Examples of cc.mallet.types.Alphabet

cc.mallet.types.Alphabet
A mapping between integers and objects where the mapping in each direction is efficient. Integers are assigned consecutively, starting at zero, as objects are added to the Alphabet. Objects can not be deleted from the Alphabet and thus the integers are never reused.
The most common use of an alphabet is as a dictionary of feature names associated with a {@link cc.mallet.types.FeatureVector} in an{@link cc.mallet.types.Instance}. In a simple document classification usage, each unique word in a document would be a unique entry in the Alphabet with a unique integer associated with it. FeatureVectors rely on the integer part of the mapping to efficiently represent the subset of the Alphabet present in the FeatureVector. @see FeatureVector @see Instance @see cc.mallet.pipe.Pipe

        LinkedHashSet<Integer> stateIndices = new LinkedHashSet<Integer>();
        stateIndices.add(labelIndex);
        labelToState.put(labelIndex, stateIndices);
      }
    } else {
      stateAlphabet = new Alphabet();


      labelIter = labelAlphabet.iterator();
      while (labelIter.hasNext()) {
        String label = (String) labelIter.next();
        labelToState.put(labelAlphabet.lookupIndex(label, false),

View Full Code Here

    
    
    
    if (p.isTargetProcessing())
    {
      Alphabet targets = p.getTargetAlphabet();
      StringBuffer buf = new StringBuffer("Labels:");
      for (int i = 0; i < targets.size(); i++)
        buf.append(" ").append(targets.lookupObject(i).toString());
      logger.info(buf.toString());
    }
    if (trainOption.value)
    {
      crf = train(trainingData, testData, eval,

View Full Code Here

     * Creates a new
     * <code>SimpleTaggerSentence2FeatureVectorSequence</code> instance.
     */
    public SimpleTaggerSentence2FeatureVectorSequence ()
    {
      super (new Alphabet(), new LabelAlphabet());
    }

View Full Code Here

    }


    public Instance pipe (Instance carrier)
    {
      Object inputData = carrier.getData();
      Alphabet features = getDataAlphabet();
      LabelAlphabet labels;
      LabelSequence target = null;
      String [][] tokens;
      if (inputData instanceof String)
        tokens = parseSentence((String)inputData);
      else if (inputData instanceof String[][])
        tokens = (String[][])inputData;
      else
        throw new IllegalArgumentException("Not a String or String[][]; got "+inputData);
      FeatureVector[] fvs = new FeatureVector[tokens.length];
      if (isTargetProcessing())
      {
        labels = (LabelAlphabet)getTargetAlphabet();
        target = new LabelSequence (labels, tokens.length);
      }
      for (int l = 0; l < tokens.length; l++) {
        int nFeatures;
        if (isTargetProcessing())
        {
          if (tokens[l].length < 1)
            throw new IllegalStateException ("Missing label at line " + l + " instance "+carrier.getName ());
          nFeatures = tokens[l].length - 1;
          target.add(tokens[l][nFeatures]);
        }
        else nFeatures = tokens[l].length;
        ArrayList<Integer> featureIndices = new ArrayList<Integer>();
        for (int f = 0; f < nFeatures; f++) {
          int featureIndex = features.lookupIndex(tokens[l][f]);
          // gdruck
          // If the data alphabet's growth is stopped, featureIndex
          // will be -1.  Ignore these features.
          if (featureIndex >= 0) {
            featureIndices.add(featureIndex);

View Full Code Here

public class SvmLight2FeatureVectorAndLabel extends Pipe {


  private static final long serialVersionUID = 1L;
  
  public SvmLight2FeatureVectorAndLabel () {
    super (new Alphabet(), new LabelAlphabet());
  }

View Full Code Here

    
    
    
    if (p.isTargetProcessing())
    {
      Alphabet targets = p.getTargetAlphabet();
      StringBuffer buf = new StringBuffer("Labels:");
      for (int i = 0; i < targets.size(); i++)
        buf.append(" ").append(targets.lookupObject(i).toString());
      logger.info(buf.toString());
    }
    if (trainOption.value)
    {
      crf = train(trainingData, testData, eval,

View Full Code Here

   * @return ArrayList with the int indices of the selected features.
   */
  public static ArrayList<Integer> selectTopLDAFeatures(int numSelFeatures, ParallelTopicModel lda, Alphabet alphabet) {
    ArrayList<Integer> features = new ArrayList<Integer>();


    Alphabet seqAlphabet = lda.getAlphabet();
    
    int numTopics = lda.getNumTopics();
    
    Object[][] sorted = lda.getTopWords(seqAlphabet.size());


    for (int pos = 0; pos < seqAlphabet.size(); pos++) {
      for (int ti = 0; ti < numTopics; ti++) {
        Object feat = sorted[ti][pos].toString();
        int fi = alphabet.lookupIndex(feat,false);
        if ((fi >=0) && (!features.contains(fi))) {
          logger.info("Selected feature: " + feat);

View Full Code Here

     * Creates a new
     * <code>SimpleTaggerSentence2FeatureVectorSequence</code> instance.
     */
    public SimpleTaggerSentence2FeatureVectorSequence ()
    {
      super (new Alphabet(), new LabelAlphabet());
    }

View Full Code Here

    }


    public Instance pipe (Instance carrier)
    {
      Object inputData = carrier.getData();
      Alphabet features = getDataAlphabet();
      LabelAlphabet labels;
      LabelSequence target = null;
      String [][] tokens;
      if (inputData instanceof String)
        tokens = parseSentence((String)inputData);
      else if (inputData instanceof String[][])
        tokens = (String[][])inputData;
      else
        throw new IllegalArgumentException("Not a String or String[][]; got "+inputData);
      FeatureVector[] fvs = new FeatureVector[tokens.length];
      if (isTargetProcessing())
      {
        labels = (LabelAlphabet)getTargetAlphabet();
        target = new LabelSequence (labels, tokens.length);
      }
      for (int l = 0; l < tokens.length; l++) {
        int nFeatures;
        if (isTargetProcessing())
        {
          if (tokens[l].length < 1)
            throw new IllegalStateException ("Missing label at line " + l + " instance "+carrier.getName ());
          nFeatures = tokens[l].length - 1;
          target.add(tokens[l][nFeatures]);
        }
        else nFeatures = tokens[l].length;
        ArrayList<Integer> featureIndices = new ArrayList<Integer>();
        for (int f = 0; f < nFeatures; f++) {
          int featureIndex = features.lookupIndex(tokens[l][f]);
          // gdruck
          // If the data alphabet's growth is stopped, featureIndex
          // will be -1.  Ignore these features.
          if (featureIndex >= 0) {
            featureIndices.add(featureIndex);

View Full Code Here

    public double [] initialWeights; // indexed by state index
    public double [] finalWeights; // indexed by state index
    
    /** Construct a new empty Factors with a new empty weightsAlphabet, 0-length initialWeights and finalWeights, and the other arrays null. */
    public Factors () {
      weightAlphabet = new Alphabet();
      initialWeights = new double[0];
      finalWeights = new double[0];
      // Leave the rest as null.  They will get set later by addState() and addWeight()
      // Alternatively, we could create zero-length arrays
    }

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of cc.mallet.types.Alphabet

cc.mallet.classify.BalancedWinnowTrainer

cc.mallet.classify.FeatureConstraintUtil

cc.mallet.classify.MaxEnt

cc.mallet.classify.MaxEntOptimizableByLabelDistribution

cc.mallet.classify.MaxEntOptimizableByLabelLikelihood

cc.mallet.classify.MCMaxEnt

cc.mallet.classify.MCMaxEntTrainer$MaximizableTrainer

cc.mallet.classify.RankMaxEnt

cc.mallet.classify.RankMaxEntTrainer$MaximizableTrainer

cc.mallet.classify.WinnowTrainer

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.