edu.ucla.sspace.index.GaussianVectorGenerator Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of sspace-wordsi Show documentation

The S-Space Package is a collection of algorithms for building Semantic Spaces as well as a highly-scalable library for designing new distributional semantics algorithms. Distributional algorithms process text corpora and represent the semantic for words as high dimensional feature vectors. This package also includes matrices, vectors, and numerous clustering algorithms. These approaches are known by many names, such as word spaces, semantic spaces, or distributed semantics and rest upon the Distributional Hypothesis: words that appear in similar contexts have similar meanings.

The newest version!

/*
 * Copyright 2009 Keith Stevens 
 *
 * This file is part of the S-Space package and is covered under the terms and
 * conditions therein.
 *
 * The S-Space package is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as published
 * by the Free Software Foundation and distributed hereunder to you.
 *
 * THIS SOFTWARE IS PROVIDED "AS IS" AND NO REPRESENTATIONS OR WARRANTIES,
 * EXPRESS OR IMPLIED ARE MADE.  BY WAY OF EXAMPLE, BUT NOT LIMITATION, WE MAKE
 * NO REPRESENTATIONS OR WARRANTIES OF MERCHANT- ABILITY OR FITNESS FOR ANY
 * PARTICULAR PURPOSE OR THAT THE USE OF THE LICENSED SOFTWARE OR DOCUMENTATION
 * WILL NOT INFRINGE ANY THIRD PARTY PATENTS, COPYRIGHTS, TRADEMARKS OR OTHER
 * RIGHTS.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program. If not, see .
 */

package edu.ucla.sspace.index;

import edu.ucla.sspace.vector.DenseVector;
import edu.ucla.sspace.vector.DoubleVector;
import edu.ucla.sspace.vector.Vectors;

import java.io.Serializable;

import java.util.Properties;
import java.util.Random;


/**
 * A Generator for vectors where each entry is from a guassian distribution
 * having some mean and standard deviation.  This class supports the following
 * properties:
 *
 * 
 *
 *  Property: {@value #STANDARD_DEVIATION_PROPERTY}
 *       

 *      Default: {@value #DEFAULT_STANDARD_DEVIATION }
 *
 * 
This variable is the standard deviation used
 * when generating random numbers from a gaussian distribution
 *
 * 
 Property: {@value #MEAN_PROPERTY}
 *       

 *      Default: {@value #DEFAULT_MEAN}
 *
 * 
This variable is the mean used for generating
 * values from gaussian distribution for vectors
 *
 * 
 *
 * @author Keith Stevens
 */
public class GaussianVectorGenerator
        implements DoubleVectorGenerator, Serializable {

    private static final long serialVersionUID = 1L;

    /**
     * The base property prefix.
     */
    private static final String PROPERTY_PREFIX =
        "edu.ucla.sspace.index.GuassianVectorGenerator";

    /**
     * The property for setting the standard deviation.
     */
    public static final String STANDARD_DEVIATION_PROPERTY =
        PROPERTY_PREFIX + ".stdev";

    /**
     * The property for setting the standard deviation.
     */
    public static final String MEAN_PROPERTY =
        PROPERTY_PREFIX + ".mean";

    public static final double DEFAULT_STANDARD_DEVIATION = 1;

    public static final String DEFAULT_MEAN = "0";

    /**
     * The standard deviation used for generating a new index vector for terms.
     */
    private double stdev;

    private final int indexVectorLength;

    /**
     * The mean used each element in an generated {@link Vector}.
     */
    private double mean;

    /**
     * A random number generator which produces values for index vectors.
     */
    private Random randomGenerator;

    /**
     * Create a {@code GaussianVectorGenerator} that uses the system properties
     * for setup.
     */
    public GaussianVectorGenerator(int indexVectorLength) {
        this(indexVectorLength, System.getProperties());
    }

    /**
     * Create a {@code GaussianVectorGenerator} which uses {@code
     * vectorLength} as the size of each generated {@code Vector}.
     *
     * @param vectorLength The length of each index and semantic {@code Vector}
     *                     used in this {@code IndexVectorGenerator}.
     */
    public GaussianVectorGenerator(int indexVectorLength, Properties prop) {
        // Generate utility classes.
        randomGenerator = new Random();

        this.indexVectorLength = indexVectorLength;

        String stdevProp = prop.getProperty(
                STANDARD_DEVIATION_PROPERTY);
        stdev = (stdevProp != null)
            ? Double.parseDouble(stdevProp)
            : DEFAULT_STANDARD_DEVIATION;
        mean = Double.parseDouble(prop.getProperty(
                    MEAN_PROPERTY, DEFAULT_MEAN));
    }

    /**
     * Generate a new random vector using a guassian distribution for each
     * value.
     */
    public synchronized DoubleVector generate() {
        DoubleVector termVector = new DenseVector(indexVectorLength);
        for (int i = 0; i < indexVectorLength; i++)
            termVector.set(i, mean + (randomGenerator.nextGaussian() * stdev));
        return termVector;
    }
}