# Dataset Card for GOOD-Sum_v2
## Description
GOOD-Sum_v2 is an expanded version of the original [GOOD-Sum]([https://www.goud.ma](https://huggingface.co/datasets/Goud/Goud-sum)) dataset, containing a total of 189,000 articles for training, 9,497 articles for validation, and 9,497 articles for testing. The dataset includes headlines and categories for each article. The articles are written in Moroccan Darija, Modern Standard Arabic (MSA), or a mix of both. Both versions of the dataset were scraped from the [GOUD.ma](https://www.goud.ma) website.

## About the Source
[GOUD.ma](https://www.goud.ma) is a news website established by Ahmed Najim. Journalists contribute articles in Modern Standard Arabic and Moroccan Darija, spanning various topics. This dataset captures the linguistic diversity of Moroccan media.

## Languages:
- Moroccan Darija.
- MSA.
- A mix of both.

## Data Fields:
- Headline: A concise summary of the article.
- Content: The full article text.
- Category: The topic category of the article.


## Intended Use

The dataset is suitable for Natural Language Processing (NLP) tasks, such as:
- Text summarization (extractive or abstractive).
- Pretraining or fine-tuning language models for Arabic dialects.


## Dataset Splits

| **Split** | **Number of Articles** |
|------------------|-------------------------|
| Training Set | 189,000 |
| Validation Set | 9,497 |
| Test Set | 9,497 |

## Citation
If you use this dataset, please cite it as follows:

@inproceedings{aftiss2025empirical,
title={Empirical Evaluation of Pre-trained Language Models for Summarizing Moroccan Darija News Articles},
author={Aftiss, Azzedine and Lamsiyah, Salima and Schommer, Christoph and El Alaoui, Said Ouatik},
booktitle={Proceedings of the 4th Workshop on Arabic Corpus Linguistics (WACL-4)},
pages={77--85},
year={2025}
}