Addressing overlapping communities in multiple-source detection: An edge clustering approach for complex networks

2025-07-10

AI Summary

Detecting multiple diffusion sources in complex networks—such as disease outbreaks or misinformation cascades—is challenging, especially when nodes belong to multiple overlapping communities. To address this, we propose an edge-clustering-based multi-source localization method. Our approach innovatively integrates a latent-space edge clustering model into a community-aware label propagation framework, resolving ambiguities arising from node multiplicity by performing clustering at the edge level rather than the node level. This enables precise identification of multiple diffusion sources embedded within overlapping community structures. Experiments on the ADD HEALTH social network dataset demonstrate that our method achieves significantly higher F1 scores than state-of-the-art baselines, particularly excelling in detecting sources located within overlapping community regions. The results confirm superior accuracy and robustness under realistic overlapping community topologies.

Technology Category

• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Machine Learning: Graph-based Machine Learning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Social Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

The source detection problem in network analysis involves identifying the origins of diffusion processes, such as disease outbreaks or misinformation propagation. Traditional methods often focus on single sources, whereas real-world scenarios frequently involve multiple sources, complicating detection efforts. This study addresses the multiple-source detection (MSD) problem by integrating edge clustering algorithms into the community-based label propagation framework, effectively handling mixed-membership issues where nodes belong to multiple communities. The proposed approach applies the automated latent space edge clustering model to a network, partitioning infected networks into edge-based clusters to identify multiple sources. Simulation studies on ADD HEALTH social network datasets demonstrate that this method achieves superior accuracy, as measured by the F1-Measure, compared to state-of-the-art clustering algorithms. The results highlight the robustness of edge clustering in accurately detecting sources, particularly in networks with complex and overlapping source regions. This work advances the applicability of clustering-based methods to MSD problems, offering improved accuracy and adaptability for real-world network analyses.

Problem

• Detect multiple diffusion sources in complex networks
• Handle overlapping community memberships in source detection
• Improve accuracy using edge clustering for network partitioning

Innovation

• Edge clustering for multiple-source detection
• Automated latent space edge clustering model
• Handles mixed-membership in complex networks