ModelSentence.java

package cmu.arktweetnlp.impl;

import java.util.ArrayList;
import java.util.Arrays;

import edu.stanford.nlp.util.Pair;

/**
 * One sequence structure -- typically, for one sentence
 * This is the model's view of a sentence -- only deals with non-textual numberized versions of everything
 */
public class ModelSentence {
	public int T;

	/** Runtime inferred, Trainingtime observed.
	 * dim T
	 **/
	public int labels[];
	
	public double confidences[];
	
	/** Runtime observed, Trainingtime observed.
	 * This is an array-of-arrays of (featID, featValue) pairs.
	 * dim T, then variable nnz per t.
	 **/
	public ArrayList<ArrayList< Pair<Integer, Double>>> observationFeatures;

	/** Runtime observed, Trainingtime observed (for MEMM).
	 * dim T st: edgeFeatures[t] = ID of label@(t-1).
	 * values in 0..(N_labels-1), plus extra higher numbers for markers (see Model)
	 **/
	public int edgeFeatures[];

	public ModelSentence(int T) {
		this.T = T;
		labels = new int[T];
		edgeFeatures = new int[T];
		observationFeatures = new ArrayList<ArrayList< Pair<Integer, Double> >>();
		for (int t=0; t<T; t++) {
			observationFeatures.add( new ArrayList<Pair<Integer,Double>>() );
		}
		Arrays.fill(labels, -1);
		Arrays.fill(edgeFeatures, -1);
	}
}