DarjaCore/algerian-darja-sample

Algerian Darja Sample

A growing Algerian Darja text corpus collected for NLP and language-modeling research.

This dataset is updated incrementally as new sources are collected and processed. Exact sample counts, file sizes, and statistics change between releases — refer to the Dataset Viewer on the repository page for current figures rather than any numbers in this card.

Dataset at a Glance

Sample count, file size, character/word counts, and other metrics are not fixed and grow over time as new batches are added. Check the repository's file listing and Dataset Viewer for up-to-date figures.

Data Sources

The corpus is assembled from multiple sources over time, including:

• Algerian online forums
• Telegram channels and groups
• Academic and community text sources

Each new batch is deduplicated against the existing corpus before being added.

Data Format

Single-column text dataset:

text

Example:

واش راك اليوم؟

Each row represents one independent text sample.

Data Quality

• Samples are deduplicated within the corpus.
• Empty or whitespace-only samples are removed.
• Basic text cleaning is applied (mention stripping, repeated-character normalization) but the corpus has not been normalized into a single standardized Algerian Darja orthography.

Intended Use

• Algerian Darja NLP research
• Language-model pretraining experiments
• Tokenizer evaluation
• Darja language analysis
• Text generation research
• Linguistic statistics

Limitations

This is a sample of an ongoing collection effort, not a finished or complete corpus.

The data may contain:

• spelling variations
• informal language
• Arabizi / Latin-script text
• emojis
• noisy user-generated text
• potentially sensitive or context-dependent content

Statistics such as token counts depend on the tokenizer used and are not fixed values of the dataset itself.

Preprocessing

• Empty-row detection and removal
• Duplicate detection and removal (within and across batches)
• Basic text cleaning

Citation

@dataset{darjacore_algerian_darja_sample,
  author = {DarjaCore},
  title = {Algerian Darja Sample},
  year = {2026},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/DarjaCore/algerian-darja-sample}
}

License

MIT

See the repository license for the applicable terms.

Organization

DarjaCore

Open-source research focused on Algerian Darja, NLP, and language technologies.