A signal processing perspective to community detection in dynamic networks.

Many real world systems ranging from biological to social systems can be modeled as networks. As network data becomes more ubiquitous, it is important to reduce the large-scale networks into smaller modules or communities. Most of the original research in community detection has focused on static networks while many real world networks such as social and biological networks exhibit varying topology over time. For this reason, over the past twenty years various community detection methods for dynamic networks have been developed in physics, statistics, computer science and signal processing. In this paper, we present an overview of the current state-of-the-art in community detection for dynamic networks and illustrate some of the shortcomings of existing methods. We also outline future challenges and opportunities in this research area. • An overview of existing approaches to community detection in dynamic networks is presented. • Community detection results for both synthetic and real dynamic networks are given. • The results illustrate the utility and shortcomings of current community detection methods. • Challenges and opportunities in community detection for dynamic networks are discussed. [ABSTRACT FROM AUTHOR]