Identifying Vital Nodes in Hypernetworks Based on Improved PageRank Algorithm and Information Entropy

Hypernetworks can reflect the multiple connections of entities in the real-world from multiple dimensions. Identifying the vital nodes in hypernetworks is beneficial to analyze the topology and network functions of the hypernetwork. Traditional methods of vital node identification within a hypernetwork struggles to reach an optimum between time complexity and the precision of identification. As a result, this paper proposes a global and local centrality based on the improved PageRank algorithm and information entropy, which merges the local properties of nodes with their global properties to balance accuracy and time complexity. It is then compared with other methods on real hypernetworks in four different domains by means of monotonicity as well as SI propagation model evaluation criteria. To compare the ranking differentiation and accuracy with other centrality algorithms, we perform numerical simulation experiments on four real networks using the SI model. The experimental findings demonstrate that the proposed method not only yields more precise ranking outcomes but also significantly diminishes the occurrence of identical ranking nodes.