This dataset is enhanced version of existing offensive language studies. Existing studies are highly imbalanced, and solving this problem is too costly. To solve this, we proposed contextual data mining method for dataset augmentation. Our method is basically prevent us from retrieving random tweets and label individually. We can directly access almost exact hate related tweets and label them directly without any further human interaction in order to solve imbalanced… See the full description on the dataset page:
https://huggingface.co/datasets/Toygar/turkish-offensive-language-detection.
