In Search of Dense Subgraphs: How Good is Greedy Peeling?

TL;DR

This paper evaluates the effectiveness of greedy and new heuristic algorithms for finding dense subgraphs in large graphs, demonstrating their efficiency and near-optimal solutions in practical scenarios.

Contribution

It introduces a new heuristic algorithm built on top of existing methods and provides an extensive computational study comparing various approaches.

Findings

The new heuristic often finds optimal or near-optimal solutions.

Greedy heuristic is extremely fast and effective on large instances.

Proposed methods outperform some existing heuristics in accuracy and speed.

Abstract

The problem of finding the densest subgraph in a given graph has several applications in graph mining, particularly in areas like social network analysis, protein and gene analyses etc. Depending on the application, finding dense subgraphs can be used to determine regions of high importance, similar characteristics or enhanced interaction. The densest subgraph extraction problem is a fundamentally a non-linear optimization problem. Nevertheless, it can be solved in polynomial time by an exact algorithm based on the iterative solution of a series of maximum flow sub-problems. Despite its polynomial time complexity, the computing time required by the exact algorithms on very large graphs could be prohibitive. Thus, to approach graphs with millions of vertices and edges, one has to resort to heuristic algorithms. We provide an efficient implementation of a greedy heuristic from the literature that is extremely fast and has some nice theoretical properties. We also introduce a new heurisitic algorithm that is built on top of the greedy and the exact methods. An extensive computational study is presented to evaluate the performance of various solution methods on a benchmark composed of 86 instances taken from the literature. This analysis shows that the proposed heuristic algorithm proved very effective on a large number of test instances, often providing either the optimal solution or near-optimal solution within short computing times.