DavidCBaines/ebible_corpus

Dataset Card for eBible Corpus

Dataset Summary

The eBible Corpus is a large multilingual parallel corpus containing Bible translations sourced from eBible.org. It features text from many low-resource languages and 891 languages in total. This dataset is designed to facilitate research in machine translation, cross-lingual alignment, and linguistic analysis, particularly for low-resource languages.

The dataset consists of two main files:

1. bible_corpus.parquet: Contains the verse text aligned by a common verse reference (book, chapter, verse). Each column represents a specific translation identified by its translationId.
2. metadata.csv: Provides detailed metadata for each translation included in the Parquet file, such as language names, copyright information, license, and source details.

The total compressed size of the dataset files is approximately 1GB.

Supported Tasks and Leaderboards

This dataset can be used for various NLP tasks, including:

• Machine Translation: Especially for low-resource language pairs.
• Cross-Lingual Word Alignment: Identifying corresponding words or phrases across translations.
• Linguistic Analysis: Studying translation strategies, language evolution, or typological features across a wide range of languages.

Languages

The dataset covers hundreds of languages. The metadata.csv file contains details about each translation's language information (including languageCode, languageName, languageNameInEnglish).

Dataset Structure

Data Instances

bible_corpus.parquet

An example row looks like:

{
  "book": "GEN",
  "chapter": 1,
  "verse": 1,
  "eng-engESV": "In the beginning, God created the heavens and the earth.",
  "fra-frLSG": "Au commencement, Dieu créa les cieux et la terre.",
  "spa-spaRV1909": "EN el principio crió Dios los cielos y la tierra.",
  "...": "..."
}