A computational method from the history of science for tracing how scientific knowledge systems change over time, operationalizing the semantic layer of the socio-epistemic networks framework. It pairs two diachronic techniques on a text corpus: relative-entropy measures (KL and Jensen-Shannon divergence) over unigram language models to detect and explain semantic shifts, and density estimation over document embeddings to track how semantic neighbourhoods concentrate or disperse, enabling comparison of an individual scholar's trajectory against the global evolution of a field.
This paper proposes a data-driven approach for tracking how knowledge systems evolve, operationalizing the socio-epistemic networks (SEN) framework on its semantic layer. Two complementary computatio…