Japanese to English translation quality evaluation using xCOMET
Evaluate and ensure high-quality Japanese → English translations using automated quality metrics (xCOMET) with clear decision criteria and actionable outputs.
| Input | Type | Description |
|---|---|---|
| source_file | .ja.md file | Japanese source document |
| translation_file | .md file | English translation document |
| threshold |
| number (0-1) |
| Minimum acceptable score (default: 0.85) |
| Output | Type | Description |
|---|---|---|
| quality_report | Markdown | Embeddable report with scores and recommendations |
| segment_details | Table | Per-segment scores and issues |
| action_items | List | Specific corrections needed |
xcomet:xcomet_batch_evaluate for multi-segment evaluationsource_lang: "ja" and target_lang: "en"deepl:translate-text for re-translation when --fix is specifiedDetect .ja.md ↔ .md pairs in the specified path
- If single file specified, find its pair automatically
- If directory specified, find all pairs recursively
Extract translatable segments from both files:
1. Document title (# heading)
2. Section headings (## / ### headings)
3. Key paragraphs (first 2-3 sentences per section)
4. Table content
5. List items with substantive content
Skip:
- Code blocks (```)
- URLs and links
- Mermaid diagrams
- Pure formatting elements
Call xCOMET MCP with extracted pairs: