ArabicNLPWorld/canonical-islamic-corpus

🕌 Canonical Islamic Corpus (Quran + Hadith)

Description

Comprehensive corpus of authentic Islamic texts:

• 6,236 verses of the Holy Quran from Tanzil (Simple Clean)
• 315,913 unique hadith matns from four curated collections
  Total: 322,149 texts with rich metadata.

Prepared by Mullosharaf Arabov for IslamicEval 2026 Shared Task (Subtask 2: Hallucination Detection).

📊 Corpus Statistics

Metric	Value
Total entries	322,149
Quran verses	6,236 (source: tanzil)
Unique hadith matns	315,913
Hadith sources	3 collections
Top collections	Sahih al Bukhari, Sahih Muslim, Sunan Abu Dawud, Sunan an Nasai, Sunan Ibn Majah

📦 Data Fields

Field	Type	Description
id	int32	Unique record ID (0–322148)
text	string	Arabic text (ayah or hadith matn)
source	string	Origin dataset (e.g., tanzil, fawazahmed0/hadith-data)
type	string	quran or hadith
surah	int32	Surah number (1–114, 0 for hadith)
ayah	int32	Ayah number within surah (0 for hadith)
collection	string	Hadith collection name (e.g., Sahih Bukhari, Sunan Abu Dawud)
section	string	Chapter/section within the collection
hadith_number	string	Hadith number as string

📥 Loading the Dataset

from datasets import load_dataset

dataset = load_dataset("ArabicNLPWorld/canonical-islamic-corpus", split="train")

# Quran example
for ex in dataset.filter(lambda x: x['type'] == 'quran').select(range(2)):
    print(f"Surah {ex['surah']}, Ayah {ex['ayah']}: {ex['text']}")

# Hadith example with collection info
for ex in dataset.filter(lambda x: x['type'] == 'hadith' and x['collection']).select(range(3)):
    print(f"[{ex['collection']}] ({ex['section']}) No.{ex['hadith_number']}")
    print(f"{ex['text'][:200]}...\n")

📖 Example Records

Quran:

{
  "id": 0,
  "text": "بسم الله الرحمن الرحيم",
  "source": "tanzil",
  "type": "quran",
  "surah": 1,
  "ayah": 1,
  "collection": "",
  "section": "",
  "hadith_number": ""
}

Hadith:

{
  "id": 6236,
  "text": "حَدَّثَنَا عَبْدُ اللَّهِ بْنُ مَسْلَمَةَ ...",
  "source": "fawazahmed0/hadith-data",
  "type": "hadith",
  "surah": 0,
  "ayah": 0,
  "collection": "Sunan Abu Dawud",
  "section": "Purification (Kitab Al-Taharah)",
  "hadith_number": "1.0"
}

📚 Top Hadith Collections

• Sahih al Bukhari (64,182)
• Sahih Muslim (54,799)
• Sunan Abu Dawud (40,556)
• Sunan an Nasai (32,426)
• Sunan Ibn Majah (30,263)
• Jami At Tirmidhi (26,872)
• Muwatta Malik (13,493)
• Sahih Bukhari (5,688)
• Sunan Ibn Majah (3,952)
• Sahih Muslim (3,536)

🔬 Use in IslamicEval 2026

This corpus is the canonical reference base for:

• Building FAISS index for authentic text retrieval
• Detecting hallucinated Quranic verses and hadith matns
• Generating synthetic training data for Subtask 2

📜 Sources & License

• Quran: Tanzil Project (Simple Clean)
• Hadith: fawazahmed0/hadith-data, arbml/LK_Hadith, M-AI-C/all_hadith, Mahadih534/all-hadiths-cleaned
• License: CC BY-SA 4.0

🏷️ Citation

@misc{mullosharaf_islamic_corpus_2026,
  author = {Mullosharaf Arabov},
  title = {Canonical Islamic Corpus (Quran + Hadith)},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/ArabicNLPWorld/canonical-islamic-corpus}}
}

---

Prepared with dedication for IslamicEval 2026. Inshallah, victory!