Security-Aware Malicious Hyperlinks Phishing Detection Using Combined Machine Learning Models

Abstract

Phishers utilise email phishing through URLs that are obfuscated, malicious, or phished, and they continually adapt or reinvent their techniques in order to entice victims. The problem of phishing attacks in enterprise is the next issue that is rising in wide scale and complexity. The use of visceral variables and familiarity signals has become more common in phishing attempts in order to earn the trust and confidence of victims. It would be naive to think that phishing is always focused on financial gain, even if it is usually the phisher's obvious goal when they commit identity theft. The goodwill and character of an internet user can also be taken by a phisher. In this kind of situation, a phisher has no boundaries. Making a fool of oneself in the academic or professional world might be more fatal than revealing oneself on a social media site. It is not an easy process to resolve this matter. An analysis of the available literature reveals that traditional methods of phishing detection filters are inadequate for spotting the many types of phishing attempts that can occur in a corporate setting. As a result, we provide an innovative anti-phishing solution for businesses based on an artificial neural network. This approach also successfully determines if an email is known or unknown phishing, which helps to lessen the impact of trust and familiarity-based email phishing in business environments. To improve the URL categorisation process, we use the Feed-Forward Backpropagation and Levenberg-Marquart methods of Artificial Neural Networks (ANNs). To acquire results with imprecise data of social aspects, we use the Fuzzy Inference System. When it comes to URL-based email phishing, the suggested model can correctly categorise both common and uncommon examples.