The structure of scientific collaboration networks

Area of Science:

• Network science
• Bibliometrics
• Sociology of science

Background:

• Understanding the structure of scientific collaboration is crucial for analyzing knowledge dissemination and scientific progress.
• Previous research has explored co-authorship networks, but comprehensive analysis across diverse scientific domains is needed.

Purpose of the Study:

• To investigate the structural properties of scientific collaboration networks.
• To determine if these networks exhibit small-world characteristics.
• To compare collaboration patterns across different scientific disciplines.

Main Methods:

• Constructed explicit collaboration networks using publication data from major scientific databases.
• Included data from MEDLINE (biomedical research), Los Alamos e-Print Archive (physics), and NCSTRL (computer science).
• Analyzed network properties including path lengths, collaborator numbers, and clustering coefficients.

Main Results:

• Scientific collaboration networks demonstrate a "small world" phenomenon, characterized by short average path lengths between scientists.
• Analysis revealed the distribution and mean number of collaborators per author.
• Evidence of significant clustering within the collaboration networks was observed.
• Distinct differences in collaboration patterns were identified among the studied scientific fields.

Conclusions:

• Scientific collaboration networks are small worlds, facilitating rapid information flow.
• The findings highlight the importance of network structure in scientific communication.
• Disciplinary differences in collaboration patterns suggest unique social and structural dynamics in different scientific fields.