Detección y mitigación del phishing en correos electrónicos mediante aprendizaje automático y análisis de datos preexistentes

Abstract

An academic institution is a very important organization; if its institutional email is compromised, the confidentiality of the centralized email system and access to many resources are at risk. This project seeks to develop a model for detecting fraudulent emails using supervised machine learning (M_L) techniques and natural language processing applied to pre-existing data. The process involves reviewing the textual content, including preprocessing, extracting input characters, and training binary classification models. This process establishes the threshold for distinguishing legitimate emails from malicious ones. The work is being carried out in an academic setting at the Carried of Computer Science of the UPS. It is currently in the research phase, and although the model is not yet implemented in real-world systems or in real time, its effectiveness is being assessed using metrics such as accuracy, recall, and F1 score. This research project has confirmed—in educational environments—the feasibility of using machine learning and textual analysis to detect phishing, laying the groundwork for future research and raising awareness of cybersecurity within the Computer Science program.