Parliamentary debates and Media space (1870-1940): understanding the circulation of political discourse using data-intensive methods – DECIDON

DECIDON aims to analyse the mechanisms used by the Parliament of the Third Republic (1870-1940) to set the agenda for public issues. Parliamentarians had considerable autonomy in choosing the topics to be discussed, and parliamentary debates seemed to follow their own life cycle, little influenced by the day-to-day fluctuations of public opinion, as embodied in press discourse. However, historical research has shown the complex relationship between the press and Parliament, and the constant interplay of influence between these two arenas. To gain a better understanding of how this system worked, while shedding a little more light on the daily lives of parliamentarians at the time, it is therefore necessary to assess the extent to which the agenda-setting of public issues was influenced by the choices made by MPs - and more generally, by power games within the parliamentary arena - or by the press, which embodied and gave shape to public opinion at the time. Such a study involves analysing the circulation of political discourse between the press and Parliament over the long term, and therefore exploring vast corpora of digitised historical documents. DECIDON will make it possible both to study the corpus of digitised French parliamentary debates, and to develop a complete and open pipeline for processing historical corpora, from the creation of the corpus to its online publication in the form of enriched FAIR data. Always informed by historical analyses, the results will be obtained by combining computer vision methods and natural language processing. DECIDON will demonstrate that the critical stages of document layout analysis, OCR and named entity recognition are now within reach with open, ready-to-use solutions. It also aims to model political discourse in fine detail, particularly by leveraging the capabilities of large language models.