Rumor Detection Technology Based On Graph Knowledge Embedding

With the rapid development of the Internet,more and more netizens posted their opinions on social platforms,which has promoted diversified communication in the world.But the following problem is that numerous rumors spread on social platforms.Rumors involving public safety have caused great harm to social stability.Previous rumor detection research focused on text features,user information,and propagation structure information.They are still essentially detecting the falseness of the text by analyzing the user's posting intentions,which ignores the importance of external knowledge.Besides,without abundant rumor text for training and analyzing,the performance of rumor detection is degraded.Therefore,the external knowledge embedding and the data augment of rumor are of great significance for the automatic rumor detection.This paper uses the graph knowledge embedding based method to detect rumors and generate rumor text.The main contents and innovations are as follows:1.This paper proposes a knowledge representation embedding based rumor detection method,which takes knowledge representation into account in rumor detection.Experimental results show that this method can improve the accuracy of rumor detection and explain the detection results to a certain extent.2.Aiming to augment rumor data from the view of knowledge based text generation,this paper proposes a graph based method of rumor data augment,making up for the shortcomings of insufficient data in current rumor detection tasks.In this paper,the graph based text generation model combined with generative adversarial networks is used to model the rumor text generation.The experimental results show the model trained by generative adversarial networks can simulate the process of rumor generation,and consequently improve the performance of rumor detection through data augment.