Evolving Support Vector Machines using WOA for Spam Detection in OSNs

This research paper introduces a novel machine learning approach, SVM-WOA, for the accurate detection of spam profiles on online social networks (OSNs). The model integrates Support Vector Machines (SVM) with the Whale Optimization Algorithm (WOA) to automatically identify spammers and simultaneously determine the most influential features in the detection process. The study applies and tests the SVM-WOA model on multilingual datasets collected from Twitter in Arabic, English, Spanish, and Korean. The experiments demonstrate that the proposed model outperforms other algorithms in terms of accuracy, precision, recall, F-measure, and AUC, while also successfully identifying key features. The paper provides a comprehensive overview of the model's components, including feature selection techniques and the WOA, and details the data collection, feature extraction, and evaluation criteria used in the experiments. The results highlight the effectiveness of the SVM-WOA approach in improving spam detection and understanding the characteristics of spam profiles across different languages.