Physica A 462 (2016) 386–395
Contents lists available at ScienceDirect
Physica A
journal homepage: www.elsevier.com/locate/physa
Weighted modularity optimization for crisp and fuzzy
community detection in large-scale networks
Jie Cao
a
, Zhan Bu
a,∗
, Guangliang Gao
b
, Haicheng Tao
b
a
Jiangsu Provincial Key Laboratory of E-Business, Nanjing University of Finance and Economics, Nanjing, China
b
College of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, China
a r t i c l e i n f o
Article history:
Received 14 March 2016
Received in revised form 17 May 2016
Available online 22 June 2016
Keywords:
Community detection
Weighted modularity
Cosine similarity
Potentially attractive clusters
Crisply fuzzy partition
a b s t r a c t
Community detection is a classic and very difficult task in the field of complex network
analysis, principally for its applications in domains such as social or biological networks
analysis. One of the most widely used technologies for community detection in networks
is the maximization of the quality function known as modularity. However, existing
work has proved that modularity maximization algorithms for community detection may
fail to resolve communities in small size. Here we present a new community detection
method, which is able to find crisp and fuzzy communities in undirected and unweighted
networks by maximizing weighted modularity. The algorithm derives new edge weights
using the cosine similarity in order to go around the resolution limit problem. Then a
new local moving heuristic based on weighted modularity optimization is proposed to
cluster the updated network. Finally, the set of potentially attractive clusters for each
node is computed, to further uncover the crisply fuzzy partition of the network. We give
demonstrative applications of the algorithm to a set of synthetic benchmark networks and
six real-world networks and find that it outperforms the current state of the art proposals
(even those aimed at finding overlapping communities) in terms of quality and scalability.
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
Recent years have witnessed an increasing interest in detecting communities in large-scale networks of various kinds.
Communities are clusters of closely connected nodes within a network. Since massive online social networks have been
deeply integrated into our daily life, detecting meaningful communities from them has become a critical task for research and
applications of various purposes. Early work in graph partitions [1–3] could be adopted for community detection. However,
these methods usually require the knowledge on the number of communities as part of the input, which is unrealistic in
community detection problems. Modularity [4], proposed by Newman et al., is the first successful attempt to resolve the
drawback specified above. Modularity scores high those partitions containing communities with an internal edge density
larger than that expected in a given graph model. Several strategies have been proposed for its optimization, including greedy
algorithms [5,6], simulated annealing [7], extremal optimization [8], spectral clustering [9], and genetic algorithms [10]. To
the best of our knowledge, the most competitive approach of this kind is Louvain [11], which can scale to graphs with
hundreds of millions of objects.
Recently, overlapping community detection has also been investigated [12,13], most of which typically employ a bottom-
up strategy to find joint communities. They often start by defining the properties of a node, a pair of nodes or a group
∗
Corresponding author.
E-mail address: buzhan@nuaa.edu.cn (Z. Bu).
http://dx.doi.org/10.1016/j.physa.2016.06.113
0378-4371/© 2016 Elsevier B.V. All rights reserved.