Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
110 tokens/sec
GPT-4o
56 tokens/sec
Gemini 2.5 Pro Pro
44 tokens/sec
o3 Pro
6 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Betweenness Centrality as a Driver of Preferential Attachment in the Evolution of Research Collaboration Networks (1111.6804v4)

Published 29 Nov 2011 in cs.SI and physics.soc-ph

Abstract: We analyze whether preferential attachment in scientific coauthorship networks is different for authors with different forms of centrality. Using a complete database for the scientific specialty of research about "steel structures," we show that betweenness centrality of an existing node is a significantly better predictor of preferential attachment by new entrants than degree or closeness centrality. During the growth of a network, preferential attachment shifts from (local) degree centrality to betweenness centrality as a global measure. An interpretation is that supervisors of PhD projects and postdocs broker between new entrants and the already existing network, and thus become focal to preferential attachment. Because of this mediation, scholarly networks can be expected to develop differently from networks which are predicated on preferential attachment to nodes with high degree centrality.

User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (3)
  1. Alireza Abbasi (8 papers)
  2. Liaquat Hossain (14 papers)
  3. Loet Leydesdorff (196 papers)
Citations (375)

Summary

Betweenness Centrality as a Driver of Preferential Attachment in Research Collaboration Networks

The paper under review investigates the dynamics of preferential attachment within the landscape of scientific coauthorship networks, focusing on the domain of steel structures research. It seeks to decipher how various forms of centrality—specifically betweenness, degree, and closeness—affect the likelihood of new authors collaborating with established nodes in a network. Unlike traditional perspectives that emphasized degree centrality as a key determinant for network expansion, this work identifies betweenness centrality as a more potent predictor for the attachment of new nodes.

Methodology

The authors conducted a thorough analysis using a comprehensive dataset of publications from the "steel structures" discipline over a decade (1999-2009). They focused on how new nodes (authors) attach to existing ones and how these interactions change over time. Centrality measures—degree, closeness, and betweenness—were computed yearly to evaluate their correlation with the number of new authors and collaborative ties formed subsequently.

Key Findings

  1. Centrality Measures and Attachment:
    • Betweenness centrality emerged as a significantly stronger predictor of preferential attachment compared to degree and closeness centrality. Authors with high betweenness are more likely to attract new collaborators, primarily because they broker relationships between disparate parts of the network.
    • The influence of betweenness centrality on attracting new coauthors increased over the paper period, suggesting a growing importance of authors positioned as information and communication brokers.
  2. Network Dynamics:
    • The majority of new links are established between new authors themselves or between new and existing authors. However, a notable pattern is that existing authors predominantly re-collaborate with their prior coauthors rather than forming new connections.
    • While degree centrality holds a strong correlation with the number of overall coauthors (reflecting cumulative networks), its impact on attracting first-time collaborators is less significant compared to betweenness.
  3. Practical Implications:
    • The findings imply that established authors with strong betweenness centrality serve a pivotal role in facilitating network growth by acting as bridges. This highlights the potential for strategically nurturing or supporting these authors to foster, maintain, and expand collaborative efforts across a scientific network.

Theoretical and Future Directions

This paper enriches the theoretical understanding of network evolution by drawing attention to the nuanced roles that different centrality measures play in preferential attachment processes. It challenges the traditional emphasis on degree centrality by positioning betweenness centrality as a central driver in the evolution of collaborative networks.

Future research could expand on these findings by examining whether similar patterns hold in other scientific disciplines, thus providing a broader basis for understanding network dynamics. Further investigation could also explore additional network characteristics or environmental factors that might modulate these dynamics, as well as how emerging digital tools or collaborative platforms contribute to these evolving patterns.

Overall, this paper contributes important insights into the mechanisms underpinning network formation and evolution, highlighting strategic opportunities for leveraging key actors to catalyze scientific collaboration and innovation.