Phishing URL Detection: A Real-Case Scenario Through Login URLs

Phishing is a social engineering cyberattack where criminals deceive users to obtain their credentials through login form that submits the data malicious server. In this paper, we compare machine learning and deep techniques present method capable of detecting phishing websites URL analysis. most current state-of-the-art solutions dealing with detection, legitimate class made up homepages without including forms. On contrary, use URLs from page in both classes because consider it much more representative real case scenario demonstrate existing high false-positive rate when tested pages. Additionally, datasets different years show how models decrease accuracy over time by training base model old testing recent URLs. Also, perform frequency analysis domains identify carried out phishers campaigns. To prove these statements, have created new dataset named Index Login (PILU-90K), which composed 60K URLs, index websites, 30K Finally, Logistic Regression which, combined Term Frequency - Inverse Document (TF-IDF) feature extraction, obtains 96.50% on introduced dataset.