Finding Scientific Communities In Citation Graphs: Convergent Clustering

TL;DR

This paper introduces a scalable method for detecting scientific communities in citation graphs by converging two clustering techniques, validated through expert assessment on immunology articles, revealing meaningful author communities.

Contribution

The study presents a novel convergent clustering approach for identifying scientific communities in large citation datasets, with validation through expert review and thematic analysis.

Findings

Effective detection of article clusters with strong thematic coherence

Identification of valid author communities based on article clusters

Demonstration of the approach's scalability and potential utility

Abstract

Understanding the nature and organization of scientific communities is of broad interest. The `Invisible College' is a historical metaphor for one such type of community and the search for such `colleges' can be framed as the detection and analysis of small groups of scientists working on problems of common interests. Case studies have previously been conducted on individual communities with respect to their scientific and social behavior. In this study, we introduce, a new and scalable community finding approach. Supplemented by expert assessment, we use the convergence of two different clustering methods to select article clusters generated from over two million articles from the field of immunology spanning an eleven year period with relevant cluster quality indicators for evaluation. Finally, we identify author communities defined by these clusters. A sample of the article clusters produced by this pipeline was reviewed by experts, and shows strong thematic relatedness, suggesting that the inferred author communities may represent valid communities of practice. These findings suggest that such convergent approaches may be useful in the future.