freococo/arabic_names_with_myanmar_and_english_transliteration

Arabic Names with Myanmar and English Transliteration

This dataset provides a collection of 23,074 Arabic names (primarily Hadith narrators and Islamic figures) with their corresponding English and Myanmar (Burmese) transliterations.

It is designed to help with NLP tasks involving name entity recognition, transliteration, and translation between Arabic, English, and Myanmar.

Dataset Structure

The dataset contains the following fields:

• name_id (int64): Unique identifier (matches the original Hadith Narrators source).
• name_ar (string): The original name in Arabic script.
• name_my (string): The transliteration of the name into Myanmar (Burmese) script.
• name_en (string): The transliteration of the name into English.

Data Sample

name_id	name_ar	name_my	name_en
1	محمد	မုဟမ္မဒ်	Muhammad
2	ابو بكر	အဗူ ဗကရ်	Abu Bakr
3	عمر بن الخطاب بن نفيل	အိုမရ် ဗင်န် ခတ်သွ်သွာဗ် ဗင်န် နိုဖိုင်းလ်	Umar ibn al-Khattab ibn Nufayl
4	عثمان بن عفان	အွတ်ဆ်မာန် ဗင်န် အဖ္ဖားန်	Uthman ibn Affan
5	علي بن ابي طالب بن عبد المطلب	အလီ ဗင်န် အဗူ သွာလိဗ် ဗင်န် အဗ္ဗဒုလ် မုသွာလိဗ်	Ali ibn Abi Talib ibn Abd al-Muttalib
6	طلحة بن عبيد الله	သွာလ်ဟာ ဗင်န် အိုဗိုင်ဒုလ္လာဟ်	Talhah ibn Ubaydullah

Usage

You can load this dataset easily using the Hugging Face datasets library.

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("freococo/arabic_names_with_myanmar_and_english_transliteration")

# Print the first example
print(dataset['train'][0])

# Output:
# {'name_id': 1, 'name_ar': 'محمد', 'name_my': 'မုဟမ္မဒ်', 'name_en': 'Muhammad'}

Using Pandas

import pandas as pd

df = pd.read_parquet("hf://datasets/freococo/arabic_names_with_myanmar_and_english_transliteration/data/train.parquet")
print(df.head())

Dataset Creation & Curation

Source Data

The original list of Arabic names and IDs was sourced from the Hadith Narrators Dataset on Kaggle by fahd09.

Transliteration Process

1. Extraction: The Arabic names were extracted from the original dataset.
2. AI Generation: Google Gemini Flash was used to generate initial transliterations for both English and Myanmar.
3. Human-in-the-Loop (HITL):
   • The AI output was reviewed and refined by human annotators.
   • Special attention was paid to Myanmar transliteration to ensure it follows standard phonetic conventions used in the Myanmar Islamic community (e.g., using အဗ္ဗဒုလ် for 'Abdul' rather than a raw phonetic mapping).

Limitations

• Transliteration Variants: Transliteration is often subjective. For example, "Muhammad" could also be written as "Mohammad", and "မုဟမ္မဒ်" could vary based on regional pronunciation. This dataset attempts to use the most standard forms.
• Historical Context: These are primarily historical names (Hadith narrators), so the naming conventions reflect that era (e.g., frequent use of "ibn" / "bin").

Citation & License

If you use this dataset, please credit the original source of the data and this repository for the transliteration work.

Original Data Source: Hadith Narrators by fahd09 (Kaggle)

@misc{arabic_names_myanmar_transliteration,
  author = {freococo},
  title = {Arabic Names with Myanmar and English Transliteration},
  year = {2025},
  publisher = {Hugging Face},
  journal = {Hugging Face repository},
  howpublished = {\url{https://huggingface.co/datasets/freococo/arabic_names_with_myanmar_and_english_transliteration}}
}