Phishing Uniform Resource Locator Detection

Phishing attacks have become one of the most prevalent forms of cybercrime, that can be targeting users through deceptive URLs that mimic legitimate websites. In this research, we propose a hybrid deep learning approach combining the both algorithms called Long Short-Term Memory (LSTM) and Artificial Neural Network (ANN) models to detect phishing URLs effectively. The system is implemented as a web-based application that enables real-time detection and visual representation of model performance. A benchmark phishing dataset is used, and the URLs are preprocessed using tokenization and padding techniques to prepare them for model training. The LSTM model captures sequential dependencies in URL structures, while the ANN model serves as a comparative classifier to enhance decision-making. Performance metrics such as accuracy, confusion matrix, and visual comparison graphs are presented to analyze the efficacy of both models. The LSTM model achieved higher accuracy, validating its strength in sequential data processing. The integration of both models in a web interface offers an interactive and user-friendly experience for phishing detection. This research demonstrates the feasibility and effectiveness of dual-model approaches in strengthening cybersecurity systems.