Communities detection in complex network and multilayer network systems: A flow approach

TL;DR

This paper introduces a flow-based approach for detecting communities in complex and multilayer networks, utilizing flow influence, flow cores, and aggregate networks to identify communities where traditional methods fail.

Contribution

It presents novel flow-based methods for community detection in complex and multilayer networks, including new concepts like flow influence parameters and flow cores.

Findings

Methods effectively identify communities where existing techniques fail.

Two new approaches for multilayer network community detection are proposed.

Application examples demonstrate the methods' effectiveness.

Abstract

A flow approach to community detection in complex network and multilayer network systems is proposed. Two methods have been developed to search for communities in a network system (NS). The first of them is based on the calculation of flow influence parameters of NS's subsystems, selected according to the principle of nesting hierarchy. The second method uses the concept of flow core of network system. Two methods are also proposed for community detection in multilayer network system (MLNS). The first of them is based on the concept of MLNS aggregate-network and subsequent allocation of its flow core. The second method uses the concept of flow core of the process of intersystem interactions in general. All developed methods are based on the use of flow criterion that the selected group of nodes really forms a community. The results of application of developed approaches are illustrated by examples for which known methods are ineffective.