This repository hosts the dataset associated with our paper titled "ChatGPT Role-play Dataset: Analysis of User Motives and Model Naturalness". Our work has been accepted for presentation at the LREC-COLING 2024 conference.
- vanilla.csv: Represents the interactions without specific role-play instructions.
- boss.csv: Interactions where ChatGPT plays the role of a user's boss.
- classmate.csv: Interactions with ChatGPT acting as the user's classmate.
Each turn was coded with user motives of user responses, or the perceived naturalness of ChatGPT responses.
We express our thanks to the anonymous reviewers and the members of the PortNLP research group for their constructive feedback. We are particularly grateful to the participants who helped us create the CRD dataset in this study. This research was supported by the National Science Foundation grants (CRII:RI 2246174 and SAI-P 2228783).
Please cite our work as follows:
@article{tao2024chatgpt,
title={ChatGPT Role-play Dataset: Analysis of User Motives and Model Naturalness},
author={Tao, Yufei and Agrawal, Ameeta and Dombi, Judit and Sydorenko, Tetyana and Lee, Jung In},
journal={arXiv preprint arXiv:2403.18121},
year={2024}
}