GROBID for Humanities When engineering meets History

Charles Riondet (1) , Luca Foppiano (1)

1 ALMAnaCH - Automatic Language Modelling and ANAlysis & Computational Humanities

In this presentation we explore the relationship between humanists and computer scientists and the crucial need of scientific crossover, based on our experience in the interdisciplinary team ALMAnaCH at Inria, which gathers people with very different backgrounds. We focus on the use and development of the GROBID suite. GROBID is a tool initially built for extracting metadata from scientific articles. Over the years it has evolved with new features and been used in new domains (for example archival documents), with the help of specialists in the field concerned. In our example, we use it to identify mentions corresponding to the actors of armed conflicts in historical personal diaries. This is a Named Entity Recognition task made more complex by the specialized terminology due to the period (Second world war) and the presence of constraints of writing (clandestinity).

Keywords: NER; NLP; Digital Humanities; History

Domains: History; Library and information sciences

Charles Riondet, Luca Foppiano. GROBID for Humanities When engineering meets History. Text as a Resource. Text Mining in Historical Science, Institut Historique Allemand, Jun 2017, Paris, France. ⟨hal-01585693⟩