Making Invisible College Visible: Comparing ASIS&T and CAIS Annual Conferences as Two Scientific Communities

Kwan Yi, Ping Li, Tao JinProceedings of the Annual Conference of CAIS / Actes du Congrès Annuel de l ACSI 2018 · DOI

TL;DR

This paper uses social network analysis to visualize and compare the structures of two major Library and Information Science (LIS) academic conferences, ASIS&T and CAIS, as scientific communities over 23 years. By treating conferences as "invisible colleges," the study reveals how these communities are organized and how they differ, offering insights into the dynamics of scientific collaboration.

What Problem Does This Paper Address?

The paper tackles the challenge of understanding "invisible colleges" – loosely formed groups of scientists with shared interests and norms. While these communities are crucial for scientific progress, their intangible nature makes them difficult to study. The authors propose that academic conferences, often overlooked as research subjects, can serve as tangible and observable channels to map the structure and dynamics of these scientific communities. Specifically, they aim to compare the social structures of two major LIS conferences, ASIS&T and CAIS, to identify patterns, similarities, and differences in how these communities are organized.

Methods

The study employs a Social Network Analysis (SNA) approach. This involves conceptualizing individuals as "nodes" and their connections (co-authorship) as "links" or "ties" within a network. The authors utilize three key centrality measures:

  • Degree centrality: Measures the number of relationships an individual has.
  • Closeness centrality: Focuses on the shortest path distance between an individual and all others in the network.
  • Betweenness centrality: Assesses how often an individual lies on the shortest path between other individuals.

Data was collected from conference proceedings (ASIS&T and CAIS/ACSI) between 1993 and 2015, focusing on authors of contributed or full papers. This data was used to create an author-to-paper matrix (publication matrix) and subsequently an author-to-author matrix (coauthorship matrix). SNA software (UCINET and NetDraw) was used for analysis and visualization.

Datasets

The study uses data from the ASIS&T (Association for Information Science and Technology) and CAIS/ACSI (Canadian Association for Information Science) annual conference proceedings from 1993 to 2015. This includes author, bibliographic, and affiliation information for contributed or full papers. The authors note that keynote speeches, panels, and posters were excluded to focus on a core set of active members and maintain a manageable dataset.

Key Insights and Ideas

  • Conferences as Proxies for Scientific Communities: The core idea is that academic conferences, despite their transient nature, offer a valuable and visible lens through which to study the underlying, often invisible, scientific communities.
  • Visualizing Invisible Structures: SNA provides a powerful toolkit to map and understand the complex relationships and structures within these communities, making abstract connections concrete.
  • Comparing Community Dynamics: By analyzing two distinct but related conferences, the study highlights how different organizational cultures and scopes can lead to variations in community structure and collaboration patterns.

Key Results

  • Scale Differences: The ASIS&T community is larger, with 1,113 contributed papers from 1,488 authors, compared to the CAIS/ACSI community's 872 full papers from 777 authors.
  • Collaboration Patterns: ASIS&T shows a higher prevalence of collaborative papers, with two-author papers being the dominant mode (34%). In contrast, single-author papers dominate the CAIS community (50%).
  • Average Authors per Paper: ASIS&T has an average of 2.6 authors per paper, while CAIS/ACSI averages 1.8.
  • Co-authorship Density: The number of co-authors per author is similar between the two communities (ASIS&T: 2.97, CAIS/ACSI: 2.26).
  • Community Fragmentation: The ASIS&T community exhibits more "components" (separate, unconnected subnetworks) than CAIS/ACSI, with nearly 1.5 times as many components (379 vs. 260).

Limitations and Future Work

Limitations: * The study excluded authors of keynote speeches, panels, and posters, which might not fully represent the entire community. * The data collection focused only on contributed or full papers, potentially missing other forms of scholarly contribution.

Future Work: The authors suggest that these findings can help community participants and managers better understand and nurture collaborative networks. Further research could explore how these structures evolve over time and how different types of contributions (e.g., posters, invited talks) might influence community dynamics.

Summary updated

Abstract

This paper reports an ongoing research that conceptualizes two major LIS academic conferences in NorthAmerica as two dynamic scientific communities: ASIS&T Annual Meetings and CAIS AnnualConferences. Using a social network analysis approach, this study compares and visualizes how these twoscientific communities have been structured and organized between 1993 and 2015.
BibTeX
@inproceedings{yi2018making,
  author = {Kwan Yi and Ping Li and Tao Jin},
  title = {{Making Invisible College Visible: Comparing ASIS\&T and CAIS Annual Conferences as Two Scientific Communities}},
  booktitle = {Proceedings of the Annual Conference of CAIS / Actes du Congrès Annuel de l ACSI},
  year = {2018},
  doi = {10.29173/cais1038},
}
Figure