Finding core–periphery structures in large networks

Finding core–periphery structures in networks is very useful in many disciplines such as biology and sociology. However, most of the previous works focus on the single core–periphery structure in the network. A few recent algorithms considering multiple core–periphery are usually not suitable for large networks. Inspired by the modularity maximization method for community detection, we propose a simple but effective approach to detect core–periphery structures in this work. Moreover, we propose a metric called core–periphery score to evaluate the performance of core–periphery structure detection algorithms. In the experiment, we find that the score is consistent with the normalized mutual information when ground-truth structures are given. Our approach also outperforms other core–periphery detection algorithms for randomly generated networks and real-world networks.