diff options
author | Michael Denkowski <michael.j.denkowski@gmail.com> | 2012-12-22 16:01:23 -0500 |
---|---|---|
committer | Michael Denkowski <michael.j.denkowski@gmail.com> | 2012-12-22 16:01:23 -0500 |
commit | 597d89c11db53e91bc011eab70fd613bbe6453e8 (patch) | |
tree | 83c87c07d1ff6d3ee4e3b1626f7eddd49c61095b /klm/search/rule.cc | |
parent | 65e958ff2678a41c22be7171456a63f002ef370b (diff) | |
parent | 201af2acd394415a05072fbd53d42584875aa4b4 (diff) |
Merge branch 'master' of git://github.com/redpony/cdec
Diffstat (limited to 'klm/search/rule.cc')
-rw-r--r-- | klm/search/rule.cc | 52 |
1 files changed, 26 insertions, 26 deletions
diff --git a/klm/search/rule.cc b/klm/search/rule.cc index 5b00207e..0244a09f 100644 --- a/klm/search/rule.cc +++ b/klm/search/rule.cc @@ -1,7 +1,7 @@ #include "search/rule.hh" +#include "lm/model.hh" #include "search/context.hh" -#include "search/final.hh" #include <ostream> @@ -9,35 +9,35 @@ namespace search { -template <class Model> float ScoreRule(const Context<Model> &context, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing) { - unsigned int oov_count = 0; - float prob = 0.0; - const Model &model = context.LanguageModel(); - const lm::WordIndex oov = model.GetVocabulary().NotFound(); - for (std::vector<lm::WordIndex>::const_iterator word = words.begin(); ; ++word) { - lm::ngram::RuleScore<Model> scorer(model, *(writing++)); - // TODO: optimize - if (prepend_bos && (word == words.begin())) { - scorer.BeginSentence(); - } - for (; ; ++word) { - if (word == words.end()) { - prob += scorer.Finish(); - return static_cast<float>(oov_count) * context.GetWeights().OOV() + prob * context.GetWeights().LM(); - } - if (*word == kNonTerminal) break; - if (*word == oov) ++oov_count; +template <class Model> ScoreRuleRet ScoreRule(const Model &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing) { + ScoreRuleRet ret; + ret.prob = 0.0; + ret.oov = 0; + const lm::WordIndex oov = model.GetVocabulary().NotFound(), bos = model.GetVocabulary().BeginSentence(); + lm::ngram::RuleScore<Model> scorer(model, *(writing++)); + std::vector<lm::WordIndex>::const_iterator word = words.begin(); + if (word != words.end() && *word == bos) { + scorer.BeginSentence(); + ++word; + } + for (; word != words.end(); ++word) { + if (*word == kNonTerminal) { + ret.prob += scorer.Finish(); + scorer.Reset(*(writing++)); + } else { + if (*word == oov) ++ret.oov; scorer.Terminal(*word); } - prob += scorer.Finish(); } + ret.prob += scorer.Finish(); + return ret; } -template float ScoreRule(const Context<lm::ngram::RestProbingModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); -template float ScoreRule(const Context<lm::ngram::ProbingModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); -template float ScoreRule(const Context<lm::ngram::TrieModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); -template float ScoreRule(const Context<lm::ngram::QuantTrieModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); -template float ScoreRule(const Context<lm::ngram::ArrayTrieModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); -template float ScoreRule(const Context<lm::ngram::QuantArrayTrieModel> &model, const std::vector<lm::WordIndex> &words, bool prepend_bos, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::RestProbingModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::ProbingModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::TrieModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::QuantTrieModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::ArrayTrieModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); +template ScoreRuleRet ScoreRule(const lm::ngram::QuantArrayTrieModel &model, const std::vector<lm::WordIndex> &words, lm::ngram::ChartState *writing); } // namespace search |