Preparing the Dictionnaire Universel for Automatic Enrichment


The ​Dictionnaire Universel ​(DU) is an encyclopaedic dictionary originally written by Antoine Furetière around 1676-78, later revised and improved by the Protestant jurist Henri Basnage de Beauval who expanded, corrected and included terms of arts, crafts and sciences, into the ​Dictionnaire. The aim of the ​BASNUM project is to digitize the ​DU ​in its second edition rewritten by Basnage de Beauval,​ to analyse it with computational methods in order to better assess the importance of this work for the evolution of sciences and mentalities in the 18th century, and to contribute to the contemporary movement for creating innovative and data-driven computational methods for text digitization, encoding and analysis. Based on the experience acquired within the research group, an enrichment workflow based upon a series of Natural Language Processing processes is being set up to be applied to Basnage's work. This includes, among others, automatic identification of the dictionary structure (macro-, meso- and microstructure), named-entity recognition (in particular persons and locations), classification of dictionary entries, detection and study of polysemy markers, tracking and classification of quotation use (bibliographic references), scoring semantic similarity between the ​DU ​and other dictionaries. The main challenges being the lack of available annotated data in order to train machine learning models, decreased accuracy when using modern pre-trained models due to the differences between present-day and 18t​ h century French, and even unreliable or low quality OCRisation. The talk describes methods that are useful to tackle these issues in order to prepare the the ​DU for automatic enrichment going beyond what current available tools like Grobid-dictionaries can do, thanks to the advent of deep learning NLP models. The paper also describes how these methods could be applied to other dictionaries or even other types of ancient texts.

Jun 13, 2019 11:00 AM — 11:30 AM
Leeuwarden, Pays-Bas
Pedro Javier Ortiz Suárez

Je suis doctorant en informatique à Sorbonne Université et à l'équipe de recherche ALMAnaCH à Inria