SOCFACE, research project on the evolution of French society, based on AI

The interministerial service of the Archives of France (SIAF), in partnership with the Paris School of Economics, is launching the Socface research project, conducted by the National Institute of Demographic Studies (INED) and the company Teklia. The evolution of French society from 1836 to 1936 will be studied based on the twenty population censuses conducted during this period, preserved and digitized by the Departmental Archives. The interdepartmental service of the Archives of France is one of the services that make up the general direction of heritage and architecture. With the General Inspection of Heritage, it exercises, among other things, a scientific and technical control over the public archives still in the hands of the State's public services and establishments. With the Paris School of Economics, it has just launched the innovative research project Socface, which was selected last year by the Agence Nationale de la Recherche (ANR), the French research funding agency.

The SOCFACE project

The Socface project, funded by the ANR, aims to analyze the nominal lists of the census from 1836 to 1936 to produce a database of individuals who lived in France during this period. These lists provide information on an individual's name, age, sex, and sometimes occupation. Investigators in historical demography from INED, the "Economic and Social History" group of the Paris School of Economics, Teklia, a specialist in AI and automatic document processing, and the SIAF will collaborate on this project. The project is coordinated by Lionel Kesztenbaum, historian and demographer, director of research at INED.

Methodology

Socface will use machine learning to analyze 15 million images collected from a hundred departmental archives and build automatic processing models. The results will be validated by the team's historians, demographers and archivists. More than 700 million records will be processed using Arkindex, Teklia's automatic manuscript recognition and indexing solution.

Public data

The project started in September 2021 and should be completed in March 2025, when the database will be available on the FranceArchives portal. The departmental archives will be able to publish their own archives. Genealogical research will be simplified, as will research into economic and social history and the history of democracy. Those concerning the evolution of the labor market, inequalities, and migrations will also benefit from the knowledge acquired thanks to Socface. Translated from SOCFACE, projet de recherche sur l'évolution de la société française, basé sur l'IA