001/* 002 * Licensed to the Apache Software Foundation (ASF) under one or more 003 * contributor license agreements. See the NOTICE file distributed with 004 * this work for additional information regarding copyright ownership. 005 * The ASF licenses this file to You under the Apache License, Version 2.0 006 * (the "License"); you may not use this file except in compliance with 007 * the License. You may obtain a copy of the License at 008 * 009 * http://www.apache.org/licenses/LICENSE-2.0 010 * 011 * Unless required by applicable law or agreed to in writing, software 012 * distributed under the License is distributed on an "AS IS" BASIS, 013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 014 * See the License for the specific language governing permissions and 015 * limitations under the License. 016 */ 017package org.apache.commons.collections4.bloomfilter; 018 019import java.util.function.LongBinaryOperator; 020 021/** 022 * Implementations of set operations on BitMapExtractors. 023 * 024 * @since 4.5 025 */ 026public final class SetOperations { 027 028 /** 029 * Calculates the cardinality of the logical {@code AND} of the bit maps for the two filters. 030 * @param first the first BitMapExtractor. 031 * @param second the second BitMapExtractor 032 * @return the cardinality of the {@code AND} of the filters. 033 */ 034 public static int andCardinality(final BitMapExtractor first, final BitMapExtractor second) { 035 return cardinality(first, second, (x, y) -> x & y); 036 } 037 038 /** 039 * Calculates the cardinality of a BitMapExtractor. By necessity this method will visit each bit map 040 * created by the bitMapExtractor. 041 * @param bitMapExtractor the extractor to calculate the cardinality for. 042 * @return the cardinality of the bit maps produced by the bitMapExtractor. 043 */ 044 public static int cardinality(final BitMapExtractor bitMapExtractor) { 045 final int[] cardinality = new int[1]; 046 bitMapExtractor.processBitMaps(l -> { 047 cardinality[0] += Long.bitCount(l); 048 return true; 049 }); 050 return cardinality[0]; 051 } 052 053 /** 054 * Calculates the cardinality of the result of a LongBinaryOperator using the 055 * {@code BitMapExtractor.makePredicate} method. 056 * @param first the first BitMapExtractor 057 * @param second the second BitMapExtractor 058 * @param op a long binary operation on where x = {@code first} and y = {@code second} bitmap extractors. 059 * @return the calculated cardinality. 060 */ 061 private static int cardinality(final BitMapExtractor first, final BitMapExtractor second, final LongBinaryOperator op) { 062 final int[] cardinality = new int[1]; 063 064 first.processBitMapPairs(second, (x, y) -> { 065 cardinality[0] += Long.bitCount(op.applyAsLong(x, y)); 066 return true; 067 }); 068 return cardinality[0]; 069 } 070 071 /** 072 * Calculates the Cosine distance between two BitMapExtractor. 073 * 074 * <p>Cosine distance is defined as {@code 1 - Cosine similarity}</p> 075 * 076 * @param first the first BitMapExtractor. 077 * @param second the second BitMapExtractor. 078 * @return the jaccard distance. 079 */ 080 public static double cosineDistance(final BitMapExtractor first, final BitMapExtractor second) { 081 return 1.0 - cosineSimilarity(first, second); 082 } 083 084 /** 085 * Calculates the Cosine similarity between two BitMapExtractors. 086 * <p> Also known as Orchini similarity and the Tucker coefficient of congruence or 087 * Ochiai similarity.</p> 088 * 089 * <p>If either extractor is empty the result is 0 (zero)</p> 090 * 091 * @param first the first BitMapExtractor. 092 * @param second the second BitMapExtractor. 093 * @return the Cosine similarity. 094 */ 095 public static double cosineSimilarity(final BitMapExtractor first, final BitMapExtractor second) { 096 final int numerator = andCardinality(first, second); 097 // Given that the cardinality is an int then the product as a double will not 098 // overflow, we can use one sqrt: 099 return numerator == 0 ? 0 : numerator / Math.sqrt(cardinality(first) * cardinality(second)); 100 } 101 102 /** 103 * Calculates the Cosine similarity between two Bloom filters. 104 * <p> Also known as Orchini similarity and the Tucker coefficient of congruence or 105 * Ochiai similarity.</p> 106 * 107 * <p>If either filter is empty (no enabled bits) the result is 0 (zero)</p> 108 * 109 * <p>This is a version of cosineSimilarity optimized for Bloom filters.</p> 110 * 111 * @param first the first Bloom filter. 112 * @param second the second Bloom filter. 113 * @return the Cosine similarity. 114 */ 115 public static double cosineSimilarity(final BloomFilter first, final BloomFilter second) { 116 final int numerator = andCardinality(first, second); 117 // Given that the cardinality is an int then the product as a double will not 118 // overflow, we can use one sqrt: 119 return numerator == 0 ? 0 : numerator / Math.sqrt(first.cardinality() * second.cardinality()); 120 } 121 122 /** 123 * Calculates the Hamming distance between two BitMapExtractors. 124 * 125 * @param first the first BitMapExtractor. 126 * @param second the second BitMapExtractor. 127 * @return the Hamming distance. 128 */ 129 public static int hammingDistance(final BitMapExtractor first, final BitMapExtractor second) { 130 return xorCardinality(first, second); 131 } 132 133 /** 134 * Calculates the Jaccard distance between two BitMapExtractor. 135 * 136 * <p>Jaccard distance is defined as {@code 1 - Jaccard similarity}</p> 137 * 138 * @param first the first BitMapExtractor. 139 * @param second the second BitMapExtractor. 140 * @return the Jaccard distance. 141 */ 142 public static double jaccardDistance(final BitMapExtractor first, final BitMapExtractor second) { 143 return 1.0 - jaccardSimilarity(first, second); 144 } 145 146 /** 147 * Calculates the Jaccard similarity between two BitMapExtractor. 148 * 149 * <p>Also known as Jaccard index, Intersection over Union, and Jaccard similarity coefficient</p> 150 * 151 * @param first the first BitMapExtractor. 152 * @param second the second BitMapExtractor. 153 * @return the Jaccard similarity. 154 */ 155 public static double jaccardSimilarity(final BitMapExtractor first, final BitMapExtractor second) { 156 final int[] cardinality = new int[2]; 157 first.processBitMapPairs(second, (x, y) -> { 158 cardinality[0] += Long.bitCount(x & y); 159 cardinality[1] += Long.bitCount(x | y); 160 return true; 161 }); 162 final int intersection = cardinality[0]; 163 return intersection == 0 ? 0 : intersection / (double) cardinality[1]; 164 } 165 166 /** 167 * Calculates the cardinality of the logical {@code OR} of the bit maps for the two filters. 168 * @param first the first BitMapExtractor. 169 * @param second the second BitMapExtractor 170 * @return the cardinality of the {@code OR} of the filters. 171 */ 172 public static int orCardinality(final BitMapExtractor first, final BitMapExtractor second) { 173 return cardinality(first, second, (x, y) -> x | y); 174 } 175 176 /** 177 * Calculates the cardinality of the logical {@code XOR} of the bit maps for the two filters. 178 * @param first the first BitMapExtractor. 179 * @param second the second BitMapExtractor 180 * @return the cardinality of the {@code XOR} of the filters. 181 */ 182 public static int xorCardinality(final BitMapExtractor first, final BitMapExtractor second) { 183 return cardinality(first, second, (x, y) -> x ^ y); 184 } 185 186 /** 187 * Do not instantiate. 188 */ 189 private SetOperations() { 190 } 191}