Co-Association Matrix-Based Multi-Layer Fusion for Community Detection in Attributed Networks.

Area of Science:

• Graph theory
• Network science
• Data mining

Background:

• Community detection in attributed networks is challenging due to data inconsistency between network topology and node attributes.
• Existing methods often use simple linear fusion strategies, making community detection vulnerable to data variations.
• Effective fusion of multi-source heterogeneous data is crucial for robust community detection algorithms.

Purpose of the Study:

• To develop a novel two-layer representation for capturing latent knowledge from both topological structure and node attributes in attributed networks.
• To propose a weighted co-association matrix-based fusion algorithm (WCMFA) for robust community detection.
• To extend community detection from a single-view to a multi-view approach.

Main Methods:

• Developed a novel two-layer representation to integrate network topology and node attributes.
• Proposed a weighted co-association matrix-based fusion algorithm (WCMFA).
• Employed multi-layer fusion strategies for enhanced community detection.

Main Results:

• The proposed WCMFA method effectively captures latent knowledge from attributed networks.
• Multi-layer fusion strategies improve the robustness and accuracy of community detection.
• Experimental results demonstrate superior performance compared to state-of-the-art algorithms.

Conclusions:

• The novel two-layer representation and WCMFA provide a robust solution for community detection in attributed networks.
• The multi-view fusion approach aligns with human cognitive processes for better community identification.
• This method significantly advances the field of attributed network analysis.