Machine learning and sentiment analysis for predicting insider threats on email

https://doi.org/10.55214/25768484.v9i6.8230

Authors

  • Onil Chibaya Department of Computer Science and Information Technology, Faculty of Natural and Applied Sciences, Sol Plaatje University, South Africa.
  • Ibidun Christiana Obagbuwa Department of Computer Science and Information Technology, Faculty of Natural and Applied Sciences, Sol Plaatje University, South Africa.

Insider threats remain a critical challenge in cybersecurity, particularly with the increasing complexity of digital environments. This study investigates the integration of machine learning models and sentiment analysis to enhance the detection of insider threats, specifically through the analysis of email communication. Motivated by the need for more effective security measures, the research explores four machine learning models: Random Forest, Support Vector Machine (SVM) Classifier, Logistic Regression, and Decision Tree. The Random Forest model demonstrated the highest accuracy at 91%, while the SVM Classifier and Logistic Regression models achieved 72% accuracy. The Decision Tree model performed slightly lower, with an accuracy of 90%. An ensemble approach combining these models further improved the detection accuracy to 90%. These findings underscore the potential of merging sentiment analysis with machine learning to advance cybersecurity practices. The study's implications are significant for organizations aiming to bolster their defenses against internal threats by leveraging these innovative techniques. By integrating sentiment analysis with traditional machine learning methods, this approach offers a novel and more nuanced method of detecting insider threats, with potential applications not only in cybersecurity but also in broader domains that involve monitoring digital communication for malicious intent.

Section

How to Cite

Chibaya, O. ., & Obagbuwa, I. C. . (2025). Machine learning and sentiment analysis for predicting insider threats on email. Edelweiss Applied Science and Technology, 9(6), 1718–1728. https://doi.org/10.55214/25768484.v9i6.8230

Downloads

Download data is not yet available.

Dimension Badge

Download

Downloads

Issue

Section

Articles

Published

2025-06-20