Place and Sentiment-based Life story Analysis

In 2008, the Network of Actors for the History and Memory of Immigration (RAHMI) launched an experimental gathering program to collect the forgotten memory of immigrant populations involved in the local community. Various groups of people were targeted by this collection, which made it possible to record life stories, including those of Spanish Republicans who went into exile in France between 1936 and 1939, and participated in the French Resistance. The MATRICIEL project (PEPS CNRS UPE 2016) focused on the migration of these Spanish Republicans in terms of the places mentioned in their stories, and the sentiments associated with these places. The project aimed to mainstream the migrants’ voices in the analysis; the objects of study chosen: the places, designated by a proper name: Barcelona, or a common name: internment camp, and the associated sentiments distinguished by their polarity: positive or negative, contribute to enhancing oral archives for the construction of an immigration memory. In this article, we present the approach implemented for a multidisciplinary analysis of the life story corpus, which combines methods and tools for natural language processing and mapping. The identification of common noun places mentioned in the stories was conducted through a supervised learning model. The identification and subsequent mapping of proper name places highlight the spatial distribution of the witnesses’ life courses, determined by the historical context and personal choices. The semi-automatic sentiment annotation adds polarity to the stories. In perspective, the analysis of common noun place types will make it possible to evaluate the granularity used by witnesses to describe their lived spaces; their location will help to specify the spatiality of the stories.