Extracting Data From the Bulletin de la Convention Nationale

Emilien Arnaud (History) used generative AI to “produce human-like transcriptions” of scans from the Bulletin de la Convention Nationale, “the official journal of Revolutionary France’s National Assembly.”

Despite the importance of the journals, Emilien explained, “historians have not yet engaged with them in a systematic way. My aim was to bridge that historiographical gap by obtaining transcriptions of the journal and conducting text-as-data analysis on the resulting corpus.”

Although considerable work remains, Emilien noted that “the model performed exceptionally well on brief passages (under 300 words), achieving nearly 100 percent accuracy—I found no errors in the sample of files I validated manually.”