English-Odia Parallel corpus

Overview

• 80,437 English text followed by its Odia translation text pairs can be downloaded from our NMT model repo.
• Parallel pairs have been collected from many sources by many volunteers.

Sources

• Wikipedia Data dump
• Open Parallel Corpus
• OdiEnCorp 1.0
• TDIL - Technical strings 52,000 pairs-Data needs to be cleaned
• Global Voices - 328 sentences pairs
• Mann ki baat - 1000+ pairs
• Twitter:DoctorBabu - Around 100 Botanical terms En-Or pairs
• Rupesh Ranjan Panda - Around 300 generic En-Or pairs
• Krishna Kabi - 186 En-Or pairs

Additional links

• Extracting Parallel-text pairs from Wikipedia

Odia Monolingual corpus

• Monolingual Odia data has been extracted from Wikipedia.
• You can use this repo to fetch the latest dataset.
• Ready-made monolingual corpus (with ~17,000 wikipedia articles) can be found at Kaggle created by Gaurav.

Odia dictionary

• The dictionary data has been extracted from Odia Purnachandra Bhashakosha.
• The source code repository for the dataset are in: OdiaNLP/dictionary

---

Last update: 2023-03-27