k-chirkunov/jeeran_semeval_2016_arabic_adaptation_coarse

Jeeran — SemEval-2016 Task 5 (Arabic) adaptation, coarse categories The k-chirkunov/jeeran_semeval_2016_arabic_adaptation dataset with its 48 fine-grained Arabic aspect labels collapsed onto 14 coarse families. Everything else — targets, character offsets, polarity, the human-annotated spans, the train/test split — is carried over unchanged. Contents split reviews sentences opinions train 43,365 69,524 181,406 test 10,838 17,594 45,225 The… See the full description on the dataset page: https://huggingface.co/datasets/k-chirkunov/jeeran_semeval_2016_arabic_adaptation_coarse.