☪ Chinese-Muslim: Comprehensive Chinese Muslim Knowledge Base & RAG Corpus

Dataset Description

Chinese-Muslim is the flagship dataset of the Salaamalykum project — a comprehensive knowledge base containing 279 articles covering the full spectrum of Chinese Muslim life: culture, travel, mosques, halal food, religious practice, history, and community documentation across 20+ countries. Every article preserves complete full-text content with all 16,560 embedded image references.

Key Properties

Property	Value
Total Records	279
Language	Chinese (Simplified, zh-CN)
Format	Parquet (auto Dataset Viewer enabled)
Splits	train (223), validation (27), test (29)
Avg. Article Length	14,599 characters
Total Characters	4,073,145
Total Image References	16,560
Articles with Images	275 / 279
License	MIT
Version	v2.0.0
Release Date	2026-06-13

---

Embeddable Dataset Viewer

Embed in any webpage:

<iframe
  src="https://huggingface.co/datasets/qurancn/Chinese-Muslim/embed/viewer/default/train"
  frameborder="0"
  width="100%"
  height="560px"
></iframe>

Direct viewer links:

• 🔍 Browse All (train)
• 🕌 Mosques
• 🍖 Halal Food
• ✈️ Travel
• 📖 Ramadan

---

Topic Distribution

Category	Count
Mosques & Architecture	139
Halal Food & Cuisine	80
Religious Life & Practice	29
Culture & Knowledge	11
History & Heritage	11
Travel & Exploration	7
Muslim Communities	2

---

Data Schema

Column	Type	Description	Queryable
id	string	Article ID from source platform	✅
title	string	Full article title	✅ Full-text
text	string	Complete article body with all Markdown images	✅ Full-text
text_preview	string	First 300 chars (Viewer-friendly preview)	✅
source_url	string	Canonical URL on salaamalykum.com	✅
content_hash	string	SHA-256 integrity hash	✅
author	string	Original article author	✅ Filter
language	string	BCP-47 language tag (zh-CN)	✅ Filter
topic_category	string	One of 7 categories	✅ Filter
char_count	int32	Character count of full article	✅ Sort
word_count	int32	Word/token count	✅ Sort
image_count	int32	Number of embedded image references	✅ Sort
has_images	bool	Whether article contains images	✅ Filter
source_platform	string	Source website identifier	✅ Filter

Note on images: All image URLs are preserved inline as Markdown ![image](url) within the text field. Images are hosted externally and render when viewing raw Markdown.

SQL Console Queries

Run directly in Data Studio:

-- Topic breakdown
SELECT topic_category, COUNT(*) as count, AVG(char_count) as avg_len
FROM train GROUP BY topic_category ORDER BY count DESC;

-- Longest articles
SELECT title, char_count, image_count, source_url
FROM train ORDER BY char_count DESC LIMIT 10;

-- Most image-rich articles
SELECT title, image_count, char_count
FROM train WHERE has_images = true ORDER BY image_count DESC LIMIT 10;

-- Search by keyword
SELECT title, text_preview, source_url
FROM train WHERE title LIKE '%Ramadan%';

---

Data Samples

Example: Authentic Halal Chinese Food Beijing: Xinjiang Restaurant, Y

Field	Value
id	salaamalykum.com
title	Authentic Halal Chinese Food Beijing: Xinjiang Restaurant, Yunnan Dish...
text_preview	# Authentic Halal Chinese Food Beijing: Xinjiang Restaurant, Yunnan Dishes & Niujie Snacks Reposted from the web Summa...
source_url	https://salaamalykum.com
topic_category	Halal Food & Cuisine
char_count	10332
image_count	43

Example: Authentic Hui Muslim Food in Beijing: Speed Pizza, Fujian Be

Field	Value
id	salaamalykum.com
title	Authentic Hui Muslim Food in Beijing: Speed Pizza, Fujian Beef and Tur...
text_preview	# Authentic Hui Muslim Food in Beijing: Speed Pizza, Fujian Beef and Turkish Qubbe Reposted from the web Summary: This...
source_url	https://salaamalykum.com
topic_category	Halal Food & Cuisine
char_count	15088
image_count	60

Example: Authentic Muslim Life Guide in the Muslim World: Shaban Virt

Field	Value
id	salaamalykum.com
title	Authentic Muslim Life Guide in the Muslim World: Shaban Virtues, Dua a...
text_preview	# Authentic Muslim Life Guide in the Muslim World: Shaban Virtues, Dua and Ramadan Preparation Reposted from the web S...
source_url	https://salaamalykum.com
topic_category	Religious Life & Practice
char_count	8891
image_count	1

---

Data Collection & Processing Pipeline

Source

All articles originate from salaamalykum.com, a community-driven platform documenting Chinese Muslim life, culture, and global Islamic heritage.

Extraction Method

Articles were extracted from the platform's content management system and converted to structured Markdown with YAML frontmatter. Each article preserves its original source URL, author attribution, and content integrity hash.

Processing Steps

1. Content Extraction: Articles extracted from WeCenter 3.1.9 CMS (OpenResty + PHP 7.0 + MySQL)
2. Format Standardization: Normalized to Markdown with YAML frontmatter (title, URL, author, timestamps, hash)
3. BBCode → Markdown: [img]url[/img] → ![image](url), [b]text[/b] → **text**
4. HTML Cleanup: Entity unescaping + whitespace normalization
5. SHA-256 Hashing: Content integrity verification per article
6. Topic Classification: Rule-based keyword classifier into 7 categories
7. Parquet Serialization: pyarrow with row_group_size=50, write_page_index=True
8. Split Generation: Deterministic shuffle (seed=42), 80/10/10 train/validation/test

Known Limitations

• Image hosting: 16,560 image URLs reference external hosts; not embedded as binary
• Some articles may contain residual formatting artifacts from CMS migration
• Temporal coverage: approximately 2020–2026

---

Related Datasets

Dataset	Articles	Focus
Chinese-Muslim (this)	279	Full-spectrum Chinese Muslim knowledge
Chinese-Muslim-Travel	347	Muslim travel & halal tourism
Islamic-Culture	260	Islamic culture & heritage

---

Usage

Python (Hugging Face Datasets)

from datasets import load_dataset

ds = load_dataset("qurancn/Chinese-Muslim")
print(ds)

# Access full article with all images
article = ds["train"][0]
print(article["title"])
print(article["text"])       # Complete body with ![image](...) references
print(article["image_count"])

LangChain RAG

from langchain.vectorstores import FAISS
from langchain.embeddings import HuggingFaceEmbeddings
from datasets import load_dataset

ds = load_dataset("qurancn/Chinese-Muslim", split="train")
texts = [f"{row['title']}\n{row['text']}" for row in ds]
embeddings = HuggingFaceEmbeddings(model_name="shibing624/text2vec-base-chinese")
vectorstore = FAISS.from_texts(texts, embeddings)

---

Version History

Version	Date	Changes
v2.0.0	2026-06-13	Full production rebuild: 279 articles, Parquet, 3 splits, 14-column schema, complete text + 16,560 images
v1.0.0	2026-06-13	Initial legacy upload

---

Citation

@dataset{chinese_muslim_2026,
  title={Chinese-Muslim: A Comprehensive Chinese Muslim Knowledge Base and RAG Corpus},
  author={Salaamalykum Project},
  year={2026},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/qurancn/Chinese-Muslim}
}

---

License

MIT License

Contact

• Email: bropeace@protonmail.com
• GitHub: https://github.com/salaamalykum
• Website: https://salaamalykum.com