NELA-GT-2020
收藏资源简介:
NELA-GT-2020是由伦斯勒理工学院和田纳西大学诺克斯维尔分校合作创建的大型多标签新闻数据集,旨在研究新闻文章中的虚假信息。该数据集包含2020年1月1日至12月31日期间从519个来源收集的近180万篇新闻文章。数据集不仅包括主流新闻来源,还包括替代新闻来源,并附有来自Media Bias/Fact Check(MBFC)的源级真实性标签,覆盖多个真实性维度。此外,2020年的数据集还引入了新闻文章中嵌入的推文,为数据增加了额外的信息层。NELA-GT-2020数据集的应用领域广泛,包括研究新闻生产行为、机器学习模型的训练和测试、以及探索事件驱动的媒体动态和叙事。
NELA-GT-2020 is a large-scale multi-label news dataset jointly developed by Rensselaer Polytechnic Institute and The University of Tennessee, Knoxville, targeting research on disinformation in news articles. This dataset encompasses nearly 1.8 million news articles collected from 519 distinct sources between January 1 and December 31, 2020. It includes both mainstream news outlets and alternative news sources, and is annotated with source-level authenticity labels from Media Bias/Fact Check (MBFC) that span multiple authenticity dimensions. Furthermore, the 2020 iteration of the dataset incorporates embedded tweets within news articles, adding an additional informational layer to the corpus. The NELA-GT-2020 dataset has broad application domains, including research on news production behaviors, the training and testing of machine learning models, as well as the exploration of event-driven media dynamics and narratives.




