Discovering Network Structure Beyond Communities

TL;DR

This paper introduces a novel exploratory method that combines human visual pattern recognition with computational processing to discover diverse network groups beyond traditional communities, revealing hidden structures in complex networks.

Contribution

It presents a new approach for identifying various types of node groups in networks without prior assumptions, including their number, membership, and defining properties.

Findings

Successfully applied to real networks, uncovering previously hidden group structures.

Demonstrates the method's ability to identify diverse network groupings beyond communities.

Suggests many network group structures remain undiscovered in various scientific fields.

Abstract

To understand the formation, evolution, and function of complex systems, it is crucial to understand the internal organization of their interaction networks. Partly due to the impossibility of visualizing large complex networks, resolving network structure remains a challenging problem. Here we overcome this difficulty by combining the visual pattern recognition ability of humans with the high processing speed of computers to develop an exploratory method for discovering groups of nodes characterized by common network properties, including but not limited to communities of densely connected nodes. Without any prior information about the nature of the groups, the method simultaneously identifies the number of groups, the group assignment, and the properties that define these groups. The results of applying our method to real networks suggest the possibility that most group structures lurk undiscovered in the fast-growing inventory of social, biological, and technological networks of scientific interest.