Scikit Learn CountVectorizer 的java实现

//汉字转拼音后，向量化文本，计算向量之间的余弦夹角

package sentenceSimilaryty;

import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;


public class calSimilaryty {


public double acalSimilary(String[] querys, String[] contexts) {
if (querys.length == 0) {
System.out.println("query string is null");
return -1;
}
if (contexts.length == 0) {
System.out.println("contextString is null");
return -1;
}
double score = 0;
if (contexts.length != querys.length) {
System.out.println("error arg,the length should be the same");
return 0;


}


List<String> wordList = new ArrayList<String>();
for (String i : querys)
wordList.add(i);
for (String i : contexts)