Towards transdisciplinary impact of scientific publications: A longitudinal, comprehensive, and large-scale analysis on Microsoft Academic Graph

摘要

This paper studies the transdisciplinary impact of scientific publications with a longitudinal, comprehensive, and large-scale analysis on the Microsoft Academic Graph (MAG) dataset. More specifically, this paper aims to understand to what extent publications in discipline A have impact on discipline B. To this end, we propose a novel method to characterize the degree to which a publication impacts another discipline instead of its original discipline. We consider the ratio of the number of citations in a certain discipline and that in the original discipline. We also adopt an OLS regression to identify the equation between the ratio and the affinity of discipline pair and find a clear positive relation. This inspires us to categorize a publication's degree of transdisciplinarity by setting up two thresholds, the top 95% and the bottom 95% confident interval curve (of the fitted line). Publications above the top 95% curve is categorized as transdisciplinary ones, those below the bottom 95% curve as domain-specific ones, and those between the two curve as normal publications. This categorization does not require any pre-defined framework for transdisciplinarity and offers an automatic way of definition by data distribution itself. We find that sociology, mathematics, physics, and chemistry account for a great proportion of transdisciplinary publications that influence other domains, and that medicine, biology, economics, and geology have the greatest proportion of domain-specific publications that show impact in the original discipline. Moreover, we observe a negative relation between the number of citations and the proportion of transdisciplinary publications. A longitudinal analysis presents that the proportion of transdisciplinary publications shows a slightly increase trend for years.