Fighting against Fake News on Newly-Emerging Crisis: A Case Study of COVID-19

Migyeong Yang, Chaewon Park, Jiwon Kang, Daeun Lee, Daejin Choi, Jinyoung Han

Abstract: As social media users can easily access, generate, and spread information regardless of its authenticity, the proliferation of fake news related to public health has become a serious problem. Since these rumors have caused severe social issues, detecting them in the early stage is imminent. Therefore, in this paper, we propose a deep learning model that can debunk fake news on COVID-19, as a case study, at the initial stage of emergence. The evaluation with a newly-collected dataset consisting of both the COVID-19 and Non-COVID-19 fake news claims demonstrates that the proposed model achieves high performance, indicating that the model can identify fake news on COVID-19 in the early stage with a small amount of data. We believe that our methodology and findings can be applied to detect fake news on newly-emerging and critical topics, which should be performed with insufficient resources.