Social media posts may go viral and reach large numbers of people within a short period of time. Such posts may threaten the public dialogue if they contain misleading content such as fake news. As such, early detection of viral posts may be crucial for tasks such as fact-checking. Previous works proposed their own metrics to measure virality. However, such metrics may not accurately represent viral tweets or may introduce too many false positives. In this work, we use the ground truth data provided by Twitter's "Viral Tweets" topic. We propose a dataset of tweets that are labeled by Twitter as viral and a dataset of all tweets from users who authored a viral tweet. We review the proposed metrics to represent the viral tweets and propose our own metric. We also propose a transformers-based model to predict viral tweets. The code and the tweet ids are publicly available at: https://github.com/tugrulz/ViralTweets

翻译：社交媒体帖子可能在短时间内迅速传播并触达大量用户。若此类帖子包含虚假新闻等误导性内容，则可能威胁公共对话。因此，对病毒式帖子的早期检测对于事实核查等任务至关重要。已有研究提出了各自的病毒式传播度量指标，但这些指标或无法准确反映病毒式推文的特征，或可能引入过多误报。本研究利用推特“病毒式推文”话题提供的真实标注数据，构建了两个数据集：一个由推特官方标注为病毒式传播的推文数据集，另一个涵盖所有曾发布病毒式推文的用户发布的全部推文。我们系统梳理了现有病毒式传播度量指标，并提出新的度量标准；同时基于Transformer架构提出病毒式推文预测模型。相关代码及推文ID已开源：https://github.com/tugrulz/ViralTweets