Social Spammer Detection Based on PSO-CatBoost.

With the rapid development of social networks, more and more organizations or individuals use social media to communicate with each other, passing on information and getting information, etc. However, while bringing convenience to people, social media has also become the main target of malicious attackers who try to take advantage of the system vulnerability and cause harm to other normal users, they obtain benefits mainly through sending false information, advertising links, phishing, etc. In this paper, firstly, we collect the features of spammers from the four views (profile, behavior, relationship, and interaction) for a more comprehensive analysis of spammers, secondly, we creatively combine the features of Particle Swarm Optimization (PSO) and CatBoost algorithm, and finally, we propose a novel PSO-CatBoost model based on the CatBoost model for detecting spammers. In order to validate the effectiveness of our proposed model, some ensemble learning algorithms are compared, and the experimental results show that our model outperforms other models.