Dataset Card for moroccan_darija_domain_classifier_dataset

This dataset is designed for text classification in Moroccan Darija, a dialect spoken in Morocco. It has been created synthetically using the Gemini-2.0-Flash model and is intended for fine-tuning the ModernBERT model. The dataset covers a wide range of domains, making it suitable for various NLP applications in Moroccan Darija.

Dataset Details

Dataset Creation

The dataset has been created synthetically using the gemini-2.0-flash model

For dataset creation code see the Github Repository : https://github.com/atlasia-ma/moroccan_darija_dataset_generator

Uses

Direct Use

This dataset is intended for training and evaluating text classification models in Moroccan Darija. It can be used for:

• Domain classification of Moroccan Darija text.

• Fine-tuning ModernBERT or similar transformer-based models.

• Fine-tuning LLM on Moroccan Darija

Out-of-Scope Use

This dataset is not intended for:

• Applications involving hate speech or misinformation.

• Use cases outside Moroccan Darija text classification.

Dataset Structure

The dataset consists of labeled text samples, each categorized into one of the following domains:

• Adult

• Arts_and_Entertainment

• Autos_and_Vehicles

• Beauty_and_Fitness

• Books_and_Literature

• Business_and_Industrial

• Computers_and_Electronics

• Finance

• Food_and_Drink

• Games

• Health

• Hobbies_and_Leisure

• Home_and_Garden

• Internet_and_Telecom

• Jobs_and_Education

• Law_and_Government

• News

• Online_Communities

• People_and_Society

• Pets_and_Animals

• Real_Estate

• Science

• Sensitive_Subjects

• Shopping

• Sports

• Travel_and_Transportation

Each entry in the dataset contains:

• text: A Moroccan Darija text sample.

• label: One of the predefined domain categories.

Annotations

No validation was performed on the dataset, and it needs to be reviewed, as some expressions may not be 100% Moroccan Darija samples.

Dataset Card Authors

ALABOUCH Salah-Eddine