AraFinNews: Largest Arabic Financial-News Summarisation Dataset Released

AraFinNews, the largest Arabic financial-news summarisation dataset, is released with 212,500 article-headline pairs. Available on GitHub and HuggingFace.

AraFinNews, the largest Arabic financial-news summarisation dataset to date, has been released. It includes 212,500 article–headline pairs from reputable financial media between 2015 and 2025.

The dataset is available on GitHub and HuggingFace. The associated research paper can be found on arXiv.

• Clean structured text and article-level metadata
• Ready-to-use splits for summarisation and downstream financial NLP tasks
• Evaluation of mT5, AraT5, and FinAraT5 models demonstrating gains from financial-domain pretraining

For more NLP datasets and tools, visit ArabicNLP.uk and explore the work of Dr Mo El-Haj on https://elhaj.uk and https://vinnlp.com.

Tags: Arabic NLP, Financial News Summarisation, AraFinNews Dataset, NLP Research, Natural Language Processing, AI/ML Dataset