Hugging Face's logo

Datasets:
m-aliabbas1
/
makhzan-urdu-corpus

Modalities:
Text
Formats:
parquet
Size:
100K - 1M
Libraries:
Datasets
pandas
Dataset card Data Studio Files
xet
 Community
1

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

πŸ•Œ Makhzan Urdu Corpus

The Makhzan Urdu Corpus is a high-quality, expert-curated collection of classical and modern Urdu texts. This dataset has been extracted and cleaned from the Makhzan GitHub repository and structured into the Hugging Face datasets library format.

πŸ“š Dataset Summary

The corpus contains over 6.26 million words of Urdu text from various literary and scholarly publications. Each document is tokenized into paragraphs and tagged with semantic structure when available (e.g., headings, body, lists). This dataset is valuable for Urdu NLP tasks like:

  • Language modeling
  • Text classification
  • Named entity recognition
  • Machine translation

πŸ“ Supported Tasks and Leaderboards

  • language-modeling: Build or fine-tune language models like BERT or GPT for Urdu.
  • text-classification-other: Analyze sentiment, topic, or discourse in Urdu paragraphs.

πŸ“‚ Dataset Structure

Each example in the dataset is a dictionary with the following fields:

{
  "text": "اردو Ϊ©Ψ§ ایک پیراگراف یہاں Ψ΄Ψ§Ω…Ω„ ہے۔"
}

All entries are paragraphs extracted from structured XML files.

πŸ”’ Dataset Statistics

Feature Value
Total paragraphs ~300,000
Total tokens ~6.26 million
Format Paragraph-level text
Language Urdu (ur)

πŸ“₯ How to Use 

from datasets import load_dataset

dataset = load_dataset("m-aliabbas1/makhzan-urdu-corpus")
print(dataset["train"][0])

πŸ› οΈ Dataset Creation

The original XML files were parsed using xml.etree.ElementTree. Paragraphs were extracted, cleaned, and de-duplicated. Any English text, footnotes, and formatting tags were removed during preprocessing.

πŸ“„ License

The dataset is distributed under a custom license. Usage is permitted for research and non-commercial applications. Redistribution of raw XML text may be restricted based on original source copyrights.

Refer to the original license for details.

βœ’οΈ Citation

If you use this dataset, please cite:

@misc{makhzan2020corpus,
  author = {Ahmed, Zeerak},
  title = {Makhzan: An Urdu Corpus for Computational Linguistics},
  year = {2020},
  url = {https://github.com/zeerakahmed/makhzan}
}

🀝 Contributions

  • Original data collection and annotation: Zeerak Ahmed
  • Dataset formatting and Hugging Face integration: Muhammad Ali Abbas
Downloads last month
10
Total file size:
22.5 MB