EUvsDisinfo
收藏资源简介:
EUvsDisinfo是由谢菲尔德大学创建的多语言数据集,专注于检测新闻文章中的亲克里姆林宫虚假信息。该数据集包含18,249篇文章,覆盖42种语言和508个主题,时间跨度为8.5年。数据集内容来源于专家撰写的反驳文章,旨在揭示不同语言中针对特定虚假信息主题的模式。创建过程中,利用了EUvsDisinfo项目的专家调查结果,确保数据的高质量和多样性。该数据集的应用领域主要集中在训练模型以区分多语言环境下的虚假信息和可信内容,特别是在政治和军事背景下,如乌克兰冲突期间的信息战。
EUvsDisinfo is a multilingual dataset developed by the University of Sheffield, dedicated to detecting pro-Kremlin disinformation in news articles. This dataset contains 18,249 articles, covering 42 languages and 508 topics, spanning a period of 8.5 years. The content of the dataset is sourced from expert-authored rebuttal articles, aiming to uncover patterns of specific disinformation topics across different languages. During its curation, expert survey results from the EUvsDisinfo project were utilized to ensure the high quality and diversity of the dataset. The main applications of this dataset focus on training models to distinguish between disinformation and credible content in multilingual environments, particularly in political and military contexts such as information warfare during the Ukraine conflict.

- 1EUvsDisinfo: a Dataset for Multilingual Detection of Pro-Kremlin Disinformation in News Articles谢菲尔德大学 · 2024年



