JaroDistance、JaroWinklerDistance的Java实现

最新推荐文章于 2022-12-07 11:05:24 发布

_荣耀之路_

最新推荐文章于 2022-12-07 11:05:24 发布

阅读量1.3k

点赞数

文章标签： JaroDistance JaroWinklerDistance 字符串编辑距离相似度

本文链接：https://blog.csdn.net/asty9000/article/details/81359943

版权

Java 同时被 2 个专栏收录

32 篇文章 0 订阅

订阅专栏

算法

10 篇文章 0 订阅

订阅专栏

JaroDistance、JaroWinklerDistance算法定义见字符串编辑距离之JaroWinklerDistance，本文只是对算法的Java实现。

public class JaroWinklerDistance {

	private float p = 0.1f;
	private final float MAX_P = 0.25f;
	private final int MAX_L = 4;

	public JaroWinklerDistance() {
	}
	
	public JaroWinklerDistance(float p) {
		this.p=p>MAX_P?MAX_P:p;
	}
	/**
	 * 
	 * @param s1
	 * @param s2
	 * @return Jaro相似度
	 */
	public float JaroSimilarity(String s1, String s2) {
		if (s1 == null || s2 == null) {
			if (s1 == null && s2 == null) {
				return 1.0f;
			}
			return 0.0f;
		}
		int[] matchesInfo = getMatchesInfo(s1, s2);
		if (matchesInfo[0] == 0) {
			return 0.0f;
		} else {
			float m = matchesInfo[0];
			return ((m / s1.length() + m / s2.length() + (m - matchesInfo[1]) / m)) / 3;
		}
	}
	/**
	 * 
	 * @param s1
	 * @param s2
	 * @return Jaro-Winkler相似度
	 */
	public float JaroWinklerSimilarity(String s1, String s2) {
		if (s1 == null || s2 == null) {
			if (s1 == null && s2 == null) {
				return 1.0f;
			}
			return 0.0f;
		}
		int[] matchesInfo = getMatchesInfo(s1, s2);
		if (matchesInfo[0] == 0) {
			return 0.0f;
		} else {
			float m = matchesInfo[0];
			float j = ((m / s1.length() + m / s2.length() + (m - matchesInfo[1]) / m)) / 3;
			return j + matchesInfo[2] * p * (1 - j);
		}
	}

	private int[] getMatchesInfo(String s1, String s2) {
		String min, max;
		if (s1.length() < s2.length()) {
			min = s1;
			max = s2;
		} else {
			min = s2;
			max = s1;
		}
		//计算字符匹配的范围
		int matchRange = Math.max(max.length() / 2 - 1, 0);
		//记录两个字符串字符匹配状态
		boolean[] minMatchFlag = new boolean[min.length()];
		boolean[] maxMatchFlag = new boolean[max.length()];
		//匹配数量
		int matches = 0;
		for (int i = 0; i < min.length(); i++) {
			char minChar = min.charAt(i);
			for (int j = Math.max(i - matchRange, 0); j < Math.min(i + matchRange + 1, max.length()); j++) {
				if (!maxMatchFlag[j] && minChar == max.charAt(j)) {
					maxMatchFlag[j] = true;
					minMatchFlag[i] = true;
					matches++;
					break;
				}
			}
		}
		//转换次数，需要除以2
		int transpositions = 0;
		//共同前缀长度，最大为4
		int prefix = 0;
		//计算transpositions与prefix
		int j = 0;
		for (int i = 0; i < min.length(); i++) {
			if (minMatchFlag[i]) {
				for (; !maxMatchFlag[j]; j++) {
				}
				if (min.charAt(i) == max.charAt(j) && transpositions == 0) {
					prefix++;
				} else if (min.charAt(i) != max.charAt(j)) {
					transpositions++;
				}
				j++;
			}
		}
		return new int[] { matches, transpositions / 2, prefix > MAX_L ? MAX_L : prefix };
	}
}

简单测试

import org.junit.After;
import org.junit.Before;
import org.junit.Test;

public class JaroWinklerDistanceTest {

	@Before
	public void setUp() throws Exception {
	}

	@After
	public void tearDown() throws Exception {
	}

	@Test
	public void testJaroSimilarity() {
		String s1="aboard",s2="abroad",s3="aborad";
		JaroWinklerDistance jwd=new JaroWinklerDistance(0.25f);
		System.out.println("s1 s2 JaroSimilarity:"+jwd.JaroSimilarity(s1, s2));
		System.out.println("s1 s2 JaroWinklerSimilarity:"+jwd.JaroWinklerSimilarity(s1, s2));
		System.out.println("s1 s3 JaroSimilarity:"+jwd.JaroSimilarity(s1, s3));
		System.out.println("s1 s3 JaroWinklerSimilarity:"+jwd.JaroWinklerSimilarity(s1, s3));
	}

}