Research on false news detection based on BERT model and attention mechanism

Artificial intelligence and intelligent algorithms have made remarkable progress in information processing. However, driven by the rapid development of the Internet, fake news interferes with people's access to accurate information. Aiming at the problem of false news detection, a multimodal false news detection model is proposed on the basis of the attention mechanism optimization bidirectional converter model. The designed model has an average accuracy of 98.61% in identifying news and an average response time of 1.74 seconds. It has a high recognition accuracy on various types of news data, especially on plain text news, with an accuracy of 99.82% and a response time of 1.15s. This shows that the new model can accurately and quickly determine the authenticity of news. This study provides an effective fake news identification tool that helps reduce the false information spread and improve the authenticity of network information.