Phish Check: Content-Based Phishing Website Detection Web App

This research proposes an innovative method for categorizing phishing and legitimate websites based on content features extracted from their HTML structure. Leveraging data collection with the requests library and HTML parsing using the Beautiful Soup module, our study scrutinizes 26,584 websites sourced from “phishtank.org” and “tranco-list.eu.” Employing seven diverse machine learning classifiers from the sci-kit-learn library and implementing k-fold cross-validation, this research study assesses the efficacy of different models in accurately distinguishing phishing websites. Evaluation metrics include confusion matrices, accuracy, precision, and recall scores. The proposed approach refrains from using URL-based attributes, focusing solely on the characteristics of HTML content. This study presents the results and comparisons of various classifiers, highlighting the effectiveness of our methodology.