Below are the label descriptions for each dataset in the same order as the label categories in the datasets. For instance, if an observation in SCv1 has label 1, then that observation is in the sarcasm category. Note that the Olympic and SE0714 datasets can have multiple labels associated with each observation.
- Olympic: negative & high control, positive & high control, negative & low control, positive & high control
- PsychExp: joy, fear, anger, sadness, disgust, shame, guilt
- SCv1: not sarcastic, sarcastic
- SCv2-GEN: not sarcastic, sarcastic
- SE0714: fear, joy, sadness
- SS-Twitter: negative, positive
- SS-Youtube: negative, positive
- kaggle-insults: neutral, insult
These benchmark datasets are uploaded to this repository for convenience purposes only. They were not released by us and we do not claim any rights on them. Use the datasets at your responsibility and make sure you fulfill the licenses that they were released with. If you use any of the benchmark datasets please consider citing the original authors. References are in our paper/blog post.