shiima/vejin-Dataset-Normalization-with-textColumn

Kurdish Books Dataset with Combined Text Field

Dataset Description

This dataset contains 18,565 Kurdish books with a combined text field that includes author, book name, title, and content in a structured format. All text has been preprocessed using the asosoft library.

Languages

• Central Kurdish (ckb)
• Kurdish (ku)

Dataset Structure

The dataset contains the following columns:

• author
• book
• title
• url
• content
• text

Text Field Format

The text column combines all metadata and content in the following Kurdish format:

[author] لە کتێبی [book] ناوەڕۆکی [title] :
[content]

This format makes it easy to use the entire record as a single text input for language models.

Data Processing

Text Column Creation

A new text column was created by combining:

• Author name
• Book title (with Kurdish connector "لە کتێبی")
• Chapter/section title (with Kurdish connector "ناوەڕۆکی")
• Main content

Text Preprocessing

Both the content and text columns were preprocessed using the asosoft library:

1. Text normalization (asosoft.Normalize) - Standardizes Kurdish characters
2. Punctuation normalization (asosoft.NormalizePunctuations) - Standardizes punctuation
3. Line trimming (asosoft.TrimLine) - Removes extra whitespace

Example Usage

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("shiima/vejin-Dataset-Normalization-with-textColumn")

# View a sample
sample = dataset['train'][0]
print(f"Author: {sample['author']}")
print(f"Book: {sample['book']}")
print(f"Title: {sample['title']}")
print(f"\nCombined Text:\n{sample['text']}")

# Use the text field for language model training
for sample in dataset['train']:
    text = sample['text']
    # Train your model with this text
    print(text[:100])
    break

Use Cases

This dataset is suitable for:

• Kurdish language model pretraining
• Kurdish text generation
• Kurdish question answering
• Kurdish book understanding
• Kurdish NLP research
• Document retrieval and search
• Kurdish literature analysis

Statistics

• Total Samples: 18,565
• Languages: Kurdish (Central Kurdish, Kurmanji)
• Script: Arabic script
• Format: Structured book data with combined text field
• Columns: 6 (author, book, title, url, content, text)

Data Fields

• author: Name of the book's author
• book: Book name or identifier
• title: Title of the text/chapter/section
• url: Source URL or reference (if available)
• content: Main text content (preprocessed)
• text: Combined field with author, book, title, and content in Kurdish format (preprocessed)

Quality

All text fields have been preprocessed to ensure:

• Consistent Kurdish character encoding
• Standardized punctuation
• Clean text suitable for NLP applications
• Removed formatting artifacts

Text Field Advantages

The combined text field provides:

• Ready-to-use format for language model training
• Contextual information (author and book) included with content
• Natural Kurdish language structure
• Easy to use without additional formatting

Source

This dataset was created from an Excel file containing Kurdish book data. Text fields were preprocessed using asosoft for improved quality.

Citation

If you use this dataset, please acknowledge:

• The asosoft preprocessing library: asosoft
• The original book sources

License

Apache 2.0 (or specify your license)

Acknowledgments

• Preprocessing library: asosoft by Kurdish NLP
• Original book authors and sources
• Kurdish NLP community

Notes

• The text field combines all information in a natural Kurdish format
• All text has been preprocessed for consistency
• Suitable for various Kurdish NLP tasks
• Ready for language model training without additional processing

Contact

For questions or issues, please open an issue on the dataset repository.