Paper Notes: Fast unfolding of communities in large networks

Paper title: Fast unfolding of communities in large networks

Author: Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte and Etienne Lefebvre

Journal: Journal of Statistical Mechanics: Theory and Experiment

Year: 2008

All the code in this paper can be downloaded from: http://findcommunities.googlepages.com

My notes:

This paper proposed a heuristic method which is based on modularity maximization. The algorithm is an agglomerative method that consists of two phases: (1) each node belongs to a unique community. Then for each node i, calculate the gain of modularity if i is removed from its community and join to community of j (j is the neighbor of i). Then i is placed to j’s community in which the gain is maximum and positive. If no positive gain is found, i stays in its original community. This process is iterated for all nodes until no improvement can be achieved. (2) build a network whose nodes are the communities discovered in the first phase. Then reapply the first phase until no more changes happen and a maximum of modularity is obtained. 
The computation of this method is shown to be the least among known community detection methods.