diff options
author | Ulrich Germann <ugermann@inf.ed.ac.uk> | 2014-07-09 05:41:28 +0400 |
---|---|---|
committer | Ulrich Germann <ugermann@inf.ed.ac.uk> | 2014-07-09 05:41:28 +0400 |
commit | 4d41211c2cd6eb75c5a229c10e98fdfa1acff3b4 (patch) | |
tree | 142dca24625b301147a931f728825dd40cf74e53 /moses/TranslationModel/UG/sapt_pscore_lex1.h | |
parent | 28d64e23396cba53a83b75e01e9977db636e2285 (diff) |
Major overhaul of Mmsapt. Reorganization of old and addition of new features in phrase tables. Many critical bug fixes.
Diffstat (limited to 'moses/TranslationModel/UG/sapt_pscore_lex1.h')
-rw-r--r-- | moses/TranslationModel/UG/sapt_pscore_lex1.h | 70 |
1 files changed, 70 insertions, 0 deletions
diff --git a/moses/TranslationModel/UG/sapt_pscore_lex1.h b/moses/TranslationModel/UG/sapt_pscore_lex1.h new file mode 100644 index 000000000..be994b0d3 --- /dev/null +++ b/moses/TranslationModel/UG/sapt_pscore_lex1.h @@ -0,0 +1,70 @@ +// -*- c++ -*- +// Phrase scorer that counts the number of unaligend words in the phrase +// written by Ulrich Germann + +#include "moses/TranslationModel/UG/mm/ug_bitext.h" +#include "sapt_pscore_base.h" +#include <boost/dynamic_bitset.hpp> + +namespace Moses { + namespace bitext + { + template<typename Token> + class + PScoreLex1 : public PhraseScorer<Token> + { + float m_alpha; + public: + LexicalPhraseScorer2<Token> scorer; + + PScoreLex1(string const& alpaspec, string const& lexfile) + { + this->m_index = -1; + this->m_num_feats = 2; + this->m_feature_names.reserve(2); + this->m_feature_names.push_back("lexfwd"); + this->m_feature_names.push_back("lexbwd"); + m_alpha = atof(alpaspec.c_str()); + scorer.open(lexfile); + } + + void + operator()(Bitext<Token> const& bt, + PhrasePair<Token>& pp, + vector<float> * dest = NULL) const + { + if (!dest) dest = &pp.fvals; + // uint32_t sid1=0,sid2=0,off1=0,off2=0,len1=0,len2=0; + // parse_pid(pp.p1, sid1, off1, len1); + // parse_pid(pp.p2, sid2, off2, len2); +#if 0 + cout << len1 << " " << len2 << endl; + Token const* t1 = bt.T1->sntStart(sid1); + for (size_t i = off1; i < off1 + len1; ++i) + cout << (*bt.V1)[t1[i].id()] << " "; + cout << __FILE__ << ":" << __LINE__ << endl; + + Token const* t2 = bt.T2->sntStart(sid2); + for (size_t i = off2; i < off2 + len2; ++i) + cout << (*bt.V2)[t2[i].id()] << " "; + cout << __FILE__ << ":" << __LINE__ << endl; + + BOOST_FOREACH (int a, pp.aln) + cout << a << " " ; + cout << __FILE__ << ":" << __LINE__ << "\n" << endl; + + scorer.score(bt.T1->sntStart(sid1)+off1,0,len1, + bt.T2->sntStart(sid2)+off2,0,len2, + pp.aln, m_alpha, + (*dest)[this->m_index], + (*dest)[this->m_index+1]); +#endif + scorer.score(pp.start1,0, pp.len1, + pp.start2,0, pp.len2, pp.aln, m_alpha, + (*dest)[this->m_index], + (*dest)[this->m_index+1]); + } + }; + } //namespace bitext +} // namespace Moses + |