scala - PageRank 使用 GraphX

Question

我有一个名为 list.txt 的 .txt 文件，其中包含格式为源和目标 URL 的列表

google.de/2011/10/Extract-host       link.de/2011/10/extact-host
facebook.de/2014/11/photos           facebook.de/2014/11/name.jpg
community.cloudera.com/t5/           community.cloudera.com/t10/
facebook.de/2014/11/photos           link.de/2011/10/extact-host

在这篇文章的帮助下，How to create a VertexId in Apache Spark GraphX using a Long data type?我尝试创建节点和边，例如:

val test = sc.textFile("list.txt")                                                                                       //running

val arrayForm = test.map(_.split("\t"))                                                                            // running

val nodes: RDD[(VertexId, Option[String])] = arrayForm.flatMap(array => array).
map((_.toLong None))                                                                                                  

val edges: RDD[Edge[String]] = arrayForm.
map(line => Edge(line(0), line(1), "")) 

这里的问题是我真的不知道如何从字符串数据类型创建 VertexId 和类似的边。请让我知道如何解决这个问题。

Answer 1

答案是散列。由于您的 VertexID 是字符串，您可以使用 MurmurHash3 对它们进行哈希处理，制作图表，做您想做的事，然后将哈希值与原始字符串匹配。

示例代码

package com.void

import org.apache.spark._
import org.apache.spark.rdd.RDD
import org.apache.spark.graphx.Graph
import org.apache.spark.graphx.VertexId

import scala.util.hashing.MurmurHash3

object Main {

    def main( args: Array[ String ] ): Unit = {

        val conf = 
            new SparkConf()
            .setAppName( "SO Spark" )
            .setMaster( "local[*]" )
            .set( "spark.driver.host", "localhost" )

        val sc = new SparkContext( conf )

        val file = sc.textFile("data/pr_data.txt");

        val edgesRDD: RDD[(VertexId, VertexId)] = 
            file
            .map( line => line.split( "\t" ) )
            .map( line => (
                    MurmurHash3.stringHash( line( 0 ).toString ), MurmurHash3.stringHash( line( 1 ).toString )
                )
            )

        val graph = Graph.fromEdgeTuples( edgesRDD, 1 )

        // graph.triplets.collect.foreach( println )

        // println( "####" )

        val ranks = 
            graph
            .pageRank( 0.0001 )
            .vertices

        ranks.foreach( println )

        println( "####" )

        val identificationMap = 
            file
            .flatMap( line => line.split( "\t" ) )
            .distinct
            .map( line => ( MurmurHash3.stringHash( line.toString ).toLong, line ) )

        identificationMap.foreach( println )

        println( "####" )

        val fullMap = 
            ranks
            .join( identificationMap )

        fullMap.foreach( println )

        sc.stop()
    }
}

结果

(-1578471469,1.2982456140350878)
(1547760250,0.7017543859649124)
(1657711982,1.0000000000000002)
(1797439709,0.7017543859649124)
(996122257,0.7017543859649124)
(-1127017098,1.5964912280701753)
####
(1547760250,community.cloudera.com/t5/)
(-1127017098,link.de/2011/10/extact-host)
(1657711982,facebook.de/2014/11/name.jpg)
(1797439709,facebook.de/2014/11/photos)
(-1578471469,community.cloudera.com/t10/)
(996122257,google.de/2011/10/Extract-host)
####
(-1578471469,(1.2982456140350878,community.cloudera.com/t10/))
(1797439709,(0.7017543859649124,facebook.de/2014/11/photos))
(1547760250,(0.7017543859649124,community.cloudera.com/t5/))
(996122257,(0.7017543859649124,google.de/2011/10/Extract-host))
(1657711982,(1.0000000000000002,facebook.de/2014/11/name.jpg))
(-1127017098,(1.5964912280701753,link.de/2011/10/extact-host))

您可以通过将哈希 ID 映射出来从 RDD 中删除它们，但我相信 PageRank 不是您的最终目标，因此您稍后可能需要它们。