The views expressed are those of the authors and do not represent the views of the SNF Agora Institute or Johns Hopkins University.
Authors: Samuel Backer, Louis Hyman
New LLM-based OCR and post-OCR correction methods promise to transform computational historical research, yet their efficacy remains contested. We compare multiple correction approaches, including methods for “bootstrapping” fine-tuning with LLM-generated data, and measure their effect on downstream tasks. Our results suggest that standard OCR metrics often underestimate performance gains for historical research, underscoring the need for discipline-driven evaluations that can better reflect the needs of computational humanists.
-
-
Academic Article August 13, 2025“We don’t fit at all”: the symbolic boundaries of libertarians connected to the academyIntellectual Pluralism
Amy Binder -
Academic Article July 08, 2025MapAgora, civic opportunity datasets for the study of American local politics and public policy
Hahrie Han -
Monica Prasad
