Title: Automatic detection of multilevel communities in complex networks with a scalable community fitness function

Abstract: Community detection in complex networks has been held back by two obstacles: the resolution limit problem, which restrains simultaneous detection for communities of heterogeneous sizes, and the divergent outputs of heuristic algorithm, which are unfavorable for differentiating more relevant and significant results. In this paper, we propose a renewed method for community detection with a multiresolution and rescalable community fitness function. The scalability of the community fitness function on the one hand exempts our method from the resolution limit problem in heterogeneous networks, and on the other hand enables our method to detect multilevel community structures in deep hierarchical networks. Furthermore, we suggest a strict definition of "plateau," with which we evaluate the stability of the outputs, and remove unstable and irrelevant ones automatically, without any artificial or arbitrary selection. As a result, our method outperforms most previous methods; it reproduces the expected community structures accurately for various classes of synthetic networks, as well as the ground truths for real-world networks.