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

Abstract

Community detection and analysis is an important methodology for understanding the organization of various real-world networks and has applications in problems as diverse as consensus formation in social communities or the identification of functional modules in biochemical networks. Currently used algorithms that identify the community structures in large-scale real-world networks require a priori information such as the number and sizes of communities or are computationally expensive. In this paper we investigate a simple label propagation algorithm that uses the network structure alone as its guide and requires neither optimization of a pre-defined objective function nor prior information about the communities. In our algorithm every node is initialized with a unique label and at every step each node adopts the label that most of its neighbors currently have. In this iterative process densely connected groups of nodes form a consensus on a unique label to form communities. We validate the algorithm by applying it to networks whose community structures are known. We also demonstrate that the algorithm takes an almost linear time and hence it is computationally less expensive than what was possible so far.

Detecting community structure in complex networks via node similarity

“…The Zachary's karate club network [24] is one of the classic studies in social network analysis and has been used as one of the typical test examples by many researchers to detect community structures in complex network [9,15,17]. The club network consists of 34 member nodes, and splits in two smaller clubs after a dispute arose during the course of Zachary's study.…”

“…Another near linear time algorithm is proposed in Ref. [17], which based on label propagation and used only local information to analyze community structures, in time O(m + n) for largescale complex networks. In Ref.…”

Informative Value of Individual and Relational Data Compared Through Business-Oriented Community Detection

“…LabelPropagation [44] relies on a completely different method, based on the simulation of a propagation mechanism. All nodes are initially assigned a different label.…”

Semantic Social Network Analysis

“…On the other hand algorithms producing good community partition are time and/or space consuming and can't be applicable on very large graphs (millions of nodes). However, while modularity based algorithms like [Newman 2004a] are favoured for not too large networks, more and more heuristic based algorithms are proposed and exploit smart heuristics to handle community properties with a lower cost [Raghavan et al 2007]. [Radicchi et al 2004] has opened the door for detecting communities with heuristic based betweenness centralities.…”

“…An overview of random walk based algorithm is proposed in [Pons et al 2005]. The label propagation algorithm of [Raghavan et al 2007] is the most efficient algorithm in practice, but its ending is not deterministic (however in practice it always ends). Every node is given an initial random unique label representing the community to which it belongs.…”