How Correlated are Community-aware and Classical Centrality Measures in Complex Networks?

TL;DR

This study compares classical and community-aware centrality measures in complex networks, revealing generally low to medium correlations influenced by specific macroscopic and mesoscopic network features.

Contribution

It provides a comprehensive analysis of the relationship between different centrality measures across diverse real-world networks, highlighting key influencing network properties.

Findings

Classical and community-aware centrality measures have low to medium correlation.

Transitivity and efficiency significantly influence correlation variation.

Modularity and Max-ODF strongly affect the relationship between measures.

Abstract

Unlike classical centrality measures, recently developed community-aware centrality measures use a network's community structure to identify influential nodes in complex networks. This paper investigates their relationship on a set of fifty real-world networks originating from various domains. Results show that classical and community-aware centrality measures generally exhibit low to medium correlation values. These results are consistent across networks. Transitivity and efficiency are the most influential macroscopic network features driving the correlation variation between classical and community-aware centrality measures. Additionally, the mixing parameter, the modularity, and the Max-ODF are the main mesoscopic topological properties exerting the most substantial effect.