Multi-Diffusion Information Centrality for the Identification of Influential Spreaders in Temporal Social Networks

Identifying influential spreaders in a temporal social network, which has potential applications including network immunization, epidemic control, and viral marketing, is a fundamental class of problems. In this context, various centrality algorithms have been introduced to quantify influential spreaders, focusing on three categories: topology-based methods, dynamics-based methods, and machine learning-based methods. However, topology-based methods tend to consider single temporal features, while the consideration of multi-temporal features is subject to the same challenges of high temporal complexity as dynamics-based methods, and machine learning-based methods face challenges related to dependency on the training dataset. In this paper, we propose a novel centrality algorithm based on multiple diffusion information (RPT: Multi-diffusion information centrality based on R-path trees) to identify the influential node in a temporal social network. This algorithm considers three different temporal features and has lower temporal complexity using a newly proposed representation structure known as an R-path tree (a distinctive inverted tree that encompasses the earliest arrival paths from other nodes to the root node). Through experiments carried out on 12 empirical social networks, the results show that the effectiveness of RPT in identifying influential spreaders generally exceeds that of other baseline measures.