Base class for all community detection algorithms. More...

#include <cluster.h>

Inheritance diagram for ClusterAlg< R >:

Public Member Functions
virtual void	calculate ()=0
float	get_Density (bidirected_t::graph graph, const set< bidirected_t::vertex > cluster)
float	get_Density (set< lui > *cluster)
bool	ReadGraph (ifstream *fin, string delimiters=" ,\t", double alpha=1, double beta=1)
bool	PrintGraph (ofstream *fout)
bool	PrintCommunities (string filename)
void	PrintCommunity (set< lui > *comm)
bool	PrintTranslatedCommunities (string filename)
void	PrintTranslatedCommunity (set< lui > *comm)
void	TranslateCommunities ()
void	setOutput (string filename)
Static Public Member Functions
static ClusterAlg *	Create (string type, string args, double alpha, double beta, double weightThresh, bool translate)
Public Attributes
map< lui, map< lui, R > >	network
vector< set< lui > >	communities
map< string, lui >	idmap
map< lui, string >	idLookup
string	args
bool	translate
ostream *	fileout
Static Public Attributes
static density< bidirected_t > *	dens

Detailed Description

template<typename R>
class ClusterAlg< R >

Base class for all community detection algorithms.

Holds general purpose functions for all algorithms, but main purpose is as a factory, to create the correct class at runtime.

R The type of weight used in the network (usually int or double - using double is safer)

Member Function Documentation

template<typename R >

virtual void ClusterAlg< R >::calculate ( ) [pure virtual]

Will be the actual algorithm implementation.

Implemented in IterativeDensity< R >, CliqueCluster< R >, SLPA< R >, CONGA< R >, RRW< R >, OSLOM< R >, COPRA< R >, SSDE< R >, and LCOMM< R >.

template<typename R >

template ClusterAlg< double > * ClusterAlg< R >::Create	(	string	type,
		string	args,
		double	alpha,
		double	beta,
		double	weightThresh,
		bool	translate
	)		`[static]`

Factory function

Parameters:

type	Algorithm to create class for
argList	Arguments to use for algorithm
alpha	Factor for positive weights
beta	Factor for negative weights
weightThresh	Minimum edge weight

Returns:: Cluster algorithm set up to implement 'type' algorithm

Factory function

Parameters:

type	Algorithm to create class for
argList	Arguments to use for algorithm
alpha	Factor for positive weights
beta	Factor for negative weights
weightThresh	Minimum edge weight
translate	Flag indicating whether or not to use original vertex names (from file)

Returns:: Class set up to implement 'type' detection algorithm

template<typename R >

float ClusterAlg< R >::get_Density	(	bidirected_t::graph *	graph,
		const set< bidirected_t::vertex > *	cluster
	)		`[inline]`

Retrieve the density of a set of vertices in a graph using boost structures.

Parameters:

graph	The network the cluster is from
cluster	Set of vertices in the cluster

Returns:: Cluster density value

template<typename R >

float ClusterAlg< R >::get_Density ( set< lui > * cluster ) [inline]

Retrieve density of a set of vertices

Parameters:

cluster IDs of vertices in the community

Returns:: Cluster density value

template<typename R >

bool ClusterAlg< R >::PrintCommunities ( string filename ) [inline]

Print detected communities to file. Each line in the output file is as follows:

CommunityDensity CommunitySize Vertex1ID Vertex2ID Vertex3ID ...

Parameters:

filename String representing filename to print communities to

Returns:: true

Reimplemented in IterativeDensity< R >.

template<typename R >

void ClusterAlg< R >::PrintCommunity ( set< lui > * comm ) [inline]

Print a single community to the out file

CommunityDensity CommunitySize Vertex1ID Vertex2ID Vertex3ID ...

Parameters:

comm	Set of vertices to print out

template<typename R >

bool ClusterAlg< R >::PrintGraph ( ofstream * fout ) [inline]

Print the network to a file

Parameters:

fout	Stream to print file to

Returns:: true

template<typename R >

bool ClusterAlg< R >::PrintTranslatedCommunities ( string filename ) [inline]

Print detected communities to file. Each line in the output file is as follows:

CommunityDensity CommunitySize Vertex1 Vertex2 Vertex3 ...

Parameters:

filename String representing filename to print communities to

Returns:: true

Reimplemented in IterativeDensity< R >.

template<typename R >

void ClusterAlg< R >::PrintTranslatedCommunity ( set< lui > * comm ) [inline]

Print a single community to the out file (with translated names)

CommunityDensity CommunitySize Vertex1_OriginalID Vertex2_OriginalID Vertex3_OriginalID ...

Parameters:

comm	Set of vertices to print out

template<typename R >

bool ClusterAlg< R >::ReadGraph	(	ifstream *	fin,
		string	delimiters = `" ,\t"`,
		double	alpha = `1`,
		double	beta = `1`
	)		`[inline]`

Read network from a file and save. Each vertex read in is given a numerical ID that it used throughout the detection process. (Easier comparison between numbers than strings). This can be translated back later.

Parameters:

fin	Stream to read network from
delimiters	Delimiters between the vertex IDs and edge weights in the network file.

Returns:: true

template<typename R >

void ClusterAlg< R >::TranslateCommunities ( ) [inline]

As long as the original vertex IDs were also numerical (integers), this function will retrieve the original vertex IDs without printing out the communities. Useful for comparing results to benchmark 'ground truth's.

Reimplemented in IterativeDensity< R >.