Authors: Solène Tarride, Mélodie Boillet, Christopher Kermorvant

Affiliation: TEKLIA

Description: We train DAN for HTR and NER on records, using partial annotations.
Only words associated with named entities are taken into consideration during training. As a result, the model learns to ignore non-relevant words.
We use a unique tag combining the category and person information (ex: [name_wife]Maria).