Multilayer representation of collaboration networks with higher-order interactions

Area of Science:

• Bibliometrics
• Network Science
• Scientific Collaboration Analysis

Background:

• Traditional collaboration networks primarily model pairwise interactions, limiting their ability to represent complex, multi-person research teams.
• Understanding collaboration dynamics is crucial for scientific innovation and knowledge discovery in research groups of all sizes.

Purpose of the Study:

• To develop and analyze higher-order scientific collaboration networks that capture multi-person interactions.
• To gain microscopic insights into researcher representativeness and network evolution within scientific collaborations.
• To validate findings using both researcher and publication nodes across diverse scientific disciplines.

Main Methods:

• Construction of higher-order collaboration networks representing interactions among three or more individuals.
• Analysis of layered networks based on the number of collaborators.
• Sequential graph analysis tracking the maturation of network topological features.
• Node representation comparison using researchers and publications.

Main Results:

• Higher-order networks provide a more natural representation of team-based scientific collaborations.
• Analysis reveals novel insights into individual researcher roles and network structures.
• The maturation process of collaboration networks can be tracked by progressively merging smaller collaborations.
• Robustness of findings confirmed by using both researcher and publication nodes.

Conclusions:

• Higher-order collaboration networks offer a powerful framework for understanding complex scientific teamwork.
• This approach enhances the analysis of researcher representativeness and network dynamics.
• The findings are applicable across physics, mathematics, computer science, and broader research fields.