CHECKED

新型冠状病毒肺炎影响了所有人的生活。为了保持社交距离和避免曝光，工作和生活逐渐转移到网上。在这种趋势下，获取新型冠状病毒肺炎新闻的社交媒体使用量有所增加。此外，新型冠状病毒肺炎的错误信息经常在社交媒体上传播。在这项工作中，我们开发了第一个关于新型冠状病毒肺炎错误信息的中文数据集。CHECKED提供了从2019年12月到2020年8月的总共2,104个与新型冠状病毒肺炎相关的已验证微博，这些微博通过使用特定的关键字列表进行识别。相应地，CHECKED包括1,868,175转发、1,185,702评论和56,852,736点赞，这些评论揭示了这些经过验证的微博是如何在微博上传播和反应的。该数据集包含每个微博的丰富多媒体信息集，包括地面真相标签，文本，视觉，时间和网络信息。在使用CHECKED预测假新闻时，已经进行了广泛的实验来分析已检查的数据并为完善的方法提供基准结果。我们希望检查可以促进针对冠状病毒错误信息的研究。COVID-19 has disrupted the daily lives of people worldwide. To maintain social distancing and minimize exposure risk, work and daily routines have gradually shifted online. Consequently, social media usage for accessing COVID-19 news has surged. Meanwhile, COVID-19 misinformation has spread rampantly across social media platforms. In this study, we developed the first Chinese dataset dedicated to COVID-19 misinformation, named CHECKED. CHECKED contains a total of 2,104 verified COVID-19-related Weibo posts collected from December 2019 to August 2020, which were identified using a predefined keyword list. Correspondingly, the dataset encompasses 1,868,175 reposts, 1,185,702 comments, and 56,852,736 likes associated with these posts, which reveal the propagation dynamics and public responses of these verified content on the Weibo platform. Each Weibo post in CHECKED is paired with rich multimodal information, including ground-truth labels, textual content, visual materials, temporal metadata, and network-related attributes. Extensive experiments have been conducted using CHECKED for fake news detection tasks to analyze the dataset and establish benchmark results for state-of-the-art detection models. We hope that the CHECKED dataset will facilitate research on coronavirus-related misinformation.

搜集汇总

背景与挑战

背景概述

CHECKED是首个针对COVID-19虚假信息的中文数据集，包含2,104条经过验证的微博及相关互动数据，旨在促进冠状病毒错误信息的研究。