A change of perspective in network centrality

Area of Science:

• Network Science
• Graph Theory
• Data Analysis

Background:

• Network centrality measures node importance across diverse fields.
• Existing metrics rely on ad hoc assumptions, lacking standardized comparison methods.
• Current metrics show limitations in accurately describing node centrality in complex networks.

Purpose of the Study:

• To propose a novel theoretical perspective for defining network centrality based on the adjacency matrix.
• To evaluate and compare the effectiveness and reliability of various centrality metrics.
• To introduce improved multi-component centrality metrics for complex networks.

Main Methods:

• Developed a theoretical framework where centrality arises from the network's adjacency matrix.
• Included standard metrics like degree, eigenvector, and hub-authority centrality within this framework.
• Evaluated metric performance on a large dataset of diverse networks.

Main Results:

• Demonstrated that standard centrality metrics exhibit unsatisfactory performance and intrinsic limitations.
• Identified the need for more informative centrality measures.
• Proposed multi-component centrality metrics as a natural extension of standard approaches.

Conclusions:

• The proposed adjacency matrix-based framework provides a unified perspective on centrality.
• Standard centrality metrics are insufficient for complex network analysis.
• Multi-component centrality metrics offer a more robust and informative approach to understanding node importance.