当前位置: 首页>>代码示例>>C++>>正文


C++ InputType::GetDocumentId方法代码示例

本文整理汇总了C++中InputType::GetDocumentId方法的典型用法代码示例。如果您正苦于以下问题:C++ InputType::GetDocumentId方法的具体用法?C++ InputType::GetDocumentId怎么用?C++ InputType::GetDocumentId使用的例子?那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在InputType的用法示例。


在下文中一共展示了InputType::GetDocumentId方法的1个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的C++代码示例。

示例1: EvaluateWithSourceContext

void WordTranslationFeature::EvaluateWithSourceContext(const InputType &input
    , const InputPath &inputPath
    , const TargetPhrase &targetPhrase
    , const StackVec *stackVec
    , ScoreComponentCollection &scoreBreakdown
    , ScoreComponentCollection *estimatedScores) const
{
  const Sentence& sentence = static_cast<const Sentence&>(input);
  const AlignmentInfo &alignment = targetPhrase.GetAlignTerm();

  // process aligned words
  for (AlignmentInfo::const_iterator alignmentPoint = alignment.begin(); alignmentPoint != alignment.end(); alignmentPoint++) {
    const Phrase& sourcePhrase = inputPath.GetPhrase();
    int sourceIndex = alignmentPoint->first;
    int targetIndex = alignmentPoint->second;
    Word ws = sourcePhrase.GetWord(sourceIndex);
    if (m_factorTypeSource == 0 && ws.IsNonTerminal()) continue;
    Word wt = targetPhrase.GetWord(targetIndex);
    if (m_factorTypeSource == 0 && wt.IsNonTerminal()) continue;
    StringPiece sourceWord = ws.GetFactor(m_factorTypeSource)->GetString();
    StringPiece targetWord = wt.GetFactor(m_factorTypeTarget)->GetString();
    if (m_ignorePunctuation) {
      // check if source or target are punctuation
      char firstChar = sourceWord[0];
      CharHash::const_iterator charIterator = m_punctuationHash.find( firstChar );
      if(charIterator != m_punctuationHash.end())
        continue;
      firstChar = targetWord[0];
      charIterator = m_punctuationHash.find( firstChar );
      if(charIterator != m_punctuationHash.end())
        continue;
    }

    if (!m_unrestricted) {
      if (FindStringPiece(m_vocabSource, sourceWord) == m_vocabSource.end())
        sourceWord = "OTHER";
      if (FindStringPiece(m_vocabTarget, targetWord) == m_vocabTarget.end())
        targetWord = "OTHER";
    }

    if (m_simple) {
      // construct feature name
      util::StringStream featureName;
      featureName << m_description << "_";
      featureName << sourceWord;
      featureName << "~";
      featureName << targetWord;
      scoreBreakdown.SparsePlusEquals(featureName.str(), 1);
    }
    if (m_domainTrigger && !m_sourceContext) {
      const bool use_topicid = sentence.GetUseTopicId();
      const bool use_topicid_prob = sentence.GetUseTopicIdAndProb();
      if (use_topicid || use_topicid_prob) {
        if(use_topicid) {
          // use topicid as trigger
          const long topicid = sentence.GetTopicId();
          util::StringStream feature;
          feature << m_description << "_";
          if (topicid == -1)
            feature << "unk";
          else
            feature << topicid;

          feature << "_";
          feature << sourceWord;
          feature << "~";
          feature << targetWord;
          scoreBreakdown.SparsePlusEquals(feature.str(), 1);
        } else {
          // use topic probabilities
          const vector<string> &topicid_prob = *(input.GetTopicIdAndProb());
          if (atol(topicid_prob[0].c_str()) == -1) {
            util::StringStream feature;
            feature << m_description << "_unk_";
            feature << sourceWord;
            feature << "~";
            feature << targetWord;
            scoreBreakdown.SparsePlusEquals(feature.str(), 1);
          } else {
            for (size_t i=0; i+1 < topicid_prob.size(); i+=2) {
              util::StringStream feature;
              feature << m_description << "_";
              feature << topicid_prob[i];
              feature << "_";
              feature << sourceWord;
              feature << "~";
              feature << targetWord;
              scoreBreakdown.SparsePlusEquals(feature.str(), atof((topicid_prob[i+1]).c_str()));
            }
          }
        }
      } else {
        // range over domain trigger words (keywords)
        const long docid = input.GetDocumentId();
        for (boost::unordered_set<std::string>::const_iterator p = m_vocabDomain[docid].begin(); p != m_vocabDomain[docid].end(); ++p) {
          string sourceTrigger = *p;
          util::StringStream feature;
          feature << m_description << "_";
          feature << sourceTrigger;
          feature << "_";
//.........这里部分代码省略.........
开发者ID:mitramah,项目名称:mosesdecoder,代码行数:101,代码来源:WordTranslationFeature.cpp


注:本文中的InputType::GetDocumentId方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。