Study of Hub Centrality in Complex Networks and Models of Clustered Networks

Abstract

In this thesis, the concept of hub centrality is introduced and its relationship with degree of nodes in networks is investigated. While a universal correlation between hub centrality and degree is found in networks generated by the growth method, this correlation is not present in real-world networks due to the rich-club phenomenon and local hubs. By exploring overload cascading failure, the importance of hub centrality as a parameter that provides insights beyond degree in real-world networks is demonstrated.

Moreover, this thesis proposes models for generating triangular and quadrilateral clustered networks and applies them to random regular and Barabasi-Albert networks. Through simulations of innovation diffusion within these networks, the study discovers that triangular clustered networks have the highest spreading efficiency. However, quadrilateral clustered networks still significantly contribute to enhancing spreading efficiency. Therefore, it is crucial to examine not only triangular clustering but also quadrilateral clustering to comprehensively understand the dynamics of innovation diffusion.