org.apache.flink.graph.examples

Class MusicProfiles

java.lang.Object

org.apache.flink.graph.examples.MusicProfiles

All Implemented Interfaces:

org.apache.flink.api.common.ProgramDescription

public class MusicProfiles
extends Object
implements org.apache.flink.api.common.ProgramDescription

This example demonstrates how to mix the DataSet Flink API with the Gelly API. The input is a set <userId - songId - playCount> triplets and a set of bad records, i.e. song ids that should not be trusted. Initially, we use the DataSet API to filter out the bad records. Then, we use Gelly to create a user -> song weighted bipartite graph and compute the top song (most listened) per user. Then, we use the DataSet API again, to create a user-user similarity graph, based on common songs, where users that are listeners of the same song are connected. A user-defined threshold on the playcount value defines when a user is considered to be a listener of a song. Finally, we use the graph API to run the label propagation community detection algorithm on the similarity graph.

The triplets input is expected to be given as one triplet per line, in the following format: "<userID>\t<songID>\t<playcount>".

The mismatches input file is expected to contain one mismatch record per line, in the following format: "ERROR: <songID trackID> song_title"

If no arguments are provided, the example runs with default data from MusicProfilesData.

Constructor Summary

Constructors

Constructor and Description
MusicProfiles()

Method Summary

Modifier and Type	Method and Description
String	getDescription()
static void	main(String[] args)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

MusicProfiles

public MusicProfiles()

Method Detail

main

public static void main(String[] args)
                 throws Exception

Throws:

Exception

getDescription

public String getDescription()

Specified by:

getDescription in interface org.apache.flink.api.common.ProgramDescription