“…The LFR networks have heterogeneous degree distributions and community sizes, and allow communities to overlap, which are in accordance with the features of real networks, and have been widely used for testing the community detection methods in recent years. [5,[23][24][25][26] The adjustable topology parameters for the LFR benchmark networks are network size N, average degrees k , maximum degrees k max , minimum and maximum community size c min and c max , the degree distributions γ, the community size distributions β , the mixing parameter µ, the number of communities each overlapping node belongs to o m , and the number of overlapping nodes o n . [22] In order to detect the overlapping communities and nodes and make comparisons with other wellknown fuzzy overlapping community detection methods, such as Fuzzyclust [27] and NMF, [17] we first investigate our method on the LFR networks without overlapping nodes to compare with the existing local methods in the detection of disjoint communities.…”