Using Large Language Models to understand 'the masses' in nineteenth-century Dutch newspapers

Zenodo (CERN European Organization for Nuclear Research)

Abstract

This paper presents a computational classification pipeline using fine-tuned BERT-based language models to identify and categorize mentions of "the masses" (massa) in nineteenth-century Dutch newspapers from the Delpher corpus. The study distinguishes between human and non-human senses of the term, and further classifies human references as either physical crowd manifestations or abstract political concepts.

Authors 3

1. Duvera Herfst Aachen

   Leiden University

   Affiliation as printed

   Leiden University

2. Anne Heyer Aachen

   Leiden University

   Affiliation as printed

   Leiden University

3. Yann Ryan Aachen

   Leiden University

   Affiliation as printed

   Leiden University