Lower Ricci Curvature for Efficient Community Detection

Area of Science:

• Network Science
• Graph Theory
• Data Mining

Background:

• Existing curvature-based methods for network analysis face computational challenges.
• There is a need for scalable and efficient discrete curvature measures for large networks.

Purpose of the Study:

• Introduce Lower Ricci Curvature (LRC) as a novel, scalable, and scale-free discrete curvature.
• Develop an LRC-based preprocessing method to improve community detection algorithms.

Main Methods:

• Introduced Lower Ricci Curvature (LRC), a discrete curvature measure.
• Developed a preprocessing technique utilizing LRC.
• Applied the method to diverse real-world networks (e.g., NCAA, DBLP, Amazon, YouTube).

Main Results:

• LRC offers linear computational complexity, suitable for large-scale network analysis.
• The LRC-based preprocessing method significantly enhances the performance of various community detection algorithms.
• Demonstrated efficacy across multiple real-world datasets.

Conclusions:

• Lower Ricci Curvature (LRC) is an effective and efficient tool for network analysis.
• LRC-based preprocessing improves community detection in complex networks.
• The method shows broad applicability in analyzing large-scale network structures.