This project investigates how scientific research fields emerge, evolve, and decline by analyzing large-scale publication data from arXiv. Our approach integrates data-driven metrics and physics-inspired models to characterize the evolution of research topics and the trajectories of researchers within the scientific landscape.
We begin by identifying the stages of development of scientific categories—emergence, growth, maturity, and decline—based on temporal changes in publication activity. In our PLoS ONE article, we show that different phases are associated with distinct author-level and article-level properties. For instance, growing fields tend to attract more diverse and mobile researchers, while mature ones consolidate around established contributors and dense citation patterns.
In parallel, we explore how researchers navigate this landscape. In our article in EPJ Data Science, we introduce a method to represent individual publication paths in a low-dimensional scientific embedding space. This reveals two prototypical trajectories: explorers, who frequently shift across fields, and exploiters, who deepen their specialization. These trajectories align with broader dynamics of knowledge mobility, which we find follows patterns similar to human migration flows.
Together, these studies provide a framework to map the rise and fall of research areas and understand the cognitive mobility of scientists. Our methods can be used to anticipate topic transitions and design policies for supporting interdisciplinary or emergent research.
Interaction Data Lab 