Detecting and analyzing research communities in longitudinal scientific networks

Area of Science:

• Bibliometrics
• Network Science
• Sociology of Science

Background:

• Collaborative teams and communities are increasingly recognized for producing high-impact scientific work.
• Understanding the dynamics of scientific collaboration is crucial for fostering innovation and research productivity.

Purpose of the Study:

• To propose and validate a novel method for identifying collaborative research communities within longitudinal scientific networks.
• To evaluate the influence of research institutes, services, and policies on interdisciplinary collaboration between these communities.

Main Methods:

• Community detection algorithms applied to longitudinal collaboration networks to identify stable research communities.
• Construction of cross-community interaction networks and estimation of Exponential Random Graph Models (ERGMs).
• Application of the method to publication and grant collaboration data from the University of Florida.

Main Results:

• Factors such as similar institutional affiliation, spatial proximity, and shared research services significantly predict interdisciplinary collaboration.
• The developed method successfully identifies research communities and analyzes their longitudinal network formation.
• The study quantifies the growth of interdisciplinary team science and its association with institutional factors.

Conclusions:

• The proposed methodology provides a robust framework for analyzing the structure and evolution of scientific collaboration.
• Understanding the drivers of interdisciplinary collaboration can inform the development of effective research policies and support services.
• This approach offers valuable insights into measuring and fostering team science within research institutions.