README.md

metadata

language:
  - ary
task_categories:
  - text-generation
multilinguality:
  - monolingual
language_creators:
  - machine-translated
source_datasets:
  - original
size_categories:
  - 1K<n<10K

Dataset Card for DarijaStory

Table of Contents

• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information

Dataset Description

• Repository: https://hf.co/datasets/MBZUAI-Paris/DarijaStory
• Paper: https://arxiv.org/pdf/2409.17912

Dataset Summary

DarijaStory is a story completion dataset. It consists of 4,392 long stories scraped from 9esa, a website featuring a variety of stories written in Moroccan Darija.

Supported Tasks and Leaderboards

• Task Category: Conditional Text Generation
• Task: Story Completion in Moroccan Darija

Languages

The dataset is available in Moroccan Arabic (Darija).

Dataset Structure

Data Instances

Each data instance contains a story or a chapter of story.

Example Data Instance:

{
  'id': 1170,
  'story_name': 'قصة اللؤلؤة السوداء',
  'content': 'حلات عوييناتها بتقااالة... حاسة بحرييق كيقطع فراااسها... قوي وعينيها مضببين ليها رؤيا ... هزات يديها بتقالة حطاتها فوق رااسها....'
}

Data Fields

• id: (integer) Index of the story.
• story_name: (string) The story name.
• content: (integer) The content of the story.

Data Splits

The dataset consists of a single split:

Split	Number of Instances
train	4,392

Dataset Creation

Curation Rationale

The dataset was web-scraped from 9esa.com that contains stories in Darija.

Personal and Sensitive Information

The dataset does not contain personal, private, or sensitive information. All stories are general and cover fictional or societal themes relevant to Morocco.

Considerations for Using the Data

Social Impact of Dataset

This dataset promotes the development and evaluation of language models capable of understanding and generating extended narratives in Moroccan Darija, thus contributing to the advancement of NLP in underrepresented languages and supporting cultural diversity in AI applications.

Discussion of Biases

The dataset consists of Moroccan Darija stories, which may reflect specific cultural and societal themes relevant to Morocco. Users should be aware of this when using the dataset for general language model applications.

Additional Information

Dataset Curators

• MBZUAI-Paris Team

Licensing Information

• License: ODC-BY.

Citation

If you use this dataset in your research, please cite our paper:

@article{shang2024atlaschatadaptinglargelanguage,
      title={Atlas-Chat: Adapting Large Language Models for Low-Resource Moroccan Arabic Dialect}, 
      author={Guokan Shang and Hadi Abdine and Yousef Khoubrane and Amr Mohamed and Yassine Abbahaddou and Sofiane Ennadir and Imane Momayiz and Xuguang Ren and Eric Moulines and Preslav Nakov and Michalis Vazirgiannis and Eric Xing},
      year={2024},
      eprint={2409.17912},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2409.17912}, 
}