ArabicNLPWorld/holy-quran

🕌 The Holy Quran — Tanzil Simple Clean

Description

Complete text of the Holy Quran (6,236 ayahs) in Parquet format.
Source: Tanzil Project — Simple Clean text.
Prepared by Mullosharaf Arabov for IslamicEval 2026 Shared Task (Subtask 2: Hallucination Detection).

📊 Dataset Statistics

Metric	Value
Total ayahs	6,236
Total surahs	114
Total juz	30
Total words	82,627
Total letters	332,837
Avg words/ayah	13.2
Avg letters/ayah	53.4

📦 Data Fields

Field	Type	Description
surah	int32	Surah number (1–114)
ayah	int32	Ayah number within the surah
juz	int32	Juz number (1–30)
text	string	Arabic text of the ayah
word_count	int32	Number of Arabic words
letter_count	int32	Number of Arabic letters

📥 Loading the Dataset

from datasets import load_dataset

dataset = load_dataset("ArabicNLPWorld/holy-quran", split="train")

# First 3 ayahs
for example in dataset.select(range(3)):
    print(f"Surah {example['surah']}, Ayah {example['ayah']}, Juz {example['juz']}")
    print(f"{example['text']}")
    print(f"Words: {example['word_count']}, Letters: {example['letter_count']}\n")

🔬 Use in IslamicEval 2026

This dataset serves as the canonical reference base for:

• Building FAISS index of authentic Quranic texts
• Comparing LLM-generated fragments against genuine ayahs
• Generating synthetic training data for hallucination detection

📜 Source & License

Quran text: Tanzil Project (Simple Clean).
License: CC BY-SA 4.0

🏷️ Citation

@misc{mullosharaf_quran_2026,
  author = {Mullosharaf Arabov},
  title = {The Holy Quran — Tanzil Simple Clean (Parquet)},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/ArabicNLPWorld/holy-quran}
}

---

Prepared with dedication for IslamicEval 2026. Inshallah, victory!