实时推荐-3商品标签训练模型

package com.tuijian

import org.apache.spark.mllib.feature.Word2Vec
import org.apache.spark.sql.{SaveMode, SparkSession}
import org.jblas.DoubleMatrix

object HandleGood {

  /***
    * 每件商品和其它商品之间的差距(两个向量之间的余弦度数)
    * 取出前10件商品
    * @param args
    */
  def main(args: Array[String]): Unit = {

    val spark = SparkSession.builder()
      .appName("handle")
      .master("local")
      .getOrCreate()
    val goodData = SqlUtil.read(spark, "goods")

    val tagData = SqlUtil.read(spark, "good_tag")


//    //两表结合在一起
//    val goodTagData = goodData.join(tagData, "id")

    //需要将标签转换为向量

    val words = tagData.rdd.map(
      tag => {
        tag.getString(1).split("\\|").toBuffer
      }
    )


    val wv = new Word2Vec()
    wv.setMinCount(1)
    wv.setVectorSize(3)
    val model = wv.fit(words) //训练模型


    import spark.implicits._
    val goodTagRDD = goodData
      .join(tagDa