Carlo Piccardi
FINDING AND TESTING NETWORK COMMUNITIES BY LUMPED MARKOV CHAINS
PLoS ONE, 6(11), e27028, 2011




Copyright:
2011, Carlo Piccardi, Politecnico di Milano, Italy
[see licence notes in the distribution file]

e-mail carlo.piccardi@polimi.it
web page http://home.deib.polimi.it/piccardi/

Last updated: Nov 7, 2011

PAPER

You can download the main text of the paper and the Supporting Information file directly from the PLoS ONE website.



CODES AND DATA

The distribution file FTNCbyLMC.zip contains the Matlab codes and datafiles used to produce the results and the figures of the paper (main text and supporting information). Please refer to the paper for all details about the algorithms and the data.


The codes and data can readily be adapted and used for your own application (I hope so, at least). Please report any bug or misfunctioning. If your research benefits of these methods and algorithms, please let me know (and cite the paper...).

The zip file contains:
  • FindingCommunities.m : fully automated community detection - computes the node distances, performs hierarchical analysis, generates a set of meaningful partitions and, for each one of them, computes the set of persistence probabilities.
  • TestingCommunities.m : computes the persistence probabilities of a single, "a priori" given partition.
The following datafiles are included:
  • A_toy12.mat : the toy network of Figs. 1 and 2.
  • A_LFRuu_beta1mu025.mat : undirected, unweighed LFR benchmark (Fig. 4, top-left panel).
  • A_LFRuu_beta2mu06.mat : undirected, unweighed LFR benchmark (Fig. 5, top-left panel).
  • A_netscience_gc.mat : giant component of the Netscience network (Fig. 6).
  • A_wtn2008_gc.mat : giant component of the world trade network in 2008 (Fig. 8).
  • A_ErdosRenyi.mat : an instance of an Erdos-Renyi network (Fig. S1.1).
  • A_Zachary.mat : Zachary's karate club network (Fig. S1.2).
  • A_LFdw_mu03.mat : directed, weighted LFR benchmark (Fig. S1.3, top panel).
  • A_LFdw_mu06.mat : directed, weighted LFR benchmark (Fig. S1.3, bottom panel).
  • A_linkrank.mat : LinkRank benchmark network (Fig. S1.5).
  • A_neural_gc : giant component of the neural network of the Caenorhabditis elegans (Fig. S1.5).
  • com_3_toy12.mat : an example of a file describing a partition, to be used as input for TestingCommunities.m. It contains the "natural" 3-cluster partition of the net A_toy12.mat (Figs. 1 and 2).



SOFTWARE REQUIREMENTS

The .m codes work correctly with Matlab 7.11.0 operating under Windows 7 (but they probably work on previous versions as well).
FindingCommunies.m requires the availability of the Statistics Toolbox for the hierarchical analysis.