An Efficient Mechanism of Modified Sand Cat Swarm Optimization and DCNN For Detection of Spam In Online Social Media Network

Abstract In recent years, social media has become an indispensable part of people's daily lives. The use of social media networks, particularly WhatsApp, Facebook, and email, has significantly surged worldwide due to their widespread popularity. These platforms serve as potent tools for users to share information and opinions, post photos and videos, and disseminate news. The growing popularity and importance of social media platforms attract spammers to a greater extent. Detecting social network spam is particularly challenging due to its brief content and constantly streaming nature. Many spam detection methods for social media rely on machine learning, utilizing a variety of classification and clustering algorithms. Deep learning techniques have achieved significant success globally in solving diverse problems across different domains. This research aims to develop a robust spam detection model for social media networks using deep CNN methods and to evaluate the effectiveness of the models using Modified Sand Cat Swarm (MCSO) based optimization models. N-gram and Bag of Words features are utilized for ensemble feature extraction and selection to rank and analyze the significance of features in spam detection. The performance of the proposed models is compared with other deep learning algorithms. A comparative analysis of spam detection models reveals that the random forest classifier outperforms when trained with randomly chosen data of varying sizes. A DCNN-based MSCSO successfully validates 99.02% of spam on social media networks, as confirmed by the classification. These unconventional pairings reinforce the original premise behind these classification findings.