ChatGPT-FC

ChatGPT-FC数据集是由加州大学戴维斯分校创建，包含22337条新闻声明，涵盖2007至2023年的政治和社会问题。该数据集通过结合人类记者和大型语言模型ChatGPT的增强事实核查，旨在减少人类记者的主观偏见并提高假新闻检测的准确性。数据集的创建过程涉及从PolitiFact网站收集新闻声明和人类记者的事实核查报告，并设计特定提示以引导ChatGPT生成相应的事实核查响应。ChatGPT-FC数据集的应用领域主要集中在新闻真实性验证，旨在解决假新闻和信息误导问题。The ChatGPT-FC dataset was created by the University of California, Davis, and contains 22,337 news statements covering political and social issues from 2007 to 2023. This dataset adopts enhanced fact-checking that combines human journalists and ChatGPT, a large language model, to reduce subjective bias among human reporters and improve the accuracy of fake news detection. The dataset creation process involves collecting news statements and human fact-checking reports from the PolitiFact website, as well as designing specific prompts to guide ChatGPT in generating corresponding fact-checking responses. The main application areas of the ChatGPT-FC dataset are centered on news authenticity verification, with the goal of addressing issues related to fake news and information misinformation.