Rapid identification of rumors based on BERT and CNN

Rumor detection is a process designed to identify and analyze false or misleading content in information. The rapid development of social media is an important channel for people to obtain external information, and also a platform for the widespread dissemination of false news. In view of this phenomenon, this paper analyzes the key elements such as time, place, people and events in the news text information by algorithm, and uses the algorithm to analyze the social media, multiple information such as news sources, historical data, and comprehensively analyze the authenticity of the news between the two. This paper mainly introduces the steps and methods of Bidirectional Encoder Representations from Transformers model and Bidirectional Encoder Representations from Transformers-Convolutional Neural Network model in capturing text information, and tests and optimizes the algorithm. The test results show that these two algorithms can efficiently and quickly identify most of the fake news, and intuitively show them to users through the system output, which is worth applying and popularizing in the corresponding field.