|
|
--- |
|
|
dataset_info: |
|
|
features: |
|
|
- name: idx |
|
|
dtype: int32 |
|
|
- name: concept_name |
|
|
dtype: string |
|
|
- name: question |
|
|
dtype: string |
|
|
- name: answer |
|
|
dtype: string |
|
|
- name: answer_letter |
|
|
dtype: string |
|
|
- name: choices |
|
|
sequence: string |
|
|
- name: question_type |
|
|
dtype: string |
|
|
- name: sub_type |
|
|
dtype: string |
|
|
- name: image1 |
|
|
dtype: image |
|
|
- name: image2 |
|
|
dtype: image |
|
|
- name: image3 |
|
|
dtype: image |
|
|
- name: image4 |
|
|
dtype: image |
|
|
- name: image5 |
|
|
dtype: image |
|
|
- name: image6 |
|
|
dtype: image |
|
|
splits: |
|
|
- name: test |
|
|
num_bytes: 227731749 |
|
|
num_examples: 367 |
|
|
download_size: 115652865 |
|
|
dataset_size: 227731749 |
|
|
configs: |
|
|
- config_name: default |
|
|
data_files: |
|
|
- split: test |
|
|
path: data/test-* |
|
|
license: cc-by-nc-nd-4.0 |
|
|
task_categories: |
|
|
- visual-question-answering |
|
|
- image-to-text |
|
|
language: |
|
|
- ar |
|
|
tags: |
|
|
- Culture |
|
|
- Arabic |
|
|
--- |
|
|
## 🌍 PEARL-X Dataset |
|
|
|
|
|
**Dataset ID:** `UBC-NLP/PEARL-X` |
|
|
|
|
|
**Description:** |
|
|
PEARL-X (PEARL eXtension) is a specialized benchmark designed to evaluate nuanced cultural understanding by focusing on shared cultural concepts that manifest differently across various Arab contexts. It incorporates both single-image and multi-image question-answering tasks, challenging models to compare, contrast, and synthesize information about culturally shared yet visually distinct items (e.g., different types of coffee, traditional attire). |
|
|
|
|
|
**Key Features:** |
|
|
- **Focus:** Nuanced cultural variations of 61 shared concepts |
|
|
- **Tasks:** Single-image and multi-image question answering |
|
|
- **Size:** 367 questions, 347 images |
|
|
- **Purpose:** Assess complex reasoning and comparative understanding of cultural subtleties |
|
|
- **Languages:** Arabic (questions and answers) |
|
|
- **Modality:** Image-Text pairs (single and multiple images per question) |
|
|
|
|
|
|
|
|
|
|
|
[](http://arxiv.org/abs/2505.21979) |
|
|
[](https://huggingface.co/datasets/UBC-NLP/PEARL) |
|
|
[](https://huggingface.co/datasets/UBC-NLP/PEARL-LITE) |
|
|
[](https://huggingface.co/datasets/UBC-NLP/PEARL-X) |
|
|
[](https://github.com/UBC-NLP/pearl) |
|
|
|
|
|
|
|
|
**Paper:** Pearl: A Multimodal Culturally-Aware Arabic Instruction Dataset\ |
|
|
**ArXiv Link:** [http://arxiv.org/abs/2505.21979](http://arxiv.org/abs/2505.21979)\ |
|
|
**Hugging Face Link:** [https://huggingface.co/datasets/UBC-NLP/PEARL-X](https://huggingface.co/datasets/UBC-NLP/PEARL-X)\ |
|
|
|
|
|
**Citation:** |
|
|
```bibtex |
|
|
@misc{alwajih2025pearlmultimodalculturallyawarearabic, |
|
|
title={Pearl: A Multimodal Culturally-Aware Arabic Instruction Dataset}, |
|
|
author={Fakhraddin Alwajih and Samar M. Magdy and Abdellah El Mekki and Omer Nacar and Youssef Nafea and Safaa Taher Abdelfadil and Abdulfattah Mohammed Yahya and Hamzah Luqman and Nada Almarwani and Samah Aloufi and Baraah Qawasmen and Houdaifa Atou and Serry Sibaee and Hamzah A. Alsayadi and Walid Al-Dhabyani and Maged S. Al-shaibani and Aya El Aatar and Nour Qandos and Rahaf Alhamouri and Samar Ahmad and Mohammed Anwar Al-Ghrawi and Aminetou Yacoub and Ruwa AbuHweidi and Vatimetou Mohamed Lemin and Reem Abdel-Salam and Ahlam Bashiti and Aisha Alansari and Ahmed Ashraf and Nora Alturayeif and Alcides Alcoba Inciarte and Adel Ammar and Abdelrahim A. Elmadany and Mohamedou Cheikh Tourad and Ismail Berrada and Mustafa Jarrar and Shady Shehata and Muhammad Abdul-Mageed}, |
|
|
year={2025}, |
|
|
eprint={2505.21979}, |
|
|
archivePrefix={arXiv}, |
|
|
primaryClass={cs.CL}, |
|
|
url={https://arxiv.org/abs/2505.21979}, |
|
|
} |
|
|
``` |
