Historical insights at scale - A corpus-wide machine learning analysis of early modern astronomic tables

Aug 5, 2026

Upload by Ezra

Understanding how human knowledge evolves and spreads over time is challenged by the vast quantity of historical materials and the limited availability of specialist expertise. Digitization enables AI-supported analysis, and this study presents an atomization–recomposition workflow combining unsupervised machine learning with explainable AI. Using the “Sacrobosco Collection” (359 early modern printed astronomy editions, 76,000 pages), it reveals temporal and geographic patterns in knowledge transformation and the role of textbooks in shaping a unified mathematical culture.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Introduction

Scientific revolutions and historiography

Challenges of large-scale historical sources

Machine learning for historical archives

Sci E nc E AdvA nc ES | R ESEARch A Rticl E COMPUTER SCIENCE Historical insights at scale: A corpus-wide machine learning analysis of early modern astronomic tables Oliver Eberle1,2, Jochen Büttner2,3, Hassan el-Hajj2,4, Grégoire Montavon1,2,5, Klaus-Robert Müller1,2,6,7*, Matteo Valleriani2,4,8,9* Understanding the evolution and dissemination of human knowledge over time faces challenges due to the abundance of historical materials and limited specialist resources. However, the digitization of historical archives presentsan opportunity for AI-supported analysis. This study advances historical analysis by using an atomizationrecomposition method that relies on unsupervised machine learning and explainable AI techniques. Focusing on the“Sacrobosco Collection,” consisting of 359 early modern printed editions of astronomy textbooks from European universities (1472–1650), totaling 76,000 pages, our analysis uncovers temporal and geographic patterns in knowledge transformation. We highlight the relevant role of astronomy textbooks in shaping a unified mathematical culture, driven by competition among educational institutions and market dynamics. This approach deepens our understanding by grounding insights in historical context, integrating with traditional methodologies. Case studies illustrate how communities embraced scientific advancements, reshaping astronomic and geographical views and exploring scientific roots amidst a changing world. copyright © 2024 the Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. no claim to original U.S. Government Works. distributed under a creative commons Attribution license 4.0 (cc BY) . INTRODUCTION The European early modern period has traditionally been regarded asthe cradle of modern society, particularly highlighting advancementsin science and technology. Science itself has often been portrayed as a progressive process, culminating in a scientific revolution that propelled Europe into modernity. The works of alleged heroes of science such as Nikolaus Copernicus, Galileo Galilei, and Johannes Kepler were given special significance in this process. Their publications were frequently and sometimes still are regarded as pivotal moments, encapsulating the essence of the astronomical revolution of this era (1–9). These views, largely still dominating public reception, are rightly being challenged today. Recent history of science is beginning to overcome its Eurocentrism and adopt a more differentiated view on the processes that led to the emergence of science. In his influential work, The Structure of Scientific Revolutions (1962), Thomas Kuhn had already emphasized the role of scientific paradigms, moving away from focusing solely on the contributions of a few selected individuals to viewing scientific progress as a collective achievement of the broader scientific community (10) . Even earlier, scholars like Braudel (11) and Bloch (12, 13) aimed to bridge the spatial and temporal gaps between wellstudied singular events by analyzing a broader collection of sources. Today, modern approaches, informed by this historiographical legacy, 1Machine learning Group, technische Universität Berlin, Marchstr. 23, 10587 Berlin, Germany. 2BiFOld–Berlin institute for the Foundations of learning and data, 10587 Berlin, Germany. 3Max Planck institute of Geoanthropology, Kahlaische Str. 10, 07745 Jena, Germany. 4Max Planck institute for the history of Science, Boltzmannstr. 22, 14195 Berlin, Germany. 5department of Mathematics and computer Science, Freie Universität Berlin, Arnimallee 14, 14195 Berlin, Germany. 6department of Artificial intelligence, Korea University, Seoul 136-713, South Korea. 7Max Planck institute for informatics, Stuhlsatzenhausweg 4, 66123 Saarbrücken, Germany. 8institute of history and Philosophy of Science, technology, and literature, Faculty i–humanities and Educational Sciences, technische Universität Berlin, Stra

FAQs

What problem does the study address in analyzing historical knowledge?

It tackles the difficulty of studying the evolution and dissemination of human knowledge given the abundance of historical materials and limited specialist resources for traditional historical investigation.

Which dataset and time span are used for the analysis?

The study focuses on the “Sacrobosco Collection,” covering 359 early modern printed astronomy textbook editions from European universities, dated 1472–1650, totaling about 76,000 pages.

How does the proposed method support historical analysis?

It uses an atomization–recomposition approach built on unsupervised machine learning together with explainable AI, enabling discovery of temporal and geographic patterns in knowledge transformation.

Aug 5, 2026

Upload by Ezra