Análisis de patrones de uso de internet a través del monitoreo de url maliciosos por definición de políticas de seguridad con algoritmos de machine learning.

Abstract

Abstract: This paper studies Internet usage patterns in a corporate network using machine learning algorithms to classify access to malicious or unauthorized URLs according to defined security policies. To train the model, 145 URL examples grouped into Games, Business and Economy, Phishing, Shopping, and Streaming Media were used. The models were then presented with 1,609 additional examples containing all the information regarding the access date, the URL itself, the connection time, the computer name, and the location (campus buildings or university sites). The analysis included Random Forest, Support Vector Machines (SVM), and Logistic Regression algorithms, which achieved similar results, reaching 84% accuracy in URL classifications. In conclusion, the results indicate that the use of machine learning algorithms is useful for establishing browsing patterns and detecting access to malicious or unauthorized URLs to strengthen security policies and network management.