EntropyWalker, a Fast Algorithm for Small Community Detection in Large Graphs

TL;DR

EntropyWalker is a fast, entropy-based algorithm designed for overlapping community detection in large graphs, effectively handling size constraints through random walk simulations.

Contribution

It introduces a simple, entropy-measure approach for community detection that efficiently manages overlapping communities with size constraints.

Findings

Effective in large graphs

Handles overlapping communities

Fast due to simple entropy-based method

Abstract

This report presents a very simple algorithm for overlaping community-detection in large graphs under constraints such as the minimum and maximum number of members allowed. The algorithm is based on the simulation of random walks and measures the entropy of each random walk to detect the discovery of a community.