FakeFinder: Twitter Fake News Detection on Mobile

FakeFinder: Twitter Fake News Detection on Mobile

연구 분야: Safety

학회: WWW '20: Companion Proceedings of the Web Conference 2020

Misinformation, or fake news, spreads quickly on the social media platform Twitter. Mobile devices are widely used to read Twitter posts. A mobile app that can detect fake news from the live Twitter stream and alert users in real time is an effective way to contain the spread of misinformation on Twitter. Towards this objective, the prediction model needs to be small to achieve fast prediction. In this paper, we design and develop a fake news detection mobile app with a device-based prediction model based on the small language model ALBERT. Experiments show that it can achieve real-time, accurate detection of fake news.