Merge branch 'master' of github.com:pks/cdec-dtrain

4 files changed, 59 insertions, 0 deletions

diff --git a/mteval/Makefile.am b/mteval/Makefile.am
index 83adee17..681e798e 100644
--- a/mteval/Makefile.am
+++ b/mteval/Makefile.am
@@ -19,6 +19,7 @@ libmteval_a_SOURCES = \
   ns_comb.h \
   ns_docscorer.h \
   ns_ext.h \
+  ns_ssk.h \
   ns_ter.h \
   scorer.h \
   ter.h \
@@ -31,6 +32,7 @@ libmteval_a_SOURCES = \
   ns_comb.cc \
   ns_docscorer.cc \
   ns_ext.cc \
+  ns_ssk.cc \
   ns_ter.cc \
   scorer.cc \
   ter.cc
diff --git a/mteval/ns.cc b/mteval/ns.cc
index 7d73061c..d8214558 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_ssk.h"
 
 #include <cstdio>
 #include <cassert>
@@ -252,6 +253,8 @@ EvaluationMetric* EvaluationMetric::Instance(const string& imetric_id) {
       m = new BleuMetric<4, NIST>;
     } else if (metric_id == "KOEHN_BLEU") {
       m = new BleuMetric<4, Koehn>;
+    } else if (metric_id == "SSK") {
+      m = new SSKMetric;
     } else if (metric_id == "TER") {
       m = new TERMetric;
     } else if (metric_id == "METEOR") {
diff --git a/mteval/ns_ssk.cc b/mteval/ns_ssk.cc
new file mode 100644
index 00000000..c94e62ca
--- /dev/null
+++ b/mteval/ns_ssk.cc
@@ -0,0 +1,32 @@
+#include "ns_ssk.h"
+
+#include <vector>
+
+#include "kernel_string_subseq.h"
+#include "tdict.h"
+
+static const unsigned kNUMFIELDS = 2;
+static const unsigned kSIMILARITY = 0;
+static const unsigned kCOUNT = 1;
+
+unsigned SSKMetric::SufficientStatisticsVectorSize() const {
+  return kNUMFIELDS;
+}
+
+void SSKMetric::ComputeSufficientStatistics(const std::vector<WordID>& hyp,
+                                            const std::vector<std::vector<WordID> >& refs,
+                                            SufficientStats* out) const {
+  out->fields.resize(kNUMFIELDS);
+  out->fields[kCOUNT] = 1;
+  float bestsim = 0;
+  for (unsigned i = 0; i < refs.size(); ++i) {
+    float s = ssk<4>(hyp, refs[i], 0.8);
+    if (s > bestsim) bestsim = s;
+  }
+  out->fields[kSIMILARITY] = bestsim;
+}
+
+float SSKMetric::ComputeScore(const SufficientStats& stats) const {
+  return stats.fields[kSIMILARITY] / stats.fields[kCOUNT];
+}
+
diff --git a/mteval/ns_ssk.h b/mteval/ns_ssk.h
new file mode 100644
index 00000000..0d418770
--- /dev/null
+++ b/mteval/ns_ssk.h
@@ -0,0 +1,22 @@
+#ifndef _NS_SSK_H_
+#define _NS_SSK_H_
+
+#include "ns.h"
+
+class SSKMetric : public EvaluationMetric {
+  friend class EvaluationMetric;
+ private:
+  unsigned EditDistance(const std::string& hyp,
+                        const std::string& ref) const;
+ protected:
+  SSKMetric() : EvaluationMetric("SSK") {}
+
+ public:
+  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