Effective synthetic data generation for fake user detection

Abstract

Nowadays, with the pervasiveness of social networks among the people, the possibility of publishing incorrect information has increased more than before. Therefore, detecting fake news and users who publish this incorrect information is of great importance. This paper has proposed a system based on combining context-user and context-network features with the help of a conditional generative adversarial network for balancing the data set to detect users who publish incorrect information in the Persian language on Twitter. Moreover, by conducting numerous experiments, the proposed system in terms of evaluation metrics compared to its competitors, has produced good performance results in detecting fake users.