metadata

license: mit
task_categories:
  - text-classification
  - translation
  - text-generation
  - text2text-generation
language:
  - en
  - hmr
size_categories:
  - 10K<n<100K
tags:
  - machine-translation

Dataset Card for Hmar Bible Dataset

This dataset contains parallel verses in English and Hmar languages, curated from the American Standard Version (ASV) for English and the Christian Literature by Bible Society of India (CLBSI) for Hmar. It is structured to facilitate linguistic and translation studies.

Dataset Details

Dataset Description

This dataset provides verse-level parallel translations of the Bible in English (ASV) and Hmar (CLBSI). Each book is saved as a separate CSV file with the structure:

en	hmr
"In the beginning..."	"A tîrin Pathienin hnuoi..."

Curated by: Hmar Language Development Project
Language(s) (NLP): English, Hmar
License: CC-BY-4.0

Dataset Sources

Repository: Hmar Bible Dataset on GitHub
Paper [optional]: N/A
Demo [optional]: N/A

Uses

Direct Use

This dataset can be used for:

Linguistic analysis and translation studies.
Training or fine-tuning multilingual and low-resource language models.
Cross-language information retrieval tasks.

Out-of-Scope Use

The dataset is not suitable for non-academic or commercial purposes that violate the licensing agreement.
It is not recommended for tasks requiring contextual annotations, as it only includes raw verses.

Dataset Structure

The dataset consists of separate CSV files for each book of the Bible, named as <Book_Name>.csv (e.g., Genesis.csv).

Columns:
- en: Verse content in English from ASV.
- hmr: Verse content in Hmar from CLBSI.

Example CSV structure for Genesis.csv:

en	hmr
"In the beginning..."	"A tîrin Pathienin hnuoi..."
"And the earth was..."	"Chun, hnuoi hi a sie..."

Corrections

During dataset creation, mismatched rows between the ASV Bible and CLBSI Bible were identified and corrected. The discrepancies were found in the following books:

3 John: ASV contains 14 verses, while CLBSI contains 15 verses.
Ephesians: ASV contains 155 verses, while CLBSI contains 154 verses.
Revelation: ASV contains 404 verses, while CLBSI contains 405 verses.

These mismatches were manually reviewed and resolved to ensure alignment in the final dataset.

Observations

The translations, while attempting to be literal, cannot achieve a 100% word-for-word alignment due to inherent grammatical and linguistic differences between English and Hmar. The dataset reflects these nuances and captures the intended meaning as closely as possible.

Dataset Creation

Curation Rationale

The dataset was created to promote the preservation of the Hmar language and to facilitate linguistic studies and translation model development for low-resource languages.

Source Data

Data Collection and Processing

Source: The English data was collected from the ASV Bible, and the Hmar data was collected from the CLBSI Bible.
Processing: JSON files containing verses were parsed and matched on a verse-by-verse basis to create CSV files for each book.

Who are the source data producers?

The original texts are authored and compiled by:

The ASV Bible Committee.
Christian Literature by Bible Society of India (CLBSI).

Annotations

No additional annotations were made. The dataset consists solely of raw verse content.

Personal and Sensitive Information

This dataset does not contain personal or sensitive information. It includes only publicly available religious texts.

Bias, Risks, and Limitations

Bias:
- Translational differences between ASV and CLBSI may introduce semantic mismatches.
- Cultural and doctrinal differences may influence the translations.
Technical Limitations:
- Limited to verse-level translations. Contextual analysis may require additional data.
- Verses in ASV and Hmar may not always align perfectly due to structural differences in the source texts.

Recommendations

Users should ensure appropriate preprocessing and validation before using the dataset for critical applications.

Related Datasets

Citation

Glossary

ASV: American Standard Version of the Bible.

CLBSI: Christian Literature by Bible Society of India.

More Information

For more details, contact: [email protected]

Dataset Card Authors

Hmar Language Dataset Project

Dataset Card Contact

[email protected]

BibTeX:

@dataset{asv_clbsi_bible,
  title = {Hmar Bible Dataset},
  author = {Hmar Language Developme t Project},
  year = {2024},
  publisher = {https://huggingface.co/datasets/hmar-lang/Hmar-Bible-Dataset},
  license = {MIT}
}