Measuring academic cocoon from disparity and diversity perspectives

Guoxiu He, Jia Yuan, Yunhan Yang

Abstract: Personal academic preferences and information filtering mechanisms may create information cocoon akin to those observed in social media within the academic sphere. In this study, we propose a novel methodology for measuring academic cocoon by examining both academic disparity and diversity. Disparity indicates the distance between scholars’ current research and both their previous work and that of their peers; diversity reflects the variety of topics covered in scholars’ publications and the breadth of their references. Particularly, we derive the semantic and topic representations of the paper using Sentence-BERT and BERTopic, respectively. From these insights, we formulate five distinct metrics that capture these two dimensions. We conduct empirical experiments on a collection of papers in S2ORC from 1980 to 2020. The findings indicate that scholars across nearly all disciplines are experiencing a downward trend in academic cocooning. The correlation analysis between citation count, Disruptive Index (DI), and the degree of academic cocooning indicates that a smaller academic cocoon tends to result in higher academic impact. This study provides fresh insights into academic cocoon and individual career development within academia.

External IDs:dblp:journals/scientometrics/HeYY25