Emerging Knowledge Trend in Statistical Research: A Content-Based Analysis Using Covariate-Assisted Dynamic Topic Model

Exploring the emerging knowledge trends within a particular discipline is of great interest to scientific researchers. It helps researchers to understand the historical development of their disciplines and to guide their future research directions. In this work, we focus on the fast-developing discipline, statistics, to investigate its knowledge trend and emerging topics in statistical research. To this end, we collect publications in top-tier statistical journals and statistical-related conferences, and develop a covariate-assisted dynamic topic model (CDTM). It captures the dynamic evolution of topics in statistical publications and also finds the time-varying effects of covariates on topic discussions. To estimate CDTM, a variational inference procedure is applied. The theoretical properties are studied and finite-sample performance is evaluated through simulation experiments. Last, we apply CDTM to the collected academic data. We highlight the advantages of CDTM over other alternative methods and uncover topics that characterize the evolution of statistical knowledge over the past four decades. We also observe that certain research topics in statistics are increasingly aligning with advancements in artificial intelligence. Based on these findings, we gain valuable insights into the historical progression of statistical research, which enables us to better anticipate future trends and guide innovation in the field.