Lucene4.3.1字符串距离接口StringDistance实现之JaroWinklerDistance源码解析

版权声明：本文为博主原创文章，未经博主允许不得转载。 https://blog.csdn.net/asty9000/article/details/81349129

JaroWinklerDistance是JaroWinklerDistance算法的实现，在原本JaroWinklerDistance算法的基础上做了一些小的调整。在计算Jaro-Winkler相似度时与JaroWinklerDistance算法不同的是其在计算时使用的系数取值为两个字符串最大长度的倒数，最大为0.1。

package org.apache.lucene.search.spell;

import java.util.Arrays;


public class JaroWinklerDistance implements StringDistance {

  private float threshold = 0.7f;
  
  public JaroWinklerDistance() {}

  private int[] matches(String s1, String s2) {
    String max, min;
    if (s1.length() > s2.length()) {
      max = s1;
      min = s2;
    } else {
      max = s2;
      min = s1;
    }
    //计算字符匹配范围
    int range = Math.max(max.length() / 2 - 1, 0);
    //记录与min字符串对应位置字符匹配的字符在max字符串中的位置
    int[] matchIndexes = new int[min.length()];
    Arrays.fill(matchIndexes, -1);
    //记录max字符串对应位置字符是否已被匹配
    boolean[] matchFlags = new boolean[max.length()];
    //记录两个字符串匹配的字符数
    int matches = 0;
    for (int mi = 0; mi < min.length(); mi++) {
      char c1 = min.charAt(mi);
      for (int xi = Math.max(mi - range, 0), xn = Math.min(mi + range + 1, max
          .length()); xi < xn; xi++) {
        if (!matchFlags[xi] && c1 == max.charAt(xi)) {
          matchIndexes[mi] = xi;
          matchFlags[xi] = true;
          matches++;
          break;
        }
      }
    }
    //ms1,ms2分别存储min与max中匹配的字符
    char[] ms1 = new char[matches];
    char[] ms2 = new char[matches];
    for (int i = 0, si = 0; i < min.length(); i++) {
      if (matchIndexes[i] != -1) {
        ms1[si] = min.charAt(i);
        si++;
      }
    }
    for (int i = 0, si = 0; i < max.length(); i++) {
      if (matchFlags[i]) {
        ms2[si] = max.charAt(i);
        si++;
      }
    }
    //计算min、max相匹配的字符需要换位的字符数，计算结果需除2
    int transpositions = 0;
    for (int mi = 0; mi < ms1.length; mi++) {
      if (ms1[mi] != ms2[mi]) {
        transpositions++;
      }
    }
    //计算min、max相匹配的字符的共同前缀的长度
    int prefix = 0;
    for (int mi = 0; mi < min.length(); mi++) {
      if (s1.charAt(mi) == s2.charAt(mi)) {
        prefix++;
      } else {
        break;
      }
    }
    return new int[] { matches, transpositions / 2, prefix, max.length() };
  }

  @Override
  public float getDistance(String s1, String s2) {
    int[] mtp = matches(s1, s2);
    float m = mtp[0];
    if (m == 0) {
      return 0f;
    }
    //计算Jaro Distance
    float j = ((m / s1.length() + m / s2.length() + (m - mtp[1]) / m)) / 3;
    //如果达到设置的阈值则使用Jaro-Winkler Distance
    float jw = j < getThreshold() ? j : j + Math.min(0.1f, 1f / mtp[3]) * mtp[2]
        * (1 - j);
    return jw;
  }

  public void setThreshold(float threshold) {
    this.threshold = threshold;
  }

  public float getThreshold() {
    return threshold;
  }

  @Override
  public int hashCode() {
    return 113 * Float.floatToIntBits(threshold) * getClass().hashCode();
  }

  @Override
  public boolean equals(Object obj) {
    if (this == obj) return true;
    if (null == obj || getClass() != obj.getClass()) return false;
    
    JaroWinklerDistance o = (JaroWinklerDistance)obj;
    return (Float.floatToIntBits(o.threshold) 
            == Float.floatToIntBits(this.threshold));
  }

  @Override
  public String toString() {
    return "jarowinkler(" + threshold + ")";
  }

}