Community detection in networks: A user guide

By Santo Fortunato, Darko Hric. 2016

2,066 citations recorded by the source corpus.

Abstract Community detection in networks is one of the most popular topics of modern network science. Communities, or clusters, are usually groups of vertices having higher probability of being connected to each other than to members of other groups, though other patterns are possible. Identifying communities is an ill-defined problem. There are no universal protocols on the fundamental ingredients, like the definition of community itself, nor on other crucial issues, like the validation of algorithms and the comparison of their performances. This has generated a number of confusions and misconceptions, which undermine the progress in the field. We offer a guided tour through the main aspects of the problem. We also point out strengths and weaknesses of popular methods, and give directions to their use.

Subfields: Computer Networks and Communications, Statistical and Nonlinear Physics

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Networks and Communications

1. Richard F. Betzel
2. M. A. Porter
3. D. Bassett
4. Naoki Masuda
5. Neil Shah
6. Xiao Wang
7. Xiaorui Liu
8. Peng Cui

In Statistical and Nonlinear Physics

1. Xiao Wang
2. Peng Cui
3. Richard F. Betzel
4. Wenwu Zhu
5. D. Bassett
6. Jian Pei
7. Philip S. Yu
8. M. A. Porter

Citations in the corpus

• Debiasing Community Detection: The Importance of Lowly Connected Nodes (2019, neutral): “More information can be found in [6] which is a user guide for community detection and its methods.”
• Open Community Challenge Reveals Molecular Network Modules with Key Roles in Diseases (2018, methodology): “Module identification, also called community detection or graph clustering, is a key problem in network science for which a wide range of methods have been proposed (Fortunato and Hric, 2016).”
• A comprehensive review of Rasch measurement in language assessment: Recommendations and guidelines for research (2020, background): “The fundamental assumption is that having more connections between papers (nodes) makes them more relevant (connected) to each other, which is the basis of all community detection algorithms (Fortunato & Hric, 2016).”
• A connectome and analysis of the adult Drosophila central brain (2020, background): “Partitioning a network into clusters of nodes that exhibit similar connectivity is known as 895 community detection or graph clustering(Fortunato and Hric, 2016).”
• Community Preserving Network Embedding (2017, background): “To cover all community detection meth-ods is beyond the scope of this paper, and an elaborate review can be found in (Xie, Kelley, and Szymanski 2013; Fortunato and Hric 2016).”
• Community Discovery in Dynamic Networks (2017, neutral): “…they do not allow to make direct use of temporal information, thus reducing dynamic problems to static ones: for an in-depth overview of community discovery approaches defined for such scenarios refer to (Coscia et al. , 2011; Fortunato, 2010; Fortunato and Hric, 2016; Xie et al. , 2013b).”
• What are higher-order networks? (2021, background): “determine important nodes [112,123,224], reveal modular structure of a system [108, 249], or—if each node is a dynamical unit—elucidate collective network dynamics such as synchronization [269].”
• Is Homophily a Necessity for Graph Neural Networks? (2021, methodology): “2 Analysis based on CSBM model with two classes To have a clear control on the assumptions, we study the contextual stochastic block model(CSBM), which is a generative model for random graphs; such models have been previously adopted for benchmarking graph clustering [Fortunato and Hric, 2016] and G”
• Multi-scale brain networks (2016, background): “Whether the plurality of methods is viewed as a shortcoming or an advantage, the enterprise of community detection is one of the better-developed and continually growing sub-fields of network analysis (Fortunato, 2010; Fortunato and Hric, 2016).”
• Nine quick tips for analyzing network data (2019, methodology): “Community detection methods [23] implement this approach, which implicitly assumes the existence of modules in the network.”
• Random walks and diffusion on networks (2016, methodology): “It is common to use the “modularity” objective function Q to quantify the quality of a partition of a network into nonoverlapping communities, and many community-detection methods are based on maximizing Q [65].”