Malicious URL Detection Using Decision Tree-based Lexical Features Selection and Multilayer Perceptron Model

Network information security risks multiply and become more dangerous. Hackers today generally target end-to-end technology take advantage of human weaknesses. Furthermore, hackers weaknesses by applying various methods to attack. Nowadays, one the greatest dangers modern digital world is malicious URLs, stopping them biggest challenges in field cyber security. Detecting harmful URLs using machine learning deep algorithms have been subject academic papers. However, time accuracy are two these tools. This paper proposes a multilayer perceptron (MLP) model that utilizes significant aspects make it practical, lightweight, fast: Using only lexical features decision tree (DT) algorithm select best relevant subset features. The effectiveness experimental outcomes evaluated terms time, accuracy, error reduction. results show MLP 35 could achieve an 94.51% utilizing URL improved after DT as feature selection with slight improvement loss.