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.text.similarity; 018 019import java.util.Arrays; 020 021import org.apache.commons.lang3.StringUtils; 022 023/** 024 * A similarity algorithm indicating the percentage of matched characters between two character sequences. 025 * 026 * <p> 027 * The Jaro measure is the weighted sum of percentage of matched characters 028 * from each file and transposed characters. Winkler increased this measure 029 * for matching initial characters. 030 * </p> 031 * 032 * <p> 033 * This implementation is based on the Jaro Winkler similarity algorithm 034 * from <a href="http://en.wikipedia.org/wiki/Jaro%E2%80%93Winkler_distance"> 035 * http://en.wikipedia.org/wiki/Jaro%E2%80%93Winkler_distance</a>. 036 * </p> 037 * 038 * <p> 039 * This code has been adapted from Apache Commons Lang 3.3. 040 * </p> 041 * 042 * @since 1.7 043 */ 044public class JaroWinklerSimilarity implements SimilarityScore<Double> { 045 046 /** 047 * This method returns the Jaro-Winkler string matches, half transpositions, prefix array. 048 * 049 * @param first the first string to be matched 050 * @param second the second string to be matched 051 * @return mtp array containing: matches, half transpositions, and prefix 052 */ 053 protected static int[] matches(final CharSequence first, final CharSequence second) { 054 final CharSequence max; 055 final CharSequence min; 056 if (first.length() > second.length()) { 057 max = first; 058 min = second; 059 } else { 060 max = second; 061 min = first; 062 } 063 final int range = Math.max(max.length() / 2 - 1, 0); 064 final int[] matchIndexes = new int[min.length()]; 065 Arrays.fill(matchIndexes, -1); 066 final boolean[] matchFlags = new boolean[max.length()]; 067 int matches = 0; 068 for (int mi = 0; mi < min.length(); mi++) { 069 final char c1 = min.charAt(mi); 070 for (int xi = Math.max(mi - range, 0), xn = Math.min(mi + range + 1, max.length()); xi < xn; xi++) { 071 if (!matchFlags[xi] && c1 == max.charAt(xi)) { 072 matchIndexes[mi] = xi; 073 matchFlags[xi] = true; 074 matches++; 075 break; 076 } 077 } 078 } 079 final char[] ms1 = new char[matches]; 080 final char[] ms2 = new char[matches]; 081 for (int i = 0, si = 0; i < min.length(); i++) { 082 if (matchIndexes[i] != -1) { 083 ms1[si] = min.charAt(i); 084 si++; 085 } 086 } 087 for (int i = 0, si = 0; i < max.length(); i++) { 088 if (matchFlags[i]) { 089 ms2[si] = max.charAt(i); 090 si++; 091 } 092 } 093 int halfTranspositions = 0; 094 for (int mi = 0; mi < ms1.length; mi++) { 095 if (ms1[mi] != ms2[mi]) { 096 halfTranspositions++; 097 } 098 } 099 int prefix = 0; 100 for (int mi = 0; mi < Math.min(4, min.length()); mi++) { 101 if (first.charAt(mi) != second.charAt(mi)) { 102 break; 103 } 104 prefix++; 105 } 106 return new int[] {matches, halfTranspositions, prefix}; 107 } 108 109 /** 110 * Computes the Jaro Winkler Similarity between two character sequences. 111 * 112 * <pre> 113 * sim.apply(null, null) = IllegalArgumentException 114 * sim.apply("foo", null) = IllegalArgumentException 115 * sim.apply(null, "foo") = IllegalArgumentException 116 * sim.apply("", "") = 1.0 117 * sim.apply("foo", "foo") = 1.0 118 * sim.apply("foo", "foo ") = 0.94 119 * sim.apply("foo", "foo ") = 0.91 120 * sim.apply("foo", " foo ") = 0.87 121 * sim.apply("foo", " foo") = 0.51 122 * sim.apply("", "a") = 0.0 123 * sim.apply("aaapppp", "") = 0.0 124 * sim.apply("frog", "fog") = 0.93 125 * sim.apply("fly", "ant") = 0.0 126 * sim.apply("elephant", "hippo") = 0.44 127 * sim.apply("hippo", "elephant") = 0.44 128 * sim.apply("hippo", "zzzzzzzz") = 0.0 129 * sim.apply("hello", "hallo") = 0.88 130 * sim.apply("ABC Corporation", "ABC Corp") = 0.91 131 * sim.apply("D N H Enterprises Inc", "D & H Enterprises, Inc.") = 0.95 132 * sim.apply("My Gym Children's Fitness Center", "My Gym. Childrens Fitness") = 0.92 133 * sim.apply("PENNSYLVANIA", "PENNCISYLVNIA") = 0.88 134 * </pre> 135 * 136 * @param left the first CharSequence, must not be null 137 * @param right the second CharSequence, must not be null 138 * @return result similarity 139 * @throws IllegalArgumentException if either CharSequence input is {@code null} 140 */ 141 @Override 142 public Double apply(final CharSequence left, final CharSequence right) { 143 final double defaultScalingFactor = 0.1; 144 145 if (left == null || right == null) { 146 throw new IllegalArgumentException("CharSequences must not be null"); 147 } 148 149 if (StringUtils.equals(left, right)) { 150 return 1d; 151 } 152 153 final int[] mtp = matches(left, right); 154 final double m = mtp[0]; 155 if (m == 0) { 156 return 0d; 157 } 158 final double j = (m / left.length() + m / right.length() + (m - (double) mtp[1] / 2) / m) / 3; 159 return j < 0.7d ? j : j + defaultScalingFactor * mtp[2] * (1d - j); 160 } 161 162}