An Overlapping Community Discovery Algorithm Based on Label Propagation

Abstract

Aiming at the defects of random update and poor stability of COPRA algorithm, this paper proposes an overlapping community discovery algorithm based on label propagation (COLBN). The algorithm calculates the value of the node by traversing each node, and finds the reference node to determine the number of divided communities. Then, the reference node is used to update the label for the adjacent node to identify the overlapping node, and the overlapping community can be divided when all overlapping nodes are found. This paper makes use of the network dataset of the bottlenose dolphins in the Magic Bay and the book dataset on American politics, and the comparison test of COPRA algorithm, which proves that the COLBN algorithm significantly reduces the update randomness of the label propagation process and improves the stability of the generated network communities.