jasonwiki / recommend Goto Github PK

val userMatrixsRDD = itemsUserCollectionRDD.groupBy(_.1).flatMap(
      x =>
        for (
          ri <- x._2.iterator;
          rj <- x._2.iterator
          if ri.id < rj.id
        )
          yield {
            ((ri.id, rj.id), 1L
          }
    )

    val userAndUserMatrixCollection = userMatrixsRDD 
      .reduceByKey((x, y) => x + y)
      .map { x =>
        val ((uid, vid), counts) = x
        (uid, vid, similarity(uid, vid, counts) // similarity是自定义计算相似度的函数
      }