AhmadHakami/saudipedia-arabic-qa

Saudipedia Q&A Dataset

Dataset Description

Summary

This dataset contains question-answer pairs scraped from Saudipedia, a comprehensive Arabic encyclopedia focused on Saudi Arabia. The dataset includes 1,082 Q&A entries covering various topics related to Saudi culture, history, economy, government, society, geography, religion, and notable personalities.

The data was collected by scraping the website's question-answer section, which provides detailed answers to common questions about Saudi Arabia in Arabic. Each entry includes the question, the detailed answer, the section/category, and the original page link.

Supported Tasks and Leaderboards

This dataset can be used for:

• Question Answering: Training and evaluating Arabic QA models
• Information Retrieval: Building search systems for Saudi-related information
• Text Classification: Categorizing content by topic sections
• Natural Language Understanding: Understanding Arabic text in cultural context
• Chatbot Development: Creating conversational AI for Saudi topics

Languages

The dataset is primarily in Arabic (العربية), with some potential English terms in technical contexts.

Dataset Structure

Data Instances

Each data instance is a dictionary with the following fields:

• section (string): The category/topic of the question (e.g., "حكومة وسياسة", "اقتصاد وأعمال", "تاريخ")
• question (string): The question text in Arabic
• answer (string): The detailed answer text in Arabic
• page_link (string): URL to the original Saudipedia page

Data Fields

Field	Type	Description
section	string	Topic category in Arabic
question	string	Question text
answer	string	Detailed answer text
page_link	string	Source URL

Data Statistics

• Total entries: 1,082
• Sections and their counts:
  • حكومة وسياسة: 308
  • اقتصاد وأعمال: 190
  • مجتمع: 159
  • تاريخ: 131
  • جغرافيا: 96
  • ثقافة: 92
  • شؤون دينية: 88
  • الأسرة المالكة: 9
  • شخصيات: 9

Data Splits

The dataset is provided as a single CSV file without predefined train/validation/test splits. Users can split the data according to their needs.

Dataset Creation

Curation Rationale

This dataset was created to provide a comprehensive collection of factual information about Saudi Arabia in Arabic, sourced from a reputable encyclopedia. It serves as a valuable resource for Arabic NLP research and applications focused on Saudi culture and knowledge.

Source Data

• Initial Data Collection and Normalization: Data was scraped from Saudipedia's question-answer section using Python scripts with BeautifulSoup and requests libraries.
• Who are the source language producers?: Saudipedia content creators and editors.

Annotations

The dataset includes automatic categorization by topic sections as defined on the source website. No additional manual annotations were performed.

Personal and Sensitive Information

The dataset contains factual information about Saudi Arabia and does not include personal or sensitive information about individuals.