Discovering Dense Correlated Subgraphs in Dynamic Networks

TL;DR

This paper introduces a scalable approximate algorithm to identify dense, correlated subgraphs in dynamic networks by analyzing temporal edge activity patterns, with promising results on real and synthetic data.

Contribution

It presents a novel framework for enumerating dense, correlated subgraphs in dynamic networks, including an efficient approximation algorithm.

Findings

High accuracy of the approximation demonstrated on synthetic data.

Framework scales well with network size.

Effective in identifying correlated dense subgraphs in real datasets.

Abstract

Given a dynamic network, where edges appear and disappear over time, we are interested in finding sets of edges that have similar temporal behavior and form a dense subgraph. Formally, we define the problem as the enumeration of the maximal subgraphs that satisfy specific density and similarity thresholds. To measure the similarity of the temporal behavior, we use the correlation between the binary time series that represent the activity of the edges. For the density, we study two variants based on the average degree. For these problem variants we enumerate the maximal subgraphs and compute a compact subset of subgraphs that have limited overlap. We propose an approximate algorithm that scales well with the size of the network, while achieving a high accuracy. We evaluate our framework on both real and synthetic datasets. The results of the synthetic data demonstrate the high accuracy of the approximation and show the scalability of the framework.