Investigating Centrality Measures in Social Networks with Community Structure

TL;DR

This study compares classical and community-aware centrality measures in social networks, revealing two distinct groups of community-aware measures with varying degrees of similarity to classical measures across real-world networks.

Contribution

It provides a comprehensive analysis of the interactions between classical and community-aware centrality measures using real-world social network data.

Findings

Community-aware measures split into two groups based on their similarity to classical measures.

Some community-aware measures offer unique insights not captured by classical measures.

Results are consistent across multiple real-world online social networks.

Abstract

Centrality measures are crucial in quantifying the influence of the members of a social network. Although there has been a great deal of work dealing with this issue, the vast majority of classical centrality measures are agnostic of the community structure characterizing many social networks. Recent works have developed community-aware centrality measures that exploit features of the community structure information encountered in most real-world complex networks. In this paper, we investigate the interactions between 5 popular classical centrality measures and 5 community-aware centrality measures using 8 real-world online networks. Correlation as well as similarity measures between both type of centrality measures are computed. Results show that community-aware centrality measures can be divided into two groups. The first group, which includes Bridging centrality, Community Hub-Bridge and Participation Coefficient, provides distinctive node information as compared to classical centrality. This behavior is consistent across the networks. The second group which includes Community-based Mediator and Number of Neighboring Communities is characterized by more mixed results that vary across networks.