Tasks/Named Entity Recognition

Named Entity Recognition

নামীয় সত্তা শনাক্তকরণ

Detecting person, location, organization and other entities in Bangla text.

6 datasets·35 papers
DatasetSizeLicenseYearSource paperLink
ANCHOLIK-NER
Bangla regional NER across five dialects with 10 entity types, BIO-tagged.
17,405 sentencesCC BY 4.02025ANCHOLIK-NER (2025)
✓ verified 2026-07-19
Bangla-MedER
Manually annotated medical text covering six entity types, with a separate English translation.
2,980 textsCC BY 4.02025Bangla-MedER (Data in Brief 2026)
✓ verified 2026-07-26
BanNERD
Human-annotated Bangla NER corpus spanning 29 domains and 10 entity classes.
85,175 sentencesCC BY-NC-ND 4.02025BanNERD (Findings NAACL 2025)
✓ verified 2026-08-23
B-NER
Largest gold-standard Bangla NER corpus, 8 entity types.
22,144 sentencesMIT2023B-NER (IEEE Access 2023)
✓ verified 2026-07-19
MultiCoNER (bn)
Complex NER: creative works, groups, products; Bangla track.
15,300 sentencesCC BY 4.02022SemEval-2022 Task 11
✓ verified 2026-06-10
WikiANN (bn)
Silver standard Wikipedia PER/LOC/ORG annotations in the Rahimi et al. balanced splits; Bengali train/validation/test contain 10,000/1,000/1,000 sentences. The license field records the original release's ODC-By and 'For research use only' statements.
12,000 sentencesODC-By; research use only2019Massively Multilingual Transfer for NER (ACL 2019)
✓ verified 2026-09-08