Rumors Detection in the Persian Twitter social network based on deep learning

Due to the rapid exchange of information and the large number of social network users, these networks are focused on gathering the latest information or news from people around the world, so social networks have become the focus of many rumors. These networks can spread information much faster than ever before, misinformed or unverified information spreads just like true information in cyberspace, influencing public opinion and their decisions. Fake news and gossip are the most popular forms of false and unverified information, respectively. They should be detected as soon as possible to avoid significant effects. Therefore, in this study, a model has been presented to detect rumors in social networks as quickly as possible. The proposed model is tested on Twitter network data.

This research has been done in terms of applied type and empirically-analytically. In this research, by discovering and analyzing the importance of two categories of rumor features: structural and content-oriented features, it has identified rumors in the Persian Twitter social network. Has been.

In this study, Persian rumors have been correctly identified using a proposed model based on recursive neural network, which is one of the deep learning techniques. The accuracy of rumor detection using a combination of content and structural features is 91%.