ogozcelik/turkish-fake-news-detection

MiDe22数据集包含5,064条土耳其语推文，涵盖了2020年至2022年间的多个事件，如俄罗斯-乌克兰战争、COVID-19大流行和难民问题。每条推文都有其错误信息标签，并包括用户的互动数据（如点赞、回复、转发和引用）。数据集中的推文被标注为三类：真实信息、错误信息和其他。标注过程由五位标注者完成，每个推文至少由两位标注者标注，并通过Krippendorf’s alpha可靠性系数衡量标注者间的一致性，结果为0.791。

MiDe22 dataset contains 5,064 Turkish tweets covering several events between 2020 and 2022, including the Russia-Ukraine war, COVID-19 pandemic, and Refugees. Each tweet has its misinformation label and includes user engagement data such as likes, replies, retweets, and quotes. The tweets in the dataset are labeled into three categories: True, False, and Other. The annotation process was carried out by five annotators, with each tweet annotated by at least two annotators, and the interannotator agreement was measured using Krippendorf’s alpha reliability coefficient, resulting in 0.791.

原始信息汇总

MiDe22: An Annotated Multi-Event Tweet Dataset for Misinformation Detection

概述

• 数据集名称: MiDe22
• 任务类别: 文本分类
• 语言: 土耳其语
• 数据量: 1K<n<10K
• 样本数量: 5,064条土耳其语推文
• 事件范围: 2020年至2022年，包括俄罗斯-乌克兰战争、COVID-19疫情和难民问题
• 用户互动: 包含推文的点赞、回复、转发和引用数据

数据字段

• tweet: 字符串类型的推文内容
• label: 分类标签，可能值包括 True, False, Other

数据规模

类别	True	False	Other
推文	669	1,732	2,663

标注

• True: 推文包含关于对应事件的正确信息
• False: 推文包含关于对应事件的错误信息
• Other: 推文无法归类为错误或正确信息

标注过程

• 标注者数量: 5人
• 标注方式: 每条推文至少由两名标注者标注
• 一致性测量: 使用Krippendorf’s alpha计算标注者间一致性，结果为0.791

数据来源

• GitHub仓库: MiDe22 GitHub
• 论文: LREC-COLING 2024

引用

@inproceedings{toraman-etal-2024-mide22-annotated, title = "{M}i{D}e22: An Annotated Multi-Event Tweet Dataset for Misinformation Detection", author = "Toraman, Cagri and Ozcelik, Oguzhan and Sahinuc, Furkan and Can, Fazli", booktitle = "Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.lrec-main.986", pages = "11283--11295", }

联系信息

• 邮箱: ogozcelik[at]gmail[dot]com