A neurodynamic framework for local community extraction in networks

TL;DR

This paper introduces a neurodynamic framework that models local community detection in networks as stable states of a neuro-system, enabling multi-resolution analysis and application to real-world networks.

Contribution

It presents a novel neurodynamic approach for local community extraction that generalizes objective functions and addresses resolution limits in network analysis.

Findings

Framework successfully detects communities in model networks

Effective in analyzing real-world social and biological networks

Addresses multi-resolution community detection challenges

Abstract

To understand the structure and organization of a large-scale social, biological or technological network, it can be helpful to describe and extract local communities or modules of the network. In this article, we develop a neurodynamic framework to describe the local communities which correspond to the stable states of a neuro-system built based on the network. The quantitative criteria to describe the neurodynamic system can cover a large range of objective functions. The resolution limit of these functions enable us to propose a generic criterion to explore multi-resolution local communities. We explain the advantages of this framework and illustrate them by testing on a number of model and real-world networks.