A feature selection approach for spam detection in social networks using gravitational force-based heuristic algorithm

Nowadays, technologies cover all human life areas and expand communication platforms with suitable and low-cost space. Advertising and profiteering organizations use this large space of audience and low-cost platform to send their desired information and goals in the form of spam. In addition to creating problems for users, it causes time and bandwidth consumption. They will also be a threat to the productivity, reliability, and security of the network. Various approaches have been proposed to combat spam. The most dynamic and best methods of spam filtering are machine learning and deep learning, which perform high-speed filtering and classification of spam. In this paper, we present a new way to discover spam on various social networks by scaling up a Support Vector Machine (SVM) based on a combination of the Genetic Algorithm (GA) and Gravitational Emulation Local Search Algorithm (GELS) to select the most effective features of spam. The experiments' results show that the accuracy of the proposed method will be more optimal compared to other algorithms, and the algorithm has been able to compete with the compared algorithms.