Doubly stochastic scaling unifies community detection

Graph partitioning, or community detection, has been widely investigated in network science. Yet, the correct structure on a given is essentially data-driven. Thus, instead of formal definition, diverse measures have conceived to capture intuitive desirable properties shared by most structures. In this work, we propose preprocessing based doubly stochastic scaling adjacency matrices, highlight these properties. By investigating range detection measures, and carefully generalising them graphs, show that such unifies whole category measures—namely, so-called linear criteria—onto two unique set up. Finally, help practitioners setting up provide an extensive numerical comparison capacity uncover structures within block models, using Louvain algorithm.