A novel metric for community detection

¹ Computational Communication Collaboratory, Nanjing University - Nanjing, 210093, PRC
² Complex Systems Group, Department of Mathematics and Statistics, The University of Western Australia Crawley, Western Australia 6009, Australia
³ Mineral Resources, CSIRO - Kensington, WA, 6151, Australia

^(a) keke.shang.1989@gmail.com, kekeshang@nju.edu.cn

Received: 17 February 2020
Accepted: 2 April 2020

Abstract

Detection of dense communities has recently attracted increasing attention within network science and various metrics for detection of such communities have been proposed. The most popular metric —modularity— is based on the rule that the links within communities are denser than external links among communities. However, the principle of this metric suffers from ambiguity, and is based on a narrow intuition of what it means to form a “community”. Instead we propose that the defining characteristic of a community is that links are more predictable within a community rather than between communities. In this letter, based on the effect of communities on link prediction, we propose a novel metric for community detection based directly on this property. We find that our metric is more robust than traditional modularity measures for each specific algorithm. Finally, we provide a measure of the improvement offered by our metric.