We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Physics > Physics and Society

Title: Automatic detection of multilevel communities: scalable and resolution-limit-free

Abstract: Community detection in complex networks has been hindered by two defects: (1) the resolution limit problem, which restrains simultaneous detection for communities of heterogeneous sizes, and (2) divergent outputs of the optimization algorithm, which set hurdles for the differentiation of more relevant and significant results. In this paper, we suggest a renewed method for community detection via a scalable community fitness function. Due to its scalability, this method is on the one hand free of the resolution limit problem, even in large heterogeneous networks, and on the other hand capable of detecting multiple levels of communities in deep hierarchical networks. Moreover, we propose a strict definition for the term "plateau," which has been always loosely used in previous literature, to help us remove random and irrelevant outputs automatically--without any artificial selection. As a result, our method has "neat" outputs that include only stable and informative plateaus. On synthetic networks with prearranged community structures, our method outperforms most previous methods by recreating the prearranged communities accurately, while on real-world networks, it discovers reasonable community structures that fit the ground truth we already know about the network.
Comments: 22 pages, 7 figures, 1 table, 1 supplementary text, 4 supplementary figures
Subjects: Physics and Society (physics.soc-ph); Social and Information Networks (cs.SI)
Cite as: arXiv:2201.09544 [physics.soc-ph]
  (or arXiv:2201.09544v2 [physics.soc-ph] for this version)

Submission history

From: Kun Gao Ph.D [view email]
[v1] Mon, 24 Jan 2022 09:32:39 GMT (2629kb)
[v2] Tue, 10 May 2022 16:33:24 GMT (2596kb)

Link back to: arXiv, form interface, contact.