Package com.clearnlp.reader

Examples of com.clearnlp.reader.JointReader


public class DEPDeveloper extends DEPTrainer implements IDeveloper
{
  @Override
  public void develop(Element eConfig, JointFtrXml[] xmls, String[] trainFiles, String[] devFiles, String mode, boolean generate, int devId) throws Exception
  {
    JointReader reader = getJointReader(UTXml.getFirstElementByTagName(eConfig, TAG_READER));
    developComponentBoot(eConfig, reader, xmls, trainFiles, devFiles, null, generate, devId);
  }
View Full Code Here


public class RoleDeveloper extends RoleTrainer implements IDeveloper
{
  @Override
  public void develop(Element eConfig, JointFtrXml[] xmls, String[] trainFiles, String[] devFiles, String mode, boolean generate, int devId) throws Exception
  {
    JointReader reader = getJointReader(UTXml.getFirstElementByTagName(eConfig, TAG_READER));
    AbstractRolesetClassifier collector = getCollector(xmls);
    developComponent(eConfig, reader, xmls, trainFiles, devFiles, collector, generate, devId);
  }
View Full Code Here

public class PredDeveloper extends PredTrainer implements IDeveloper
{
  @Override
  public void develop(Element eConfig, JointFtrXml[] xmls, String[] trainFiles, String[] devFiles, String mode, boolean generate, int devId) throws Exception
  {
    JointReader reader = getJointReader(UTXml.getFirstElementByTagName(eConfig, TAG_READER));
    developComponent(eConfig, reader, xmls, trainFiles, devFiles, null, generate, devId);
  }
View Full Code Here

*/
abstract public class AbstractNLPTrainer extends AbstractNLP
{
  public void train(Element eConfig, JointFtrXml[] xmls, String[] trainFiles, String modelDir) throws Exception
  {
    JointReader reader = getJointReader(UTXml.getFirstElementByTagName(eConfig, TAG_READER));
    AbstractStatisticalComponent<?> component = getComponent(eConfig, reader, xmls, trainFiles, -1);
    component.save(new ObjectOutputStream(new BufferedOutputStream(new GZIPOutputStream(new FileOutputStream(modelDir+"/"+getMode())))));
  }
View Full Code Here

    s_outputFile = outputFile;
  }
 
  private List<DEPTree> getTrees(String inputFile)
  {
    JointReader reader = new JointReader(0, 1, 2, 3, 4, 5, 6);
    reader.open(UTInput.createBufferedFileReader(inputFile));
    List<DEPTree> trees = new ArrayList<DEPTree>();
    DEPTree tree;
 
    while ((tree = reader.next()) != null)
      trees.add(tree);
   
    reader.close();
    return trees;
  }
View Full Code Here

    catch (Exception e) {e.printStackTrace();}
  }
 
  public void develop(Element eConfig, JointFtrXml[] xmls, String[] trainFiles, String[] devFiles, String mode, boolean generate, int devId) throws Exception
  {
    JointReader reader = getJointReader(UTXml.getFirstElementByTagName(eConfig, TAG_READER));
    Element eTrain = UTXml.getFirstElementByTagName(eConfig, mode);
    int dfc = getDocumentFrequencyCutoff(eTrain);
    int dtc = getDocumentTokenCount(eTrain);
    DEPTree tree;
   
    Set<String> sLsfs = getSimplifiedFormsByDocumentFrequencies(reader, trainFiles, dfc, dtc);
    EnglishOnlinePOSTagger component = new EnglishOnlinePOSTagger(xmls, sLsfs);
   
    LOG.info("Collecting lexica:");
    int total = 0;
   
    for (String trainFile : trainFiles)
    {
      reader.open(UTInput.createBufferedFileReader(trainFile));
     
      while ((tree = reader.next()) != null)
      {
        component.collect(tree);
        if (++total%5000 == 0) LOG.info(".");
      }
     
      reader.close();
    LOG.info("\n");

    LOG.info("Trainig:");
   
    Object[] lexica = component.getLexica();
    component = new EnglishOnlinePOSTagger(xmls, lexica);
   
    for (String trainFile : trainFiles)
    {
      reader.open(UTInput.createBufferedFileReader(trainFile));
     
      while ((tree = reader.next()) != null)
      {
        component.train(tree);
        if (++total%5000 == 0) LOG.info(".");
      }
     
      reader.close();
    }
   
    IOnlineAlgorithm algorithm = new OnlineAdaGradHinge(0.01, 0.1);
    List<DEPTree> devTrees = getTrees(reader, devFiles);
   
    component.develop(LOG, algorithm, 5, devTrees);
   
    LOG.info("Bootsrapping:");
   
    for (String trainFile : trainFiles)
    {
      reader.open(UTInput.createBufferedFileReader(trainFile));
     
      while ((tree = reader.next()) != null)
      {
        component.bootstrap(tree);
        if (++total%5000 == 0) LOG.info(".");
      }
     
      reader.close();
    }
   
    component.develop(LOG, algorithm, 5, devTrees);
  }
View Full Code Here

public class CheckMentions
{
  public CheckMentions(String inDir)
  {
    JointReader reader = new JointReader(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10);
    File file = new File(inDir);
    DEPTree tree;
   
    for (String filename : file.list())
    {
      if (!filename.endsWith("c"))  continue;
      reader.open(UTInput.createBufferedFileReader(inDir+File.separator+filename));
      System.out.println(filename);
     
      while ((tree = reader.next()) != null)
      {
        tree.setDependents();
        check(tree);
      }
     
      reader.close();
    }
  }
View Full Code Here

    int iXHeads = map.get(AbstractColumnReader.FIELD_XHEADS) - 1;
    int iSHeads = map.get(AbstractColumnReader.FIELD_SHEADS) - 1;
    int iNament = map.get(AbstractColumnReader.FIELD_NAMENT) - 1;
    int iCoref  = map.get(AbstractColumnReader.FIELD_COREF- 1;
   
    JointReader reader = new JointReader(iId, iForm, iLemma, iPos, iFeats, iHeadId, iDeprel, iXHeads, iSHeads, iNament, iCoref);
    reader.initGoldPOSTag(map.get(AbstractColumnReader.FIELD_GPOS) - 1);
   
    return reader;
  }
View Full Code Here

TOP

Related Classes of com.clearnlp.reader.JointReader

Copyright © 2018 www.massapicom. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.