DiffusionScore: A Framework for Assessing Institutional Research Impact Through Influence Diffusion

Assessing an institution’s research output has traditionally relied mostly on the opinion of peers, but as the field of scientometrics progressed dramatically in recent years, the evolution of quantitative indicators gained popularity and became essential. The present approaches for evaluating an institution’s research output rely on citation data, that has been quantitatively analyzed and are predominantly data-centric models. It is observed that research productivity and impact are taken into account by many ranking systems, and the measurement of research impact is mostly dependent on citation-counting techniques and the volume of research produced. The recent ranking systems evaluate the reputation of academic institutions by employing network-based algorithms, like Page Rank which examine the citation network at the institution level. An Article’s outreach and causal impact are not completely taken into consideration by the aforementioned approaches which leads to incomplete quantification of metrics. In this study, we propose a framework that uses a data-agnostic influence diffusion model to measure the academic impact of the institutions in the citation network. The diffusion model is inspired by the phenomenon of heat diffusion. The suggested framework investigates the citation trajectory for each article published by the university in the research area, analyses its influence diffusion in citation networks, and provides methods, metrics, and processes that others could apply to similar problems or datasets. By leveraging transfer entropy-based techniques, our research goes beyond citation mapping to investigate the causal link between academic influence and the reputation of publication venues employed by universities.