MACHINE LEARNING BASED SECURITY MODEL FOR SOCIAL ENGINEERING ATTACK DETECTION

Finally, I am most grateful to the most High God for empowering me to complete this phase of life. Social engineering attacks that exploit human emotions are becoming increasingly common, as cybercriminals manipulate victims online. This study investigates the use of machine learning to automatically detect such attacks by analyzing patterns in textual conversations. The aim is to develop an artificial intelligence-based system that can identify high-risk exchanges by exposing deceitful tactics. The proposed approach involves extracting linguistic and semantic features. These features, which are indicative of fraudulent attempts to build relationships and exploit trust, are then used to train classification algorithms on datasets of real and fake conversations. Evaluation metrics are used to assess the model's performance in accurately categorizing legitimate and malicious interactions. Decision Tree model had an 81% accuracy rate. The model performs well in identifying both legitimate and fraudulent cases, as evidenced by the precision, recall, and F1-scores of about 0.81.Even though the Naïve Bayes model's accuracy was only 72%, its precision, recall, and F1 scores of 0.72 still showed respectable results. However, the results could be further improved by using larger and more diverse datasets, conducting syntactic analysis, and developing customized word vector models. Overall, the intelligent conversation screening system shows promise for uncovering emerging social engineering threats through formal validation. As online data vulnerabilities continue to grow, this research provides an artificially intelligent capability for threat awareness and victim protection by exposing the deception hiding within communication content and context.