CEDRR talk: From Archives to Algorithms: Digital Methods for Studying Imperial Commoners of Brazil and West Africa (1640-1822) by Agata Błoch

This presentation examines how digital and AI-assisted methods can broaden the study of imperial commoners in the Portuguese Empire, focusing on Brazil and West Africa between 1640 and 1822. Drawing on case studies from the Portuguese Overseas Historical Archives, it shows how computational approaches, such as Named Entity Recognition, Sentiment Analysis, Topic Modeling, and LLMs, among others, allow historians to process and analyze vast archival collections that would otherwise take decades to review manually. The study also emphasizes the continued importance of human expertise in AI-assisted historical research. Historians remain responsible for curating datasets, designing conceptual categories, interpreting results, and validating algorithmic outputs. Rather than replacing historical scholarship, artificial intelligence transforms the historian’s role, shifting it toward that of a domain expert who formulates research questions, oversees computational workflows, and evaluates algorithmic interpretations. The presentation also discusses methodological challenges in applying digital methods to early modern archival materials, including OCR noise, linguistic variation, limited annotated datasets, and structural biases in digital corpora. These challenges are especially significant for African archival materials, whose languages and historical vocabularies are underrepresented in digital resources.

Agata Błoch