Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Comprehensive Review of Community Detection in Graphs

Jiakang Li, Songning Lai, Zhihao Shuai, Yuan Tan, Yifan Jia, Mianyang Yu, Zichen Song, Xiaokang Peng, Ziyang Xu, Yongxin Ni, Haifeng Qiu, Jiayu Yang, Yutong Liu, Yonggang Lu

TL;DR

This paper surveys four major families of graph-based community detection methods—modularity-based, spectral, probabilistic, and deep learning—and introduces a novel RMS method that combines Medoid-Shift with KNN to detect communities. It provides a structured taxonomy, analyzes representative algorithms (e.g., Louvain, Infomap, SBM, GCN, VGAE), and evaluates performance on datasets with and without ground truth using $Q$ and $NMI$ metrics. The RMS method demonstrates competitive modularity on weighted graphs and strong alignment with ground-truth partitions on several datasets, underscoring the potential of distance- or similarity-matrix–driven clustering in non-Euclidean spaces. The work highlights future directions in integrating multiple perspectives, improving scalability, and applying these methods to real-world social, biological, and information networks.

Abstract

The study of complex networks has significantly advanced our understanding of community structures which serves as a crucial feature of real-world graphs. Detecting communities in graphs is a challenging problem with applications in sociology, biology, and computer science. Despite the efforts of an interdisciplinary community of scientists, a satisfactory solution to this problem has not yet been achieved. This review article delves into the topic of community detection in graphs, which serves as a thorough exposition of various community detection methods from perspectives of modularity-based method, spectral clustering, probabilistic modelling, and deep learning. Along with the methods, a new community detection method designed by us is also presented. Additionally, the performance of these methods on the datasets with and without ground truth is compared. In conclusion, this comprehensive review provides a deep understanding of community detection in graphs.

A Comprehensive Review of Community Detection in Graphs

TL;DR

This paper surveys four major families of graph-based community detection methods—modularity-based, spectral, probabilistic, and deep learning—and introduces a novel RMS method that combines Medoid-Shift with KNN to detect communities. It provides a structured taxonomy, analyzes representative algorithms (e.g., Louvain, Infomap, SBM, GCN, VGAE), and evaluates performance on datasets with and without ground truth using and metrics. The RMS method demonstrates competitive modularity on weighted graphs and strong alignment with ground-truth partitions on several datasets, underscoring the potential of distance- or similarity-matrix–driven clustering in non-Euclidean spaces. The work highlights future directions in integrating multiple perspectives, improving scalability, and applying these methods to real-world social, biological, and information networks.

Abstract

The study of complex networks has significantly advanced our understanding of community structures which serves as a crucial feature of real-world graphs. Detecting communities in graphs is a challenging problem with applications in sociology, biology, and computer science. Despite the efforts of an interdisciplinary community of scientists, a satisfactory solution to this problem has not yet been achieved. This review article delves into the topic of community detection in graphs, which serves as a thorough exposition of various community detection methods from perspectives of modularity-based method, spectral clustering, probabilistic modelling, and deep learning. Along with the methods, a new community detection method designed by us is also presented. Additionally, the performance of these methods on the datasets with and without ground truth is compared. In conclusion, this comprehensive review provides a deep understanding of community detection in graphs.
Paper Structure (42 sections, 3 equations, 2 figures, 2 tables, 2 algorithms)

This paper contains 42 sections, 3 equations, 2 figures, 2 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Motivation
  3. Contribution
  4. Road Map
  5. Research Background of Community Detection
  6. Overview of The Community Detection Methodology and Our Methodology
  7. Modularity-based Methods
  8. Louvain method
  9. Infomap method
  10. Kernighan-Lin method
  11. EdMot
  12. LPA
  13. Spectral Clustering Methods
  14. Normalized Cut algorithm
  15. RatioCut algorithm
  16. ...and 27 more sections

Figures (2)

  • Figure 1: Overview of the community detection methodology
  • Figure 2: For the current point in the shifting, Medoid-Shift chooses the data point to shift, while Mean-Shift chooses the location to shift