PageRank using GraphX

Question

I have a .txt file say list.txt which consists of list of source and destination URL in the format

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

With the help of this post, How to create a VertexId in Apache Spark GraphX using a Long data type? I tried to create node and edges like :

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), ""))

The problem here is I don't really know how to create VertexId and similarly edge from string datatype. Please let me know how to resolve this.

score 4 · Accepted Answer · answered Jan 14 '18 at 19:48

The answer is hashing. Since your VertexIDs are strings you can hash them using MurmurHash3, make a graph, do what you want to do and then match the hash values with original strings.

Example code

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()
    }
}

Results

(-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))

You can remove hashed IDs from the RDD by mapping them out but I believe that PageRank isn't your end goal so you'll probably need them later.

And what about collisions? If such graph has over 1 billion vertex the collisions will become noticeable, over 100billion vertex and you will have complete trash calc... — let4be, Mar 06 '20 at 17:01

PageRank using GraphX

1 Answers1