edu.ucla.sspace.wordsi.GeneralContextExtractor 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 2010 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.wordsi;

import edu.ucla.sspace.text.IteratorFactory;

import edu.ucla.sspace.vector.SparseDoubleVector;

import java.io.BufferedReader;

import java.util.ArrayDeque;
import java.util.Iterator;
import java.util.Queue;


/**
 * A general purpose {@link ContextExtractor}.  This extractor assumes that
 * documents are simply raw text and contexts should be defined by word
 * co-occurrences.  This class depends on a {@link ContextGenerator} for
 * generating context vectors.
 *
 * @author Keith Stevens
 */
public class GeneralContextExtractor implements ContextExtractor {

    /**
     * The generator responsible for creating context vectors from a sliding
     * window of text.
     */
    private final ContextGenerator generator;

    /**
     * The size of the sliding window of text, in a single direction.
     */
    private final int windowSize;

    /**
     * Set to true if the first token should be considered the header of the
     * context and be discarded..
     */
    private final boolean readHeader;

    /**
     * Creates a new {@link GeneralContextExtracto}.
     *
     * @param generator The {@link ContextGenerator} responsible for creating
     *        context vectors
     * @param windowSize The number of words before and after the focus word
     *        which compose a context
     */
    public GeneralContextExtractor(ContextGenerator generator,
                                   int windowSize,
                                   boolean readHeader) {
        this.generator = generator;
        this.windowSize = windowSize;
        this.readHeader = readHeader;
    }

    /**
     * {@inheritDoc}
     */
    public int getVectorLength() {
        return generator.getVectorLength();
    }

    /**
     * {@inheritDoc}
     */
    public void processDocument(BufferedReader document, Wordsi wordsi) {
        Queue prevWords = new ArrayDeque();
        Queue nextWords = new ArrayDeque();

        Iterator it = IteratorFactory.tokenizeOrdered(document);

        // Skip empty documents.
        if (!it.hasNext())
            return;

        // Read the header and use it as the secondary key for wordsi, if told
        // to do so.
        String header = null;
        if (readHeader)
            header = it.next();

        // Fill up the words after the context so that when the real processing
        // starts, the context is fully prepared.
        for (int i = 0 ; i < windowSize && it.hasNext(); ++i)
            nextWords.offer(it.next());

        // Iterate through each of the words in the context, generating context
        // vectors for each acceptable word.
        String focus = null;
        while (!nextWords.isEmpty()) {
            focus = nextWords.remove();
            String secondaryKey = (header == null) ? focus : header;

            // Advance the sliding window to the right.
            if (it.hasNext())
                nextWords.offer(it.next());

            // Represent the word if wordsi is willing to process it.
            if (wordsi.acceptWord(focus)) {
                SparseDoubleVector contextVector = generator.generateContext(
                        prevWords, nextWords);
                wordsi.handleContextVector(focus, secondaryKey, contextVector);
            }

            // Advance the sliding window to the right.
            prevWords.offer(focus);
            if (prevWords.size() > windowSize)
                prevWords.remove();
        }
    }
}