# HA: An Influential Node Identification Algorithm Based on Hub-Triggered Neighborhood Decomposition and Asymmetric Order-by-Order Recurrence Model

## TL;DR

This paper introduces a new algorithm to identify influential nodes in power networks, which helps improve network security against attacks.

## Contribution

The novel algorithm combines hub-triggered decomposition and an asymmetric recurrence model to better evaluate node influence.

## Key findings

- The proposed algorithm outperforms existing methods in identifying influential nodes in power networks.
- The algorithm achieves better performance in SIR correlation coefficients and algorithmic resolution.
- The method effectively integrates infected and infecting potentials of multiple-order neighbors.

## Abstract

In recent years, the rise of power network security incidents caused by malicious attacks has drawn considerable attention to identifying influential nodes in power networks. Power networks are a special class of complex networks characterized by a high relative clustering coefficient, which reflects a more intricate connection between nodes. This paper proposes a novel node influence evaluation algorithm based on hub-triggered neighborhood decomposition and asymmetric order-by-order recurrence model. First, the concepts of network directionalization strategy and hub-triggered neighborhood decomposition are introduced to distinguish the functional differences among nodes in the virus-spreading process. Second, this paper proposes the concepts of infected and infecting potential, then constructs a calculation model with asymmetric characteristics based on the order-by-order recurrence method to fully use the information in the connection structure of the adjacent neighborhood. Finally, the influence of the hub node is evaluated by integrating the infected potential and infecting potential of neighbors of multiple orders. We compare our method with the traditional and state-of-the-art algorithms on six power networks regarding Susceptible–Infected–Recovered (SIR) correlation coefficients, imprecision functions, and algorithmic resolution. The experimental results show that the algorithm proposed in this paper is superior in the above aspects.