Multitask rumor detection method based on bidirectional propagation graph

Abstract

The invention discloses a multitask rumor detection method based on a bidirectional propagation graph. The method can be used for rumor detection of social network posts and site detection of comment information. The method comprises the following steps: firstly, generating a text feature matrix, a user feature matrix and a text statistical feature matrix according to the content of a rumor post, then constructing a two-way propagation graph of a rumor, extracting propagation features of the rumor by calculating two-way graph convolution and carrying out root node feature enhancement, and finally, carrying out average pooling and feature integration on the propagation features to obtain a rumor post, training softmax classifier, and obtaining rumor detection and standing field detection results. According to the method, the sequential propagation features and the breadth distribution features of the rumors can be effectively obtained, and the standing detection of the user comments is introduced as an auxiliary task, so that the generalization of the model is improved, and the accuracy of the rumor detection task is further improved.

Description

technical field [0001] The invention relates to a multi-task rumor detection method based on a two-way propagation graph, which can be used for rumor detection of social network posts and position detection of comment information, and belongs to the technical field of Internet and natural language processing. Background technique [0002] In recent years, social networks have achieved rapid development, and have quickly become one of the important ways for people to obtain news information. Due to the huge amount of information on social networks and the speed of transmission far exceeds that of traditional media, a large number of unconfirmed rumors can be spread freely in cyberspace, which has become a growing problem. The Internet has become the "fifth space" after the land, sea, sky, and space, and the use of social networking platforms to spread rumors, take the opportunity to commit cyber crimes, spread violent and terrorist information, and instigate color revolutions...

AI Technical Summary

Problems solved by technology

Most mainstream social network platforms still use artificial rumor detection methods as the mainstream mechanism for identifying suspected rumor information. testing needs

The method based on machine learning boils down the rumor detection work to a binary classification problem. Using supervised learning methods to automatically detect rumor information has high requirements for feature engineering, and it is impossible to obtain the deep features of rumors in the process of spreading and spreading, so it is impossible to obtain higher recognition rate

However, many rumor detection methods based on deep learning start from the time-sequential spread structure of rumors, and detect suspected rumor information by extracting deep features during the spread of rumors. There is still some precision loss in the task

[0004] Aiming at the serious threat posed by current social network rumors to cyberspace security and the problem that existing rumor detection methods cannot effectively extract the characteristics of rumor propagation, the present invention proposes a multi-task rumor detection method based on a two-way propagation graph, using an improved The bidirectional graph convolutional neural network (Bi-GCN) extracts the sequential propagation characteristics and breadth distribution characteristics of rumors, and introduces the position detection of comment texts as an auxiliary task to improve the performance and generalization of rumor detection tasks, thereby improving the accuracy of rumor detection. Accuracy

Images