Consistency and differences between centrality measures across distinct classes of networks

Area of Science:

• Network Science
• Graph Theory
• Computational Social Science

Background:

• Node roles in networks are often assessed using centrality analysis.
• The distinct information provided by various centrality measures and their combined utility remains unclear.
• Understanding how network topology influences centrality measure correlations is crucial for interpreting node importance.

Purpose of the Study:

• To investigate the correlations between 17 centrality measures across diverse real-world networks.
• To examine how network density and global topology affect these correlations.
• To determine if clustering nodes by centrality profiles can reveal distinct functional roles.

Main Methods:

• Calculated correlations between 17 centrality measures.
• Analyzed 212 diverse real-world networks.
• Employed data-driven clustering based on centrality profiles.

Main Results:

• Centrality measures generally show positive correlations, with variation across networks.
• Network modularity significantly influences the strength of these correlations.
• Clustering successfully identified distinct node roles, such as topological cores and peripheries.

Conclusions:

• Network topology shapes the correlations between centrality measures.
• A comparative approach using multiple centrality measures enhances the interpretation of nodal roles in complex systems.