Identifying vital nodes in complex networks by adjacency information entropy

Area of Science:

• Network Science
• Graph Theory
• Information Theory

Background:

• Identifying vital nodes is crucial for understanding network structure and function.
• Existing centrality indices (e.g., Betweenness Centrality, Degree Centrality, PageRank) have limitations in applicability across different network types (directed vs. undirected).

Purpose of the Study:

• To develop novel, broadly applicable centrality measures for identifying vital network nodes.
• To propose two new algorithms based on node adjacency information entropy.

Main Methods:

• Development of two vital node identification algorithms utilizing node adjacency information entropy.
• Conducting contrast experiments comparing the proposed algorithms with established centrality indices (BC, EC, CC, SH, DC, PR, VC).
• Validation across diverse network structures.

Main Results:

• The proposed information entropy-based index shows a high correlation with Degree Centrality (DC).
• A notable correlation was observed with PageRank (PR) and Eigenvector Centrality (VC) in directed networks.
• Experimental results confirm the effectiveness of the proposed algorithms in identifying vital nodes across different network types.

Conclusions:

• The novel algorithms offer a more universally applicable method for vital node identification compared to traditional indices.
• Information entropy provides a robust basis for developing effective centrality measures in network analysis.