Finding multiple core-periphery pairs in networks.

Area of Science:

• Network science
• Graph theory
• Computational social science

Background:

• Networks often exhibit core-periphery structures, where central nodes (cores) are highly connected and peripheral nodes have fewer connections.
• Existing models typically identify a single core-periphery structure, which may not capture the complexity of all real-world networks.
• Many empirical networks display community structures, suggesting multiple, distinct organizational principles might coexist.

Purpose of the Study:

• To develop a scalable algorithm for detecting multiple, non-overlapping core-periphery structures within a single network.
• To provide a more nuanced understanding of network organization beyond single core-periphery models.
• To demonstrate the algorithm's applicability to diverse network types.

Main Methods:

• A novel scalable algorithm is proposed to identify multiple, distinct core-periphery groups.
• The algorithm is tested on both synthesized network models and real-world empirical datasets.
• Network analysis techniques are employed to validate the detected structures.

Main Results:

• The algorithm successfully identifies multiple, non-overlapping core-periphery structures in various networks.
• Analysis of a political blog network revealed distinct core-periphery pairs with differing political affiliations.
• Examination of a global airport network showed a clear separation between international and domestic subnetworks within specific countries.

Conclusions:

• Networks can possess multiple, independent core-periphery structures, challenging the traditional single-core model.
• The developed algorithm offers a powerful tool for uncovering complex organizational patterns in diverse network systems.
• Understanding these multiple structures provides insights into network function, such as political alignment or logistical segregation.