Odia-English Parallel and Monolingual Text Corpus

OdiEnCorp 2.0 is a collection of Odia-English parallel sentences and Odia monolingual text. It was created by bigscience-data and served in the WAT 2020 English-Odia Indic Task. The dataset was last updated in December 2022.

Use Cases

• Train machine translation models using Odia-English parallel sentences.
• Build language models for Odia using the monolingual text data.
• Analyze text features from sources like Odia Wikipedia, websites, and books.

Strengths

• Data collected from multiple sources including Odia Wikipedia, websites, books, and dictionaries.
• Includes both parallel and monolingual text data for Odia language processing.
• Dataset was used in the WAT 2020 English-Odia Indic Task.

Limitations

• Specific row count, column structure, and file size are unknown.
• Data collection methods include OCR and web scraping, which may introduce noise.
• The dataset's temporal coverage and geographic scope are not specified.

Provenance

Source

bigscience-data

Collection Method

Collected via manual and machine learning techniques including web scraping and optical character recognition.

Freshness

Last updated on 2022-12-12.