Datasets:

---

dogdoh

/

dari-enko-corpus-sample

like 0

Hi everyone!

I'm sharing Dari (다리, Korean for "bridge") — the largest open EN↔KO parallel corpus for technical translation.

What's included

• This sample: ~5,000 parallel sentence pairs across Patent, Medical, IT, Legal, General domains
• Glossary: 5,000 bilingual terminology entries with domain tags
• Quality metrics: Each pair includes quality scores (0-1 scale)

Full corpus stats

• 67.2 million EN↔KO sentence pairs
• 844K bilingual glossary terms
• LaBSE + COMET quality scoring

Who is this for?

• NLP researchers working on EN↔KO translation
• Professional translators needing reference concordance
• Localization teams building domain-specific MT engines
• MTPE practitioners benchmarking MT output

API Access

Free tier: 50K chars/month at https://dari.is-a.dev/docs

Questions and feedback welcome!