Detecting community structure in signed and unsigned social networks by using weighted label propagation

Area of Science:

• Complex Networks Analysis
• Social Network Analysis
• Community Detection Algorithms

Background:

• Community structure detection is crucial for understanding complex networks (technological, informational, biological, social).
• Existing methods often overlook communication intensity, a key property of social networks.
• Accurate community detection requires incorporating nuanced interaction data beyond simple connectivity.

Purpose of the Study:

• To propose a novel algorithm, the Weighted Label Propagation Algorithm (WLPA), for community detection in social networks.
• To integrate communication intensity into community detection, addressing limitations of prior methods.
• To evaluate WLPA's effectiveness on both signed and unsigned social networks.

Main Methods:

• Utilized MinHash to estimate node similarity, quantifying communication intensity between users.
• Assigned edge weights based on estimated node similarity, reflecting communication strength.
• Employed weighted label propagation for determining the final community structure.

Main Results:

• WLPA effectively incorporates communication intensity into the community detection process.
• Experimental results on benchmark networks demonstrate WLPA's efficiency and effectiveness.
• The algorithm performs well on both signed and unsigned social network structures.

Conclusions:

• The Weighted Label Propagation Algorithm (WLPA) offers an improved approach to community detection in social networks.
• Incorporating communication intensity significantly enhances the accuracy and robustness of community structure identification.
• WLPA provides a valuable tool for analyzing the organization and operation of complex social systems.