Research On Rumor Detection Based On Deep Learnin

The rapid development of social media has not only brought convenience to the society,but also produced many rumors,which has caused great harm to the society.Rumors not only occupy a lot of public resources,but also seriously mislead the public and affect the development of social economy.The research of rumor detection has become the focus of current research.As an important computer technology,deep learning has been widely used in various fields.In this paper,deep learning is applied to the research of rumor detection,which provides strong support for the research of rumor detection and improves the accuracy of rumor detection.The specific work is as follows:An early rumor detection method based on DDR model is proposed.This method first uses the data enhancement method of account filtering and data information standardization to process the data,then uses Bi GRU network to analyze the context,finally uses Q-learning algorithm to detect rumors,and analyzes the accuracy and timeliness of detection.The results show that compared with the typical early rumor detection model(ERD),the accuracy of this method is improved by 0.5% while ensuring the timeliness of rumor detection.A rumor detection method based on hierarchical capsule network is proposed.At present,most of the research on rumor detection mainly starts from the text content,analyzes and detects the text content,and lacks the consideration of the text spatial structure,this paper uses capsule network to analyze the text structure,so as to improve the accuracy of rumor detection.The proposed method is tested in dynamic routing and static routing.The experimental results show that the proposed rumor detection model improves the accuracy of rumor detection by 5.6% and 6.1%respectively compared with the traditional GRU-2 model.A method based on BiGRU＿capsule network rumor detection is proposed.The rumor detection method based on hierarchical capsule network lacks consideration of context,in this paper,Bi GRU network is used to improve the capsule network,so that the model can consider the text spatial structure and analyze the context at the same time.At the same time,the influence and authority of users are calculated based on the portrait features,the input data are processed,and the users with high influence and authority are selected for analysis to improve the accuracy of rumor detection.Experiments show that the model improves the accuracy of static routing and dynamic routing by 0.6% and 0.5% respectively.