COMB_SCORER_
+#ifndef COMB_SCORER_H_
+#define COMB_SCORER_H_
 
 #include "scorer.h"
 
diff --git a/mteval/external_scorer.h b/mteval/external_scorer.h
index 85535655..9565d5af 100644
--- a/mteval/external_scorer.h
+++ b/mteval/external_scorer.h
@@ -1,5 +1,5 @@
-#ifndef _EXTERNAL_SCORER_H_
-#define _EXTERNAL_SCORER_H_
+#ifndef EXTERNAL_SCORER_H_
+#define EXTERNAL_SCORER_H_
 
 #include <vector>
 #include <string>
diff --git a/mteval/levenshtein.h b/mteval/levenshtein.h
new file mode 100644
index 00000000..3ae56cf5
--- /dev/null
+++ b/mteval/levenshtein.h
@@ -0,0 +1,29 @@
+#ifndef LEVENSHTEIN_H_
+#define LEVENSHTEIN_H_
+
+namespace cdec {
+
+template <typename V>
+inline unsigned LevenshteinDistance(const V& a, const V& b) {
+  const unsigned m = a.size(), n = b.size();
+  std::vector<unsigned> edit((m + 1) * 2);
+  for (unsigned i = 0; i <= n; i++) {
+    for (unsigned j = 0; j <= m; j++) {
+      if (i == 0)
+        edit[j] = j;
+      else if (j == 0)
+        edit[(i % 2) * (m + 1)] = i;
+      else
+        edit[(i % 2) * (m + 1) + j] = std::min(std::min(
+                                edit[(i % 2) * (m + 1) + j - 1] + 1,
+                                edit[((i - 1) % 2) * (m + 1) + j] + 1),
+                                edit[((i - 1) % 2) * (m + 1) + (j - 1)] 
+                                    + (a[j - 1] == b[i - 1] ? 0 : 1));
+    }
+  }
+  return edit[(n % 2) * (m + 1) + m];
+}
+
+}
+
+#endif
diff --git a/mteval/ns.cc b/mteval/ns.cc
index c1ea238b..075e0121 100644
--- a/mteval/ns.cc
+++ b/mteval/ns.cc
@@ -3,6 +3,7 @@
 #include "ns_ext.h"
 #include "ns_comb.h"
 #include "ns_cer.h"
+#include "ns_wer.h"
 #include "ns_ssk.h"
 
 #include <cstdio>
@@ -285,6 +286,8 @@ EvaluationMetric* EvaluationMetric::Instance(const string& imetric_id) {
       m = new CombinationMetric(metric_id);
     } else if (metric_id == "CER") {
       m = new CERMetric;
+    } else if (metric_id == "WER") {
+      m = new WERMetric;
     } else {
       cerr << "Implement please: " << metric_id << endl;
       abort();
diff --git a/mteval/ns.h b/mteval/ns.h
index 153bf0b8..f6329b65 100644
--- a/mteval/ns.h
+++ b/mteval/ns.h
@@ -1,5 +1,5 @@
-#ifndef _NS_H_
-#define _NS_H_
+#ifndef NS_H_
+#define NS_H_
 
 #include <string>
 #include <vector>
diff --git a/mteval/ns_cer.cc b/mteval/ns_cer.cc
index a843d471..da6683b1 100644
--- a/mteval/ns_cer.cc
+++ b/mteval/ns_cer.cc
@@ -1,5 +1,6 @@
 #include "ns_cer.h"
 #include "tdict.h"
+#include "levenshtein.h"
 
 static const unsigned kNUMFIELDS = 2;
 static const unsigned kEDITDISTANCE = 0;
@@ -13,27 +14,6 @@ unsigned CERMetric::SufficientStatisticsVectorSize() const {
   return 2;
 }
 
-unsigned CERMetric::EditDistance(const std::string& hyp,
-                                 const std::string& ref) const {
-  const unsigned m = hyp.size(), n = ref.size();
-  std::vector<unsigned> edit((m + 1) * 2);
-  for(unsigned i = 0; i < n + 1; i++) {
-    for(unsigned j = 0; j < m + 1; j++) {
-      if(i == 0)
-        edit[j] = j;
-      else if(j == 0)
-        edit[(i%2)*(m+1)] = i;
-      else
-        edit[(i%2)*(m+1) + j] = std::min(std::min(edit[(i%2)*(m+1) + j-1] + 1,
-                                                   edit[((i-1)%2)*(m+1) + j] + 1),
-                                                   edit[((i-1)%2)*(m+1) + (j-1)] 
-                                                   + (hyp[j-1] == ref[i-1] ? 0 : 1));
-      
-    }
-  }
-  return edit[(n%2)*(m+1) + m];
-}
-
 void CERMetric::ComputeSufficientStatistics(const std::vector<WordID>& hyp,
                                             const std::vector<std::vector<WordID> >& refs,
                                             SufficientStats* out) const {
@@ -42,7 +22,7 @@ void CERMetric::ComputeSufficientStatistics(const std::vector<WordID>& hyp,
   float best_score = hyp_str.size();
   for (size_t i = 0; i < refs.size(); ++i) {
     std::string ref_str(TD::GetString(refs[i]));
-    float score = EditDistance(hyp_str, ref_str);
+    float score = cdec::LevenshteinDistance(hyp_str, ref_str);
     if (score < best_score) {
       out->fields[kEDITDISTANCE] = score;
       out->fields[kCHARCOUNT] = ref_str.size();
@@ -50,6 +30,8 @@ void CERMetric::ComputeSufficientStatistics(const std::vector<WordID>& hyp,
     }
   }
 }
+
 float CERMetric::ComputeScore(const SufficientStats& stats) const {
   return stats.fields[kEDITDISTANCE] / stats.fields[kCHARCOUNT];
 }
+
diff --git a/mteval/ns_cer.h b/mteval/ns_cer.h
index 9d211181..d9927f78 100644
--- a/mteval/ns_cer.h
+++ b/mteval/ns_cer.h
@@ -1,13 +1,10 @@
-#ifndef _NS_CER_H_
-#define _NS_CER_H_
+#ifndef NS_CER_H_
+#define NS_CER_H_
 
 #include "ns.h"
 
 class CERMetric : public EvaluationMetric {
   friend class EvaluationMetric;
- private:
-  unsigned EditDistance(const std::string& hyp,
-                        const std::string& ref) const;
  protected:
   CERMetric() : EvaluationMetric("CER") {}
 
diff --git a/mteval/ns_comb.h b/mteval/ns_comb.h
index 140e7e6a..22cba169 100644
--- a/mteval/ns_comb.h
+++ b/mteval/ns_comb.h
@@ -1,5 +1,5 @@
-#ifndef _NS_COMB_H_
-#define _NS_COMB_H_
+#ifndef NS_COMB_H_
+#define NS_COMB_H_
 
 #include "ns.h"
 
diff --git a/mteval/ns_docscorer.h b/mteval/ns_docscorer.h
index b3c28fc9..5feae2df 100644
--- a/mteval/ns_docscorer.h
+++ b/mteval/ns_docscorer.h
@@ -1,5 +1,5 @@
-#ifndef _NS_DOC_SCORER_H_
-#define _NS_DOC_SCORER_H_
+#ifndef NS_DOC_SCORER_H_
+#define NS_DOC_SCORER_H_
 
 #include <vector>
 #include <string>
diff --git a/mteval/ns_ext.h b/mteval/ns_ext.h
index 78badb2e..77be14b9 100644
--- a/mteval/ns_ext.h
+++ b/mteval/ns_ext.h
@@ -1,5 +1,5 @@
-#ifndef _NS_EXTERNAL_SCORER_H_
-#define _NS_EXTERNAL_SCORER_H_
+#ifndef NS_EXTERNAL_SCORER_H_
+#define NS_EXTERNAL_SCORER_H_
 
 #include "ns.h"
 
diff --git a/mteval/ns_ssk.h b/mteval/ns_ssk.h
index 0d418770..fdace6eb 100644
--- a/mteval/ns_ssk.h
+++ b/mteval/ns_ssk.h
@@ -1,5 +1,5 @@
-#ifndef _NS_SSK_H_
-#define _NS_SSK_H_
+#ifndef NS_SSK_H_
+#define NS_SSK_H_
 
 #include "ns.h"
 
diff --git a/mteval/ns_ter.h b/mteval/ns_ter.h
index c5c25413..cffd1bd7 100644
--- a/mteval/ns_ter.h
+++ b/mteval/ns_ter.h
@@ -1,5 +1,5 @@
-#ifndef _NS_TER_H_
-#define _NS_TER_H_
+#ifndef NS_TER_H_
+#define NS_TER_H_
 
 #include "ns.h"
 
diff --git a/mteval/ns_wer.cc b/mteval/ns_wer.cc
new file mode 100644
index 00000000..f9b2bbbb
--- /dev/null
+++ b/mteval/ns_wer.cc
@@ -0,0 +1,35 @@
+#include "ns_wer.h"
+#include "tdict.h"
+#include "levenshtein.h"
+
+static const unsigned kNUMFIELDS = 2;
+static const unsigned kEDITDISTANCE = 0;
+static const unsigned kCHARCOUNT = 1;
+
+bool WERMetric::IsErrorMetric() const {
+  return true;
+}
+
+unsigned WERMetric::SufficientStatisticsVectorSize() const {
+  return 2;
+}
+
+void WERMetric::ComputeSufficientStatistics(const std::vector<WordID>& hyp,
+                                            const std::vector<std::vector<WordID> >& refs,
+                                            SufficientStats* out) const {
+  out->fields.resize(kNUMFIELDS);
+  float best_score = hyp.size();
+  for (size_t i = 0; i < refs.size(); ++i) {
+    float score = cdec::LevenshteinDistance(hyp, refs[i]);
+    if (score < best_score) {
+      out->fields[kEDITDISTANCE] = score;
+      out->fields[kCHARCOUNT] = refs[i].size();
+      best_score = score;
+    }
+  }
+}
+
+float WERMetric::ComputeScore(const SufficientStats& stats) const {
+  return stats.fields[kEDITDISTANCE] / stats.fields[kCHARCOUNT];
+}
+
diff --git a/mteval/ns_wer.h b/mteval/ns_wer.h
new file mode 100644
index 00000000..45da70c5
--- /dev/null
+++ b/mteval/ns_wer.h
@@ -0,0 +1,20 @@
+#ifndef NS_WER_H_
+#define NS_WER_H_
+
+#include "ns.h"
+
+class WERMetric : public EvaluationMetric {
+  friend class EvaluationMetric;
+ protected:
+  WERMetric() : EvaluationMetric("WER") {}
+
+ public:
+  virtual bool IsErrorMetric() const;
+  virtual unsigned SufficientStatisticsVectorSize() const;
+  virtual void ComputeSufficientStatistics(const std::vector<WordID>& hyp,
+                                           const std::vector<std::vector<WordID> >& refs,
+                                           SufficientStats* out) const;
+  virtual float ComputeScore(const SufficientStats& stats) const;
+};
+
+#endif
diff --git a/mteval/ter.h b/mteval/ter.h
index 43314791..0758c6b6 100644
--- a/mteval/ter.h
+++ b/mteval/ter.h
@@ -1,5 +1,5 @@
-#ifndef _TER_H_
-#define _TER_H_
+#ifndef TER_H_
+#define TER_H_
 
 #include "scorer.h"