Detecting Cybercrime Risks in Employee Behavior Using GraphSAGE and Personality Theories

Cybercrimes pose significant threats to online safety and privacy, targeting individuals, organizations, and societies. These crimes, including child pornography, cyberstalking, identity theft, cyber laundering, cyber terrorism, drug sale, data leakage, sexually explicit content, phishing, and other hacker behaviours, exploit computers and communication tools as instruments of misconduct. This study examines the landscape of cybercrimes and their impact on digital environments, highlighting the techniques employed by cybercriminals. The methodology focuses on robust data processing, addressing issues such as missing data, duplicate records, and data standardization. Various feature engineering approaches were implemented to enhance model performance. Utilizing the Boosting GraphSAGE model, the study achieved an F1-score of 98.32%, a recall of 94.41%, and an accuracy of $\mathbf{9 8. 2 5 \%}$, demonstrating its effectiveness in identifying and mitigating cybercrimes.