A multi-swarm optimisation approach for spam detection in online social networks

Abstract

Online Social Networks (OSNs) play a crucial role in communication systems for rapid message broadcasting and information sharing. Facebook is one of the popular OSNs that has the highest number of active users as per the statistical reports. However, it witnesses challenges due to the presence of spammers, whose actions may make the environment unfavourable for users. The spammers hold social accounts that are particularly created for personal benefit. As the number of users on Facebook increases, the number of illegitimate accounts proportionally rises, which leads to distress in the OSN environment. Several methods have been proposed in the literature to address the spam profile detection problem; however, they become obsolete as the spammers evolve. Hence, in this paper, a novel Multi-Swarm-Whale Optimisation Algorithm (MS-WOA) is proposed for feature selection to detect spam profiles on Facebook. Further IP-address-based features tailored for Facebook are also proposed. The performance of the proposed MS-WOA is compared with the recently developed methods and outperforms them in terms of accuracy and robustness.