Papers with NER.We
OpenNER 1.0: Standardized Open-Access Named Entity Recognition Datasets in 50+ Languages (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing datasets are not consistently formatted and use a variety of chunk encodings (IOB, BIO, etc.), often without documentation. |
| Approach: | They present OpenNER 1.0, a standardized collection of openly-available named entity recognition (NER) datasets. |
| Outcome: | The proposed datasets correct annotation format issues and provide a structure that enables research in multilingual and multi-ontology NER. |