Translation Quality Ratings
MQM-style error annotations by professional Japanese translators on machine translation output.
Pairwise-judgment schematic. Illustration only. This is not sample data, and no sample exists until one is produced for your request.
- Ready to License
- Partner Sourced
- Custom Collection
Recorded or built to your specification, per project.
- Sourcing on Request
Specification, volume and rights for this entry are not yet fixed and are defined with you.
- Data formats
- JSONL, TSV
- Annotation
- Error spans, Severity
- Language
- Japanese ↔ English
- Region
- Japan (scope to be defined per project)
- Modality
- Text
- Volume
- Not specified. Set per project.
- Price
- Quoted per project.
Target models
MT evaluation
Use cases
Model Evaluation
How it is collected
Collected or authored to your specification by recruited Japanese participants or experts, under written consent and a project-specific protocol.
Quality assurance
Native-speaker review of a sample set before delivery, schema and format validation, duplicate checks, and a written QA report agreed in advance.
Licensing & usage
Delivered under a written license or work-for-hire agreement defining permitted AI training and evaluation use, territory, term and any exclusivity. Contributor consent is documented.
Need this, or something adjacent?
Describe the data you need. We will respond with feasibility, rights approach and next steps.