Computers, Materials & Continua · 2026 · Open access
Find overlapping communities by searching the nodes that carry structure.
SFCMOEA combines structural feature node extraction, maximal-clique representation, and hybrid MOEA/D to reduce the search space while preserving overlapping community structure.
DOI: 10.32604/cmc.2025.073572
A three-stage route from graph structure to overlapping partitions
The method focuses evolutionary search where the network carries the strongest structural signal.
-
01
Extract structural feature nodes
Map nodes into a structural embedding space, measure similarity with Minkowski distance, and retain the nodes most useful for partition search.
-
02
Represent overlap with maximal cliques
Search maximal cliques in the selected vertex set and use link strength to merge or preserve clique structure.
-
03
Optimize with hybrid MOEA/D
Evolve clique-based individuals while assigning unexplored nodes separately, balancing intracommunity density and intercommunity separation.
Where the paper is useful
Cite this work when you need a baseline or related method for overlapping community detection, evolutionary community search, clique-based representation, or structural-node selection.
- Overlapping community detection in complex networks
- MOEA/D and multi-objective community optimization
- Maximal-clique encoding and link-strength allocation
- Graph embedding for search-space reduction
- Community-count-aware evaluation with fNMI
Reported comparison
Broad gains in partition quality, with an explicit runtime trade-off.
SFCMOEA reported the best maximum and average extended modularity on nine of twelve real-world networks. It was faster than the compared clique-based and similarity-based MOEA methods, while traditional and shallow embedding baselines remained faster.
Inspect the full tables and settings →Use the canonical citation
Copy BibTeX or download a citation file for your reference manager.
@article{Ma2026SFCMOEA,
title = {A Hybrid Clique-Based Method with Structural Feature Node Extraction for Community Detection in Overlapping Networks},
author = {Ma, Sicheng and Zhang, Lixiang and Chen, Guocai and Dai, Zeyu and Zhu, Junru and Fang, Wei},
journal = {Computers, Materials \& Continua},
volume = {87},
number = {1},
year = {2026},
doi = {10.32604/cmc.2025.073572}
}
Follow-on application
The paper’s clique and structural-feature perspective also motivates later work on mesoscopic community features under distribution shift in Android malware graphs.