A COVID-19 Rumor Dataset

Cheng, Mingxi; Wang, Songli; Yan, Xiaofeng; Yang, Tianqi; Wang, Wenshuo; Huang, Zehao; Xiao, Xiongye; Nazarian, Shahin; Bogdan, Paul

doi:10.3389/fpsyg.2021.644801

Cited by 62 publications

(41 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The categories and examples of debunking are iteratively refined through pilot tests. The examples of Others as Support, Comment, and Queries come from the previous study on rumor detection (Cheng, Nazarian, and Bogdan 2020;Cheng et al 2021). As a result, we get the labels with the Fleiss Kappa score at 0.542 for replies and 0.541 for QTs, which are moderate agreement (Landis and Koch 1977).…”

Section: Annotating Debunking Tweetsmentioning

confidence: 87%

"This is Fake News": Characterizing the Spontaneous Debunking from Twitter Users to COVID-19 False Information

Miyazaki¹,

Uchiba²,

Tanaka³

et al. 2022

Preprint

View full text Add to dashboard Cite

False information spreads on social media, and fact-checking is a potential countermeasure. However, there is a severe shortage of resources of fact-checkers, and an efficient way to scale fact-checking is desperately needed, especially in pandemics like COVID-19. In this study, we focus on spontaneous debunking by social media users, which has been missed in existing research despite its indicated usefulness for fact-checking and countering false information. Specifically, we characterize the fake tweets that tend to be debunked and Twitter users who often debunk fake tweets. For this analysis, we create a comprehensive dataset of responses to tweets with false information, annotate a subset of them, and build a classification model for detecting debunking behaviors. We find that most fake tweets are left undebunked and spontaneous debunking is slower than other forms of responses, and exhibits partisanship in political topics. These results provide actionable insights into utilizing spontaneous debunking to scale conventional fact-checking, thereby supplementing existing research from a new perspective.

show abstract

Section: Annotating Debunking Tweetsmentioning

confidence: 87%

"This is Fake News": Characterizing the Spontaneous Debunking from Twitter Users to COVID-19 False Information

Miyazaki¹,

Uchiba²,

Tanaka³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Training Testing Type k_title [34] 31k 9K Article Titles coaid [9] 5K 1K News summary c19_text [1] 2.5K 0.5K Articles cq [32] 12.5K 2K Tweets miscov [27] 4K 0.6K Headlines k_text [34] 31k 9K Articles rumor [8] 4.5K 1K Social Posts cov_fn [10] 4K 2K Tweets c19_title [1] 2.5K 0.5K Article Titles…”

Section: Datasetmentioning

confidence: 99%

Evaluating Generalizability of Fine-Tuned Models for Fake News Detection

Suprem¹,

Pu²

2022

Preprint

View full text Add to dashboard Cite

The Covid-19 pandemic has caused a dramatic and parallel rise in dangerous misinformation, denoted an 'infodemic' by the CDC and WHO. Misinformation tied to the Covid-19 infodemic changes continuously; this can lead to performance degradation of fine-tuned models due to concept drift. Degredation can be mitigated if models generalize well-enough to capture some cyclical aspects of drifted data. In this paper, we explore generalizability of pre-trained and fine-tuned fake news detectors across 9 fake news datasets. We show that existing models often overfit on their training dataset and have poor performance on unseen data. However, on some subsets of unseen data that overlap with training data, models have higher accuracy. Based on this observation, we also present KMeans-Proxy, a fast and effective method based on K-Means clustering for quickly identifying these overlapping subsets of unseen data. KMeans-Proxy improves generalizability on unseen fake news datasets by 0.1-0.2 f1-points across datasets. We present both our generalizability experiments as well as KMeans-Proxy to further research in tackling the fake news problem.

show abstract

“…Some approaches managed to create misinformation datasets by a combination of data from different sources (Patwa et al, 2020;Yang et al, 2020;Haouari et al, 2020). (Cheng et al, 2021) created a set of annotated tweets specifically containing COVID-19 misinformation. Furthermore, some misinformation datasets cover languages like Chinese (Yang et al, 2021) and Arabic (Haouari et al, 2020).…”

Section: Related Workmentioning

confidence: 99%

Multimodal Pipeline for Collection of Misinformation Data from Telegram

Sosa¹,

Sharoff²

2022

Preprint

View full text Add to dashboard Cite

The paper presents the outcomes of AI-COVID19, our project aimed at better understanding of misinformation flow about COVID-19 across social media platforms. The specific focus of the study reported in this paper is on collecting data from Telegram groups which are active in promotion of COVID-related misinformation. Our corpus collected so far contains around 28 million words, from almost one million messages. Given that a substantial portion of misinformation flow in social media is spread via multimodal means, such as images and video, we have also developed a mechanism for utilising such channels via producing automatic transcripts for videos and automatic classification for images into such categories as memes, screenshots of posts and other kinds of images. The accuracy of the image classification pipeline is around 87%.

show abstract

A COVID-19 Rumor Dataset

Cited by 62 publications

References 18 publications

"This is Fake News": Characterizing the Spontaneous Debunking from Twitter Users to COVID-19 False Information

"This is Fake News": Characterizing the Spontaneous Debunking from Twitter Users to COVID-19 False Information

Evaluating Generalizability of Fine-Tuned Models for Fake News Detection

Multimodal Pipeline for Collection of Misinformation Data from Telegram

Contact Info

Product

Resources

About