A Comprehensive Survey on Identifying Influential Nodes: From Structural Centrality-based to Learning-based Methods

Identifying influential nodes in complex networks is a fundamental challenge in network science. This problem involves measuring the influence of nodes and identifying those that exert the greatest impact on network dynamics and information dissemination. Existing approaches can be broadly divided into two categories: structural centrality-based methods and machine-learning-based methods. This article first provides background formulation on the influential node identification problem, followed by a comprehensive review of state-of-the-art methods and algorithms proposed to find the most influential nodes in complex networks. To trace the evolution of these approaches, the commonly used evaluation metrics are also presented. We then propose a framework that evaluates and compares the algorithms along key dimensions such as scalability, interpretability, computational cost, and appropriate application contexts. Finally, the article explores promising research directions, including hybrid methods, improved interpretability, more realistic performance evaluation in dynamic and large-scale networks, and the use of large language models. This survey provides a comprehensive reference for researchers and practitioners, offering clear insights into the past developments, current capabilities, and future opportunities in the field of influential node identification.