This dissertation explores the development of a web crawler that utilises machine learning for detecting malicious websites. The research focuses on enhancing detection by analysing both URL-based and content-based features, including visible text and metadata, and exploring whether sentiment-based analysis can improve the accuracy of detection models. Traditional methods rely heavily on URL lexical analysis, which is often insufficient due to evasion techniques like URL shortening. This project aims to address these gaps by developing an ensemble machine learning model capable of improving detection accuracy through multi-dimensional feature analysis, contributing to more robust cybersecurity solutions. Additionally, the dissertation includes the development of an open-source dataset and a user-friendly dashboard for practical application in cybersecurity efforts.