View Dataset
Welcome to ArabicNLP.uk
Home to open-source Arabic NLP datasets, corpora, and tools — created and maintained by Dr Mo El-Haj.
Explore ResourcesKey NLP Resources & Collaborations
Your main access points for datasets, models, and long-term collaborations
Explore the core sources powering Arabic NLP research — models, datasets, repositories, and institutional collaborations.
Featured Arabic NLP Resources
Tarab
a large-scale Arabic creative-text corpus that unifies song lyrics and poetry, containing 2,557,311 verses and 13,509,336 tokens, spanning Classical Arabic, MSA, and six major regional dialect groups.
AraFinNews
212,500 financial news articles from Argaam.com for summarisation, event extraction, and financial NLP.
ArabJobs
8,546 job ads with metadata for gender, salary, profession, and dialect-sensitive labour market analysis.
MCWC
Multilingual constitutions from 191 nations, aligned for legal NLP and MT research.
Kalimat
18,256 cleaned Arabic news articles in a modern ML-ready format.
Habibi
30,000+ Arabic song lyrics across 18 countries for dialect, authorship, and cultural NLP.
EASC
EASC contains 153 documents and 765 human-made summaries. A foundational resource for Arabic extractive summarisation.
MultiLing
Multilingual single-document and multi-document abstractive summarisation in over 10 languages.
Arabic Dialects
A bivalency and code-switching focused corpus covering five major Arabic dialect varieties.
DARES
DARES: Dataset for Arabic Readability Estimation of School Materials - Transformers based evaluation of readability of school materials in Saudi Arabia from grades 1 to 12.
OSMAN
Osman is a readability metric for Arabic that functions as the closest counterpart to the Flesch measure. The repository also includes a parallel Arabic–English dataset from the UN corpus. .
Featured Corpus
Browse Corpora
View Dataset
All Corpora
Loading...