An Intelligent Deep Learning Model for XSS Attack Detection using Enhanced LSTM

Cross-site scripting (XSS) attacks remain a significant threat to web applications, allowing attackers to inject malicious scripts into trusted sites. This paper presents a hybrid detection framework that combines Term frequency-inverse document frequency (TF-IDF) for contextual feature extraction with long short-term memory (LSTM) networks to capture sequential characteristics of input payloads. The model was evaluated on benchmark datasets and achieved a precision of 99.5% and a recall of 97.9%, outperforming traditional machine learning models. A lightweight deployment using Flask demonstrates its suitability for real-time XSS detection. Limited explainability is also incorporated to support practical auditing.