JAVA代码之余弦相似度

计算两个文本向量的余弦相似度可以使用以下Java代码：

import java.util.*;

public class CosineSimilarity {
  public static void main(String[] args) {
    String[] doc1 = {"hello", "world", "java"};
    String[] doc2 = {"hello", "world", "python"};
    Map<String, Integer> tf1 = getTermFrequency(doc1);
    Map<String, Integer> tf2 = getTermFrequency(doc2);
    double cosineSimilarity = getCosineSimilarity(tf1, tf2);
    System.out.println("Cosine Similarity: " + cosineSimilarity);
  }
  
  public static Map<String, Integer> getTermFrequency(String[] doc) {
    Map<String, Integer> tf = new HashMap<>();
    for (String term : doc) {
      tf.put(term, tf.getOrDefault(term, 0) + 1);
    }
    return tf;
  }
  
  public static double getCosineSimilarity(Map<String, Integer> tf1, Map<String, Integer> tf2) {
    Set<String> terms = new HashSet<>(tf1.keySet());
    terms.addAll(tf2.keySet());
    
    double dotProduct = 0;
    double magnitude1 = 0;
    double magnitude2 = 0;
    
    for (String term : terms) {
      int count1 = tf1.getOrDefault(term, 0);
      int count2 = tf2.getOrDefault(term, 0);
      dotProduct += count1 * count2;
      magnitude1 += count1 * count1;
      magnitude2 += count2 * count2;
    }
    
    double magnitude = Math.sqrt(magnitude1) * Math.sqrt(magnitude2);
    return magnitude == 0 ? 0 : dotProduct / magnitude;
  }
}

在这个例子中，我们使用两个字符串数组作为文本向量。`getTermFrequency()`函数接收一个字符串数组并返回一个包含每个单词出现次数的映射。`getCosineSimilarity()`函数计算两个文本向量的余弦相似度。最后，我们在`main()`函数中调用这两个函数并打印输出余弦相似度。