Fake Profile Detection Using Logistic Regression and Gradient Descent Algorithm on Online Social Networks

One of the most challenging issues on online social networks is identifying spam accounts. The concern stems from the fact that these personas pose a significant threat, as they may engage in harmful activities against other users, extending beyond mere annoyance or low-quality advertisements. The demand for accurate and effective spam detection algorithms for online social networks is increasing due to this risk. To address the problem of spam detection in online social networks, this research proposes a hybrid machine learning model based on logistic regression and a contemporary metaheuristic method called the Gradient Descent Algorithm. The proposed approach automates spammer identification and provides insights into the factors that have the greatest impact on the detection process. Additionally, the model is evaluated and implemented on multiple datasets, and the experiments and findings demonstrate that the proposed model outperforms many other algorithms in terms of accuracy and delivers robust results in terms of precision, recall, f-measure, and AUC. It also aids in identifying the factors that influence detection the most.