package test;

import org.nlp.util.LineIterator;

import org.nlp.util.Tokenizer;

import org.nlp.vec.VectorModel;

import org.nlp.vec.Word2Vec;

import java.io.*;

import java.util.Collections;

import java.util.List;

import java.util.Set;

/**

* @author siegfang

*/

public class TestWord2Vec {

public static void readByJava(String textFilePath, String modelFilePath){

Word2Vec wv = new Word2Vec.Factory()

.setMethod(Word2Vec.Method.Skip_Gram)

.setNumOfThread(1).build();

try (BufferedReader br =

new BufferedReader(new FileReader(textFilePath))) {

int lineCount = 0;

for (String line = br.readLine(); line != null;

line = br.readLine()){

wv.readTokens(new Tokenizer(line, " "));

// System.out.println(line);

lineCount ++;

}

} catch (IOException ioe) {

ioe.printStackTrace();

}

wv.training();

wv.saveModel(new File(modelFilePath));

}

public static void testVector(String modelFilePath){

VectorModel vm = VectorModel.loadFromFile(modelFilePath);

Set<VectorModel.WordScore> result1 = Collections.emptySet();

result1 = vm.similar("亲");

for (VectorModel.WordScore we : result1){

System.out.println(we.name + " :\t" + we.score);

}

}

public static void main(String[] args){

String textFilePath = "D:/data/corpus.dat";

String modelFilePath = "D:/data/corpus.nn";

readByJava(textFilePath, modelFilePath);

testVector(modelFilePath);

}

}