Detecting Malicious Insiders In Social Networks And Mitigating Cyber Threats With Advanced Security Approaches

When a fraudster uses a false identity to appear as a reliable source in an OSN, this kind of assault is known as an insider attack. The identification of users' actions inside the network is now a popular use of Machine Learning (ML) techniques. This study introduces Social Network Malicious Insider Detection (SID), a hybrid method that combines LSTM and TBTE. By keeping tabs on user data, the proposed SID hopes to spot data anomalies in OSN interactions. To accurately forecast user behaviour and detect the abnormality pattern in OSN, the suggested SID use LSTM, a more sophisticated kind of Recurrent Neural Network (RNN). A time-based trust assessment approach is used with the LSTM to distinguish between new users, broken nodes, and malicious nodes in OSNs, as well as to accurately classify anomalous nodes. In addition to effectively detecting insider assaults, the proposed SID also decreases wrong apprehensions by offering a new measurable investigation for calculating the corresponding factor over time. This analysis prevents the mistaken characterization of legitimate user behaviours as abnormalities. In comparison to the preexisting RNN, LSTM, and Bilateral LSTM, the suggested SID achieves a greater detection accuracy, according to the evaluation results.