Online community detection for large complex networks.

Area of Science:

• Network Science
• Data Mining
• Computational Social Science

Background:

• Complex networks are fundamental to understanding systems in nature and society.
• Identifying community structure is key to analyzing these networks.
• Existing methods can be computationally intensive for large-scale networks.

Purpose of the Study:

• To develop an online community detection algorithm for large complex networks.
• To achieve linear time complexity for efficient real-time processing.
• To optimize expected modularity for improved performance.

Main Methods:

• An edge-by-edge processing approach that updates community structure incrementally.
• Constant time updates for new edges, avoiding full network re-computation.
• Optimization of expected modularity at each step, rather than absolute modularity.

Main Results:

• The algorithm demonstrates linear time complexity, enabling real-time analysis.
• Experimental results on 11 public datasets show competitive performance.
• The algorithm achieves faster running times compared to the Louvain algorithm.

Conclusions:

• The proposed online algorithm offers an efficient solution for community detection in large complex networks.
• It provides a scalable and real-time approach to network analysis.
• The method achieves a favorable balance between computational speed and detection accuracy.