当前位置: 首页>>代码示例>>Java>>正文

Java TokenSequence.getProperty方法代码示例

本文整理汇总了Java中cc.mallet.types.TokenSequence.getProperty方法的典型用法代码示例。如果您正苦于以下问题：Java TokenSequence.getProperty方法的具体用法？Java TokenSequence.getProperty怎么用？Java TokenSequence.getProperty使用的例子？那么, 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类cc.mallet.types.TokenSequence的用法示例。


在下文中一共展示了TokenSequence.getProperty方法的9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Java代码示例。

示例1: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance carrier) {
  TokenSequence ts = (TokenSequence) carrier.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", carrier);

  List<String> sylls = struct.getOncGrams();
  Preconditions.checkState(ts.size() == sylls.size(), "sylls and grams dont equal size");
  for (int i = 0; i < ts.size(); i++) {
    Token t = ts.get(i);
    if (t.getText().length() != sylls.get(i).length()) {
      throw new IllegalStateException("grams doesnt match syll grams " + t + " - " + sylls);
    }
    for (int j = 0; j < windows.size(); j++) {
      TokenWindow window = windows.get(j);
      String windStr = TokenSeqUtil.getWindowFromStrings(sylls, i, window.offset, window.width);
      if (windStr == null) {
        continue;
      }
      String feature = "SYN_" + windStr + "@" + window.offset;
      t.setFeatureValue(feature, 1.0);
    }
  }
  return carrier;
}
开发者ID:steveash，项目名称:jg2p，代码行数:26，代码来源:NeighborSyllableFeature.java

示例2: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  if (struct.getSyllCount() >= 1) {
    markAll(ts, struct, 0, "SYLLREL_1");
  }
  if (struct.getSyllCount() >= 2) {
    markAll(ts, struct, struct.getLastSyllIndex(), "SYLLREL_-1");
  }
  if (struct.getSyllCount() >= 3) {
    markAll(ts, struct, 1, "SYLLREL_2");
  }
  if (struct.getSyllCount() >= 4) {
    markAll(ts, struct, struct.getLastSyllIndex() - 1, "SYLLREL_-2");
  }
  if (struct.getSyllCount() >= 5) {
    for (int i = 2; i < (struct.getLastSyllIndex() - 1); i++) {
      markAll(ts, struct, i, "SYLLRELL_X");
    }
  }

  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:26，代码来源:SyllRelativeMarkFeature.java

示例3: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  int prevSyll = -1;
  for (int i = 0; i < ts.size(); i++) {
    Token tok = ts.get(i);
    int currSyll = struct.getSyllIndexForGraphoneGramIndex(i);
    if (prevSyll != currSyll) {
      tok.setFeatureValue("SYLLMARK_" + currSyll, 1.0);
      prevSyll = currSyll;
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:17，代码来源:SyllMarkingFeature.java

示例4: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {

  TokenSequence ts = (TokenSequence) inst.getData();

  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "cant use struct without sylls", inst);
  for (int i = 0; i < ts.size(); i++) {
    Token tt = ts.get(i);
    if (!struct.graphoneGramIndexContainsNucleus(i)) {
      continue;
    }
    int syllIndex = struct.getSyllIndexForGraphoneGramIndex(i);
    String feat = null;
    if (onlyLast) {
      if (syllIndex < struct.getLastSyllIndex()) {
        feat = prefix + struct.getSyllPart(struct.getLastSyllIndex(), maxOnset, -1, maxCoda);
      }
    } else if (relativeIndex != 0) {
      int targetSyll = syllIndex + relativeIndex;
      if (targetSyll >= 0 && targetSyll <= struct.getLastSyllIndex()) {
        feat = prefix + struct.getSyllPart(targetSyll, maxOnset, -1, maxCoda);
      }
    }
    if (feat != null) {
      tt.setFeatureValue(feat, 1.0);
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:31，代码来源:VowelWindowFeature.java

示例5: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  int xx = 0;
  for (int i = 0; i < ts.size(); i++) {
    Token tok = ts.get(i);
    for (String graph : Grams.iterateSymbols(tok.getText())) {
      tok.setFeatureValue("SYLLSEQ_" + struct.getSyllSequenceForGraphemeIndex(xx), 1.0);
      xx += 1;
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:16，代码来源:SyllSequenceFeature.java

示例6: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  for (int i = 0; i < ts.size(); i++) {
    Token tok = ts.get(i);
    tok.setFeatureValue("SYLLCNT_" + struct.getSyllIndexForGraphoneGramIndex(i), 1.0);
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:12，代码来源:SyllCountingFeature.java

示例7: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  for (int i = 0; i < ts.size(); i++) {
    Token tok = ts.get(i);
    int j = 0;
    for (String graph : Grams.iterateSymbols(tok.getText())) {
      tok.setFeatureValue("SYLLCHROL_" + graph + "_" + struct.getOncCodeAtGraphoneAndSequence(i, j), 1.0);
      j += 1;
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:16，代码来源:SyllCharRoleFeature.java

示例8: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {

  TokenSequence ts = (TokenSequence) inst.getData();
  String lastToken = ts.get(ts.size() - 1).getText();
  char lastChar = lastToken.charAt(lastToken.length() - 1);
  if (lastChar != 'y' && lastChar != 'e') {
    return inst;
  }

  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  int lastSyllIndex = struct.getLastSyllIndex();
  for (int i = 0; i < ts.size(); i++) {
    String tag = null;
    Token t = ts.get(i);
    String s = struct.oncGramForGraphoneIndex(i);
    int thisSyllIndex = struct.getSyllIndexForGraphoneGramIndex(i);
    String text = t.getText();
    Preconditions.checkState(text.length() == s.length(), "grams doesnt match syll grams");
    for (int j = 0; j < text.length(); j++) {
      char textChar = text.charAt(j);
      char syllChar = s.charAt(j);
      if (Character.isWhitespace(textChar) || Character.isWhitespace(syllChar)) {
        Preconditions.checkState(textChar == syllChar, "mismatched whitespace in gram");
        continue;
      }
      if (Graphemes.isVowel(String.valueOf(textChar)) && syllChar == SyllTagTrainer.NucleusChar) {
        // we care about nucleus vowels because those are the ones influence by trailing letters
        if (tag == null) {
          tag = "TE_VOWEL_" + textChar + "_" + (thisSyllIndex < lastSyllIndex ? "BEFORE" : "END");
        }
      }
    }
    if (tag != null) {
      t.setFeatureValue(tag, 1.0);
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:41，代码来源:EndingVowelFeature.java

示例9: pipe

import cc.mallet.types.TokenSequence; //导入方法依赖的package包/类
@Override
public Instance pipe(Instance inst) {
  TokenSequence ts = (TokenSequence) inst.getData();
  SyllStructure struct = (SyllStructure) ts.getProperty(PhonemeCrfTrainer.PROP_STRUCTURE);
  checkNotNull(struct, "no sylls", inst);
  int xx = 0;
  for (int i = 0; i < ts.size(); i++) {
    Token tok = ts.get(i);
    for (String grapheme : Grams.iterateSymbols(tok.getText())) {
      int relative;
      int syllIndex = struct.getSyllIndexForGraphemeIndex(xx);
      int mySyllSeq = struct.getSyllSequenceForGraphemeIndex(xx);
      // 01201
      // YYZYZ
      if (isNext) {
        int graphsInSyllable = struct.getSyllGraphsForSyllIndex(syllIndex).length(); // no zeroes
        relative = (graphsInSyllable - mySyllSeq);
        if (syllIndex == struct.getLastSyllIndex()) {
          relative = -1;
        }
      } else {
        relative = mySyllSeq + 1;
        if (syllIndex == 0) {
          relative = -1;
        }
      }
      tok.setFeatureValue("NEARSYLL_" + (isNext ? "NEXT_" : "PREV_") + grapheme + "_" + relative, 1.0);
      xx += 1;
    }
  }
  return inst;
}
开发者ID:steveash，项目名称:jg2p，代码行数:33，代码来源:NearSyllFeature.java


注：本文中的cc.mallet.types.TokenSequence.getProperty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。