Legal case documents: A comprehensive dataset for Arabic natural language processing research and applications.

Area of Science:

• Legal Technology
• Natural Language Processing
• Computational Linguistics

Background:

• Legal data presents unique challenges due to complex language and specialized terminology.
• Advancing Natural Language Processing (NLP) in the legal sector requires high-quality, structured datasets.
• A significant gap exists in Arabic legal NLP research and available datasets.

Purpose of the Study:

• To propose and introduce a novel Arabic legal case dataset.
• To provide a comprehensive resource for Arabic legal NLP research and applications.
• To bridge the gap in Arabic legal data availability for NLP tasks.

Main Methods:

• Collected legal case data from the Board of Grievances website in Saudi Arabia.
• Structured the dataset to include cases, summaries, keywords, and categories.
• Ensured diversity in case length (100-30,000 words) and page count (1-80 pages).

Main Results:

• A dataset comprising 3170 Arabic legal cases across 47 distinct categories.
• The dataset contains detailed information including case text, summaries, keywords, and classifications.
• Demonstrated the dataset's suitability for various NLP applications.

Conclusions:

• The proposed dataset significantly enhances resources for Arabic legal NLP.
• It supports diverse NLP tasks such as text categorization, data extraction, sentiment analysis, and summarization.
• This resource is expected to improve efficiency and accuracy in the legal profession through advanced NLP.