Designing an Intelligent Ensemble System for Phishing Website Detection

As technology advances and electronic communication dominates our daily lives, protection against security threats has become increasingly important. Phishing is a common cybersecurity threat that targets human vulnerabilities, tricking internet users into accessing fake websites through communication channels such as email, SMS, or social media, ultimately benefiting the attackers. Due to the evolving nature of phishing attacks, phishing detection has become a dynamic and challenging issue, necessitating the development of countermeasures. This paper presents an ensemble classifier that combines Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) to detect phishing websites by extracting ۱۸ URL-based textual features. The proposed model was evaluated on a dataset comprising ۶۰,۰۰۰ samples, achieving an accuracy of over ۹۶%. The model's response time, including feature extraction and classification for each URL, is estimated at ۱۲.۴ milliseconds. This short response time justifies the use of this system in real-time detection cases. Furthermore, a performance comparison between the designed system and another approach demonstrates the superiority of the proposed approach.

Keywords:

cybersecurity , phishing websites , machine learning , ensemble learning , Convolutional Neural Network (CNN) , Long Short-Term Memory (LSTM).