MultiCoNER Datasets
Amazon
Multilingual named-entity recognition data with hard cases like movie titles and short text.
MultiCoNER 1 is a large multilingual named-entity recognition dataset in 11 languages, built to represent contemporary challenges: short, uncased text, syntactically complex entities such as movie titles, and long-tail entity distributions. MultiCoNER 2 extends the challenge to fine-grained entities.
Access
- Cost
- FreeNo charge to download; hosted in an AWS S3 bucket.
- License
- Open licenseCC BY 4.0
- Sign-up
- No sign-up
- How you get it
- API and Bulk downloadsee documentation
What's in it
- One row is
- One sentence with entity spans and labels
- Covers
- Not geographic (11 languages in version 1)
- Years
- Not stated
- Updated
- One-time release
- Size
- Not stated
Quality
Purpose-built to be hard, with shared-task evaluation. Mostly short, partly synthetic-style text.
- Short context
- Annotation noise in long-tail entities
Connects to
- Common Crawl: Web text source for building multilingual NER data
- Sudachi Language Resources: Language resources for one of the languages
Questions it could answer
- Which entity types are hardest across languages?
- How much does limited context hurt NER accuracy?