Paper Notes: Near linear time algorithm to detect community structures in large-scale networks

Paper title: Near linear time algorithm to detect community structures in large-scale networks

Author: UN Raghavan, R Albert, S Kumara

Year: 2007

Journal: Physical Review

My notes: 

This paper adopts the idea of label propagation in networks to find the community structure. Each node is first initialized with a unique label and then in each iterative process every node adopts the label that most of its neighbors have. 
The algorithm has synchronous and asynchronous versions according to two different label updating strategies. But synchronous method will lead to a label oscillation problem particularly in networks having bi-partite or nearly bi-partite structure. Thus asynchronous versions are used in this paper. 
Validations are performed on 6 networks:

Network	Size
Zachary’s karate club network	34
US college football network	115
Co-authorship network	16726
Actor collaboration network	374511
Protein-protein interaction network	2115
Word wide web	325729

The time complexity of the label propagation algorithm is near linear. Initialization takes O(n) time and each iteration of label propagation takes linear time O(m) ( n is the number of node and m is the number of edges in the network).