During the covid-19 pandemic, a considerable amount of data travels fast worldwide on the net, mainly on the social media platform where people all over the world have constant and easy access to submit materials and posts. A considerable amount of shared news embeds misleading information which affects negatively the cognitive and psychological health of its readers. The present case study focuses on fake news being tweeted during the coronavirus pandemic for the purpose to mislead the targeted population. In this context, this paper exhibits a new approach to detect fake news on Twitter during the Covid-19 period. The proposed method consists of a classification approach that uses new tweets’ features and it is based on natural language processing, machine learning, and deep learning. The method is implemented in parallel with apache spark. Experimental results show that our approach yields very valuable results once it is used with the random forest algorithm with an accuracy equal to 79%. We also demonstrate that the sentiment of tweets plays an important role in the detection of fake news. Indeed, the model we present outperforms those models lacking consideration of new tweets’ features.

在covid-19大流行期间，大量数据通过网络快速传输到世界各地，主要是在社交媒体平台上，世界各地的人们可以轻松便捷地提交材料和帖子。大量的共享新闻会嵌入误导性信息，从而对读者的认知和心理健康产生负面影响。本案例研究的重点是在冠状病毒大流行期间发布假新闻，目的是误导目标人群。在这种情况下，本文展示了一种在Covid-19期间检测Twitter上虚假新闻的新方法。所提出的方法包括使用新推文功能的分类方法，该方法基于自然语言处理，机器学习和深度学习。该方法与apache spark并行执行。实验结果表明，与随机森林算法一起使用时，我们的方法产生了非常有价值的结果，其准确度等于79％。我们还证明了推文的情绪在检测假新闻中起着重要作用。确实，我们提出的模型优于那些没有考虑新推文功能的模型。