Overlapping Community Discovery via Weighted Line Graphs of Networks

Abstract

We propose an approach for overlapping community discovery via weighted line graphs of networks. For undirected connected networks without self-loops, we generalize previous weighted line graphs by: 1) defining weights of a line graph based on the weights in the original network, and 2) removing self-loops in weighted line graphs, while sustaining their properties. By applying some off-the-shelf node partitioning method to the weighted line graph, a node in the original network can be assigned to more than one community based on the community labels of its adjacent links. Various properties of the proposed weighted line graphs are clarified. Furthermore, we propose a generalized quality measure for soft assignment of nodes in overlapping communities.