A Turkish aspect-based sentiment dataset with 16,031 real user-review sentences and 24,439 aspect annotations from e-commerce, supplements, and movie domains.
Each row contains pool_id, domain, text, aspects, and flags. Each aspect has a verbatim span, a normalized aspect, a polarity of positive, negative, or neutral… See the full description on the dataset page:
https://huggingface.co/datasets/ytu-ce-cosmos/absa-tr.