The ANR DALiH is also involved in several ongoing research projects involving the creation and the diffusion in open of Armenian textual corpora.
Mapping Armenian Paris – cartographie des commerces de la diaspora arménienne à partir de la presse publiée en France au début du XXe siècle
The MAP project (Mapping Armenian Paris) offers an original approach to using collections of Armenian newspapers published in France at the beginning of the 20th century: creating a map of the commercial and industrial activity of the Armenian diaspora in Paris, through the recognition and analysis of advertisements in newspaper collections using artificial intelligence.
Project duration: 2026-2027
In collaboration with: BnF Datalab, EPITA, DISTAM, Calfa
PI : Chahan Vidal-Gorène
Machine learning for Epigraphic Study and Recognition Of the Armenian Patrimony (MESROP)
Computer vision and artificial intelligence for the detection, transcription, and analysis of Armenian inscriptions.
The MESROP project is developing an integrated computational chain for the detection, transcription, and analysis of Armenian inscriptions on photographic and 3D media.
It combines recent computer vision models (object detection, semantic/instance segmentation, similarity search) and automatic transcription approaches based on Handwritten Text Recognition (HTR), extended by a generative model inspired by the work of ITHACA for the reconstruction of damaged or illegible areas. One of the project’s deliverables is an online corpus of Armenian inscriptions, based on the corpus of printed inscriptions.
Project duration: 2026-2027
In collaboration with: DISTAM, Calfa, TUMO, École nationale des chartes-PSL
PI: Chahan Vidal-Gorène
Armenian Manuscripts of the Dulaurier collection
The “Digital Enhancement of the Dulaurier Collection” project aims to apply state-of-the-art Handwritten Text Recognition (HTR) and Computer Vision models and linguistic analysis methods to Armenian manuscripts from the Dulaurier collection.
This collection, held by the BnF, contains historical Armenian texts compiled by Édouard Dulaurier or copied by himself, focusing on medieval Armenian historiography. The project aims to make these texts accessible to scholars and the public through digitized resources, searchable texts, and integration with platforms like Gallica, and linguistic databases.
Platform: https://calfa.fr/bnf-datalab-armenian-collection
Project duration: 2022-2023
In collaboration with: BnF DataLab, DISTAM, Calfa, GREgORI
PIs: Chahan Vidal-Gorène, Bastien Kindt