NEM.fr

Full Official Name: Named-Entities Multigenre French Corpus
Submission date: Feb. 20, 2026, 11:21 a.m.

The corpus gathers 51,480 tokens distributed across 12 genres (from poetry to legal decisions, biomedical texts, spoken and more). This diversity makes the corpus suitable not only for evaluating the performance of Named Entity Recognition (NER) tools in heterogeneous contexts, but also for exploring textual variation, genre effects or register differences. Each sample contains approximately 1,000 tokens, except for some speeches and emails, where the natural segmentation is preserved.

Creator(s)
Alice Millour
Marina Seghier
Distributor(s)
-
Right Holder(s)
-
Status : Accepted
ISLRN :
857-654-609-197-8
Version
1.0
Source
https://github.com/ayusekyo111/NEM.fr
Resource Type
Annotated Corpus
Media Type
Text
Language(s)
French
Access Medium
Web Download