EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Description-Length Community Detection with Noise-Adaptive Graph Coarsening

Matteo Piras

International Journal of Information and Communication Technology Trends · 2026

DOI: 10.66525/ijictt130 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Information and Communication Technology Trends, 2026. Verfasst von Matteo Piras. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Community detection remains a foundational challenge in network science, with applications ranging from social network analysis to computational biology. Existing methods often struggle with large-scale graphs heavily corrupted by structural noise, leading to suboptimal clustering performance and prohibitive computational costs. This paper introduces a novel framework for description-length community detection integrated with a noise-adaptive graph coarsening mechanism. By formulating community detection through the minimum description length principle, the proposed methodology identifies the optimal network partition that most efficiently compresses the network topology. To address the dual challenges of scalability and structural noise, we develop a noise-adaptive coarsening algorithm that selectively contracts edges based on local topological similarity and estimated noise probabilities. This preprocessing step effectively filters out spurious connections while preserving the mesoscopic community structure, significantly accelerating the subsequent description-length optimization. Comprehensive theoretical analysis demonstrates that our coarsening approach strictly bounds the approximation error of the description length objective. Extensive evaluations on diverse synthetic and real-world network datasets indicate that the proposed framework achieves superior performance in clustering accuracy and computational efficiency compared to state-of-the-art baselines. The integration of noise robustness into information-theoretic community detection offers a scalable and theoretically grounded solution for mining complex, noisy networks.

Zitieren

APA

Piras, M. (2026). Description-Length Community Detection with Noise-Adaptive Graph Coarsening. International Journal of Information and Communication Technology Trends. https://doi.org/10.66525/ijictt130

Vancouver

Piras M. Description-Length Community Detection with Noise-Adaptive Graph Coarsening. International Journal of Information and Communication Technology Trends. 2026. doi:10.66525/ijictt130

BibTeX
@article{Piras2026,
  author = {Piras, Matteo},
  title = {Description-Length Community Detection with Noise-Adaptive Graph Coarsening},
  journal = {International Journal of Information and Communication Technology Trends},
  year = {2026},
  doi = {10.66525/ijictt130}
}
Kostenlos starten →