metadata

task_categories:
  - translation
language:
  - ar
  - en
size_categories:
  - 10K<n<100K
license: cc-by-4.0

Introduction

This dataset represents a curated collection of parallel Arabic-English texts, featuring the translations of 24 historically and culturally significant books. These texts provide a portal to the intellectual and literary heritage of the Arabic-speaking world during its classical period.

Content Details

Contained within this dataset are English translations of the following texts, sourced from the Rasaif website:

A Muslim Manual of War
Al-Hanin Ila'l-Awtan
Avarice and the Avaricious
Contemplation
Diseases of the Hearts and Their Cures
Hayy ibn Yaqzan
Ibn Khallikan's Biographical Dictionary
Kitab al-I'tibar
Knowledge Mandates Action
Morals and Behaviour
Nahj al-Balagha
The Book of Strangers
The Canon Of Medicine of Avicenna
The Epistle on Legal Theory
The Heavenly Dispute
The Islamic Conquest of Syria
The Journey of the Strangers
The Key to Medicine and a Guide for Students
The Muqaddimah: An Introduction to History
The Optics of Ibn Al-Haytham
The Rare and Excellent History of Saladin
The Ring of the Dove
The Strangers
The Travels Of Ibn Battuta, 1325 – 1354

Purpose and Application

The overarching objective of this dataset is to highlight the superior literary quality of Classical Arabic, which stands in stark contrast to the language's later developments, particularly due to the mass translations of European texts in the 19th and 20th centuries. It aims to:

Refine Machine Translation (MT): With its intricate grammatical structure and rich lexicon, Classical Arabic presents an ideal challenge for MT systems, which, when honed on such high-caliber content, can achieve greater accuracy and fluency.
Language Models: By incorporating texts of such linguistic finesse, this dataset becomes a cornerstone for developing Large Language Models (LLMs) that can grasp and replicate the sophistication inherent in Classical Arabic.
Preserve Linguistic Heritage: This dataset acts as a conduit for preserving the exceptional literary form of Classical Arabic, providing a benchmark of quality against which contemporary writings can be measured.

Suggested Research Application: Iterative Translation Refinement

A novel application for this dataset involves utilizing existing translation models to back-translate the English texts into Arabic, likely resulting in a less sophisticated form of the language. This process, known as back-translation, can generate a large corpus of imperfect Arabic text. Subsequently, a new model could be trained to refine this weaker form of Arabic by comparing it to the original Classical Arabic texts in the dataset. The resultant model can be used to enhance current Arabic texts by making it sound more "Classical".

Credits

The Rasaif Website: For updates and more information about their work, follow them on Twitter, and follow Ahmad Alghamdi's Telegram channel