edu.ucla.sspace.wordsi.Wordsi 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.vector.SparseDoubleVector;


/**
 * An interface for all Wordsi implementations.  A complete {@link Wordsi}
 * implementation will likely contain four parts: a {@link ContextExtractor}, a
 * clustering method, and a {@link ContextAssignmentMap}, and a {@link
 * AssignmentReporter}.  The {@link Context Extractor} will genrate context
 * vectors for a set of words within a given BufferedReader and call {@code
 * handleContextVector} for each context vector that is generated.  Each context
 * vector can be index by two keys: the primary key, which is generally the
 * focus word for the context vectors and the secondary key, which is either the
 * same as the focus word or an additional value such as a SenseEval/SemEval
 * instance identifier.  The {@link ContextAssignmentMap} is reponsible for
 * recording which secondary keys and context id's are assigned to each focus
 * term, in many cases, this is not neccesary, but if the exact clustering for
 * each context is required, one should use a {@link ContextAssignmentMap}.  The
 * clustering method will assign the context vector to some cluster, either
 * immediately or by storing the context vectors and performing a batch
 * clustering.  The {@link AssignmentReporter} is reponsible for reporting which
 * context vectors were assigned to which clusters.  The three major components
 * to {@link Wordsi} are separated so that each various context extraction
 * algorithms can be combined with various clustering algorithms and reporting
 * methods.  
 *
 * 
 *
 * Implementations are suggested to subclass {@link BaseWordsi}, since it
 * provides some methods for accepting and rejecting terms and dispatching the
 * {@link ContextExtractor}.
 *
 *
 * @see ContextExtractor
 * @see AssignmentReporter
 *
 * @author Keith Stevens
 */
public interface Wordsi {

    /**
     * Returns true if this {@link Wordsi} implementation should generate a
     * semantic vector for {@code word}.
     */
    boolean acceptWord(String word);

    /** Performs some operation with {@code contextVector}, which can be indexed
     * by either {@code primaryKey}, {@code secondaryKey}, or both.  This
     * operation will likely assign the {@code contextVector} to some cluster
     * immediately or store the {@code contextVector} so that it may be
     * clustered with all other other context vecetors generated for {@code
     * primaryKey}.
     *
     * 
     *
     * The {@code secondaryKey} does not need to be used, but some experiments
     * may require it, such as the SenseEval/SemEval evaluation or pseudo-word
     * disambiguation.  For SenseEval/SemEval evaluations, a {@link
     * SenseEvalContextExtractor} should be used, which will provide the context
     * id as the {@code secondaryKey}; reporting should be done with a {@link
     * SenseEvalReporter}.  For pseudo-word disambiguation/discrimination, a
     * {@link PseudoWordContextExtractor} should be used, which will create
     * pseudo-words for some set of tokens.  This extractor will use the
     * pseudo-word for the {@code primaryKey} and the original token as the
     * {@code secondaryKey}.
     *
     * @param primaryKey The primary key for {@code contextVector}
     * @param secondarykey A secondary key for {@code contextVector}
     * @param contextVector a {@code SparseDoubleVector} that represents a
     *        single context for a word
     */
    void handleContextVector(String primaryKey, String secondaryKey,
                             SparseDoubleVector contextVector);
}