Dear all, I’m pleased to announce the release of AraFinNews, the largest Arabic financial-news summarisation dataset to date. GitHub: https://github.com/ArabicNLP-UK/AraFinNews HuggingFace: https://huggingface.co/datasets/drelhaj/AraFinNews Paper: https://arxiv.org/abs/2511.01265 AraFinNews includes 212,500 article–headline pairs (2015–2025) from reputable financial media, making it a strong Arabic counterpart to benchmarks like CNN/DailyMail. It provides clean structured text and article-level metadata, with ready-to-use splits for summarisation and for building downstream financial NLP tasks. Using this dataset, we evaluated mT5, AraT5, and the domain-adapted FinAraT5, with results demonstrating clear gains from financial-domain pretraining in headline generation. Best wishes, Mo
See more of our NLP datasets and tools on: ArabicNLP.uk https://arabicnlp.uk/ HuggingFace: https://huggingface.co/drelhaj GitHub: https://github.com/drelhaj ArabicNLP-UK: https://github.com/orgs/ArabicNLP-UK/ UCREL: https://github.com/UCREL ———— Dr Mo El-Haj Director of NLP @ VinUniversity Reader (Associate Professor) in NLP CECS, VinUniversity, Vietnam SCC, Lancaster University, UK https://elhaj.uk<https://elhaj.uk/> https://vinnlp.com<https://vinnlp.com/>
_______________________________________________ Corpora mailing list -- [email protected] https://list.elra.info/mailman3/postorius/lists/corpora.list.elra.info/ To unsubscribe send an email to [email protected]
