CHEAT
收藏资源简介:
CHEAT数据集是由暨南大学研究团队创建的大规模数据集,专门用于检测由ChatGPT编写的摘要。该数据集包含35,304条合成摘要,主要分为生成、润色和混合三种类型。数据集的创建过程涉及使用ChatGPT从关键词生成摘要,以及对人类编写的摘要进行润色和混合处理。CHEAT数据集的应用领域主要集中在学术界,旨在解决ChatGPT生成内容对学术严谨性和原创性的潜在威胁,帮助开发有效的检测算法。
The CHEAT dataset is a large-scale dataset created by the research team from Jinan University, specifically designed for detecting summaries written by ChatGPT. This dataset contains 35,304 synthetic summaries, which are mainly categorized into three types: generated, polished, and mixed. The construction process of the dataset involves using ChatGPT to generate summaries from keywords, as well as polishing and mixing human-authored summaries. The CHEAT dataset is primarily applied in academic circles, aiming to address the potential threats posed by ChatGPT-generated content to academic rigor and originality, and assist in the development of effective detection algorithms.

- 1CHEAT: A Large-scale Dataset for Detecting ChatGPT-writtEn AbsTracts暨南大学 · 2024年



