使用spark mllib 随机森林算法对文本进行多分类

1、数据准备

20W人工标注文本数据，样本如下：

1#k-v#*亮亮爱宠*波波宠物指甲钳指甲剪附送锉刀适用小型犬及猫特价
1#k-v#*顺丰包邮*宠物药品圣马利诺PowerIgG免疫力球蛋白犬猫细小病毒
1#k-v#*包邮*法国罗斯蔓草本精华宠物浴液薰衣草护色润泽香波拍套餐
1#k-v#*包邮*家朵102宠物沐浴液
1#k-v#*包邮*家朵102宠物沐浴液猫

2、分词

使用ansj包对文本数据去除停用词分词。代码如下：

import java.io.File;
import java.io.IOException;
import java.util.HashSet;
import java.util.List;
import java.util.Set;

import org.ansj.domain.Result;
import org.ansj.domain.Term;
import org.ansj.splitWord.analysis.ToAnalysis;
import org.apache.commons.io.FileUtils;
import org.apache.commons.lang3.StringUtils;

public class Seg{

    private static Set<String> stopwords = new HashSet<String>();
    static{
        File f = new File("");
        try {
            List<String> lines = FileUtils.readLines(f);
            for(String str : lines){
                stopwords.add(str);
            }
        } catch (IOException e) {
            e.printStackTrace();
        }


    }

    public static void main(String[] args) throws IOException {
        File f = new File("");
        File resultFile = new File("");

        List<String> lists = FileUtils.readLines(f);

        int count = 0;
        for(String str : lists){
            count++;
            String index = str.split("#k-v#")[0];
//          System.out.println(count + " " + Integer.parseInt(index));

            Result res = ToAnalysis.parse(str.split("#k-v#")[1]);

            List<Term> terms  = res.getTerms();

            String wordStr = "";
            for(Term t : terms){
                String word = t.getName();
                if(word.length()>1&&!stopwords.contains(word)){
                    wordStr = wordStr + " " +  word;
                }
            }
            if(!StringUtils.isEmpty(wordStr)){
                FileUtils.write(resultFile, index + "#k-v#" + wordStr + "\n" , true);
            }