Efficient hybrid PageRank centrality computation for multilayer networks

Quantifying node centrality in multilayer networks is crucial for identifying influential nodes across various applications. Building on the PageRank model for single-layer networks, Lv et al. recently introduced a promising multilayer PageRank model for assessing node and layer centrality. In this paper, we reformulate this model within a discrete Markov chain framework. Our approach incorporates link diversity to enhance centrality measurement and ensures irreducibility within the internal Markov chains. This refinement enables an efficient computational strategy leveraging numerical algebra techniques. Experiments across diverse multilayer networks demonstrate the model’s effectiveness and computational efficiency, particularly for large-scale networks.