Generative models for local network community detection

TL;DR

This paper introduces a novel probabilistic generative model approach for local network community detection, enabling efficient identification of communities around a seed node with competitive accuracy.

Contribution

It presents the first local community detection methods based on probabilistic models, using approximations of stochastic block models for improved detection.

Findings

Comparable or improved detection results on real datasets

First probabilistic model-based local community detection methods

Approximation relates to conductance metric in the limit

Abstract

Local network community detection aims to find a single community in a large network, while inspecting only a small part of that network around a given seed node. This is much cheaper than finding all communities in a network. Most methods for local community detection are formulated as ad-hoc optimization problems. In this work, we instead start from a generative model for networks with community structure. By assuming that the network is uniform, we can approximate the structure of unobserved parts of the network to obtain a method for local community detection. We apply this local approximation technique to two variants of the stochastic block model. To our knowledge, this results in the first local community detection methods based on probabilistic models. Interestingly, in the limit, one of the proposed approximations corresponds to conductance, a popular metric in this field. Experiments on real and synthetic datasets show comparable or improved results compared to state-of-the-art local community detection algorithms.