org.apache.lucene.search.similarities.IBSimilarity Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of aem-sdk-api Show documentation
The Adobe Experience Manager SDK
There is a newer version: 2024.11.18751.20241128T090041Z-241100
/*
 * COPIED FROM APACHE LUCENE 4.7.2
 *
 * Git URL: [email protected]:apache/lucene.git, tag: releases/lucene-solr/4.7.2, path: lucene/core/src/java
 *
 * (see https://issues.apache.org/jira/browse/OAK-10786 for details)
 */

package org.apache.lucene.search.similarities;

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

import org.apache.lucene.search.Explanation;
import org.apache.lucene.search.similarities.Normalization.NoNormalization;

/**
 * Provides a framework for the family of information-based models, as described
 * in Stéphane Clinchant and Eric Gaussier. 2010. Information-based
 * models for ad hoc IR. In Proceeding of the 33rd international ACM SIGIR
 * conference on Research and development in information retrieval (SIGIR '10).
 * ACM, New York, NY, USA, 234-241.
 * The retrieval function is of the form RSV(q, d) = ∑
 * -x^q_w log Prob(X_w ≥
 * t^d_w | λ_w), where
 * 

 *   x^q_w is the query boost;
 *   X_w is a random variable that counts the occurrences
 *   of word w;
 *   t^d_w is the normalized term frequency;
 *   λ_w is a parameter.
 * 
 * 
 * The framework described in the paper has many similarities to the DFR
 * framework (see {@link DFRSimilarity}). It is possible that the two
 * Similarities will be merged at one point.
 * To construct an IBSimilarity, you must specify the implementations for 
 * all three components of the Information-Based model.
 * 

 *     {@link Distribution}: Probabilistic distribution used to
 *         model term occurrence
 *         
 *             {@link DistributionLL}: Log-logistic
 *             {@link DistributionLL}: Smoothed power-law
 *         
 *     
 *     {@link Lambda}: λ_w parameter of the
 *         probability distribution
 *         
 *             {@link LambdaDF}: N_w/N or average
 *                 number of documents where w occurs
 *             {@link LambdaTTF}: F_w/N or
 *                 average number of occurrences of w in the collection
 *         
 *     
 *     {@link Normalization}: Term frequency normalization 
 *         Any supported DFR normalization (listed in
 *                      {@link DFRSimilarity})
 *     
 * 
 * 
 * @see DFRSimilarity
 * @lucene.experimental 
 */
public class IBSimilarity extends SimilarityBase {
  /** The probabilistic distribution used to model term occurrence. */
  protected final Distribution distribution;
  /** The lambda (λ_w) parameter. */
  protected final Lambda lambda;
  /** The term frequency normalization. */
  protected final Normalization normalization;
  
  /**
   * Creates IBSimilarity from the three components.
   * 
   * Note that null values are not allowed:
   * if you want no normalization, instead pass 
   * {@link NoNormalization}.
   * @param distribution probabilistic distribution modeling term occurrence
   * @param lambda distribution's λ_w parameter
   * @param normalization term frequency normalization
   */
  public IBSimilarity(Distribution distribution,
                      Lambda lambda,
                      Normalization normalization) {
    this.distribution = distribution;
    this.lambda = lambda;
    this.normalization = normalization;
  }
  
  @Override
  protected float score(BasicStats stats, float freq, float docLen) {
    return stats.getTotalBoost() *
        distribution.score(
            stats,
            normalization.tfn(stats, freq, docLen),
            lambda.lambda(stats));
  }

  @Override
  protected void explain(
      Explanation expl, BasicStats stats, int doc, float freq, float docLen) {
    if (stats.getTotalBoost() != 1.0f) {
      expl.addDetail(new Explanation(stats.getTotalBoost(), "boost"));
    }
    Explanation normExpl = normalization.explain(stats, freq, docLen);
    Explanation lambdaExpl = lambda.explain(stats);
    expl.addDetail(normExpl);
    expl.addDetail(lambdaExpl);
    expl.addDetail(distribution.explain(
        stats, normExpl.getValue(), lambdaExpl.getValue()));
  }
  
  /**
   * The name of IB methods follow the pattern
   * {@code IB  }. The name of the
   * distribution is the same as in the original paper; for the names of lambda
   * parameters, refer to the javadoc of the {@link Lambda} classes.
   */
  @Override
  public String toString() {
    return "IB " + distribution.toString() + "-" + lambda.toString()
                 + normalization.toString();
  }
  
  /**
   * Returns the distribution
   */
  public Distribution getDistribution() {
    return distribution;
  }
  
  /**
   * Returns the distribution's lambda parameter
   */
  public Lambda getLambda() {
    return lambda;
  }

  /**
   * Returns the term frequency normalization
   */
  public Normalization getNormalization() {
    return normalization;
  }
}