The spatiotemporal relationship between usage data and topic popularity in scientific literature

Abstract

This study explored the spatiotemporal relationship between usage data (measured by PDF downloads and HTML views) and topic popularity (measured by the number of publications) in scientific literature. Using a panel dataset of over 2.3 million papers and 130 million usage records from IEEE Xplore, we develop a theoretical framework grounded in attention economy theory and the competitive exclusion principle. By using fixed effects model, the instrumental variable method, and the spatial Durbin model, we discover that how often a topic is used greatly increases its future popularity, while usage data from related topics have a negative impact. This study provides solid preliminary evidence for using usage data in detecting research hotspots. Additionally, this study innovatively proposes two methods for constructing spatial weight matrices based on topic semantic vectors, offering a concrete pathway for integrating spatial econometrics with spatial scientometrics.