Safeguarding Web Environments Through Supervised Learning-Based Cross-Site Scripting Detection

Cross-site scripting (XSS) attacks have emerged as an important issue in the Internet of Things(IoT) ecosystem as a result of the expansion of IoT devices. The use of machine learning(ML) methods to predict XSS threats in IoT systems is examined in this paper. Several ML approaches for XSS detection in IoT including Naive Bayes(NB), Logistics Regression(LR), Decision Tree(DT) and Adaptive Gradient Boosting(AdaBoost) approaches have been compared that make use of a customised feature set that includes device conversations, communication patterns, and packet characteristics. The framework shows its effectiveness in accurately detecting XSS assaults through the use of supervised learning techniques. The effectiveness of the framework in a variety of device kinds and attack situations has been validated by extensive testing utilising various IoT datasets. Additionally, it looks into the model's accessibility, illuminating the warning signs of XSS attacks particular to the IoT environment. The research fills a crucial role in the protection of linked devices from more sophisticated cyber threats, making a contribution to the emerging subject of IoT security. The research provides a foundation for creating resilient defence systems that utilise the power of ML to protect IoT ecosystems against XSS vulnerabilities as the IoT environment continues to develop. The paper predicts XSS attacks that used various ML approaches, including NB, LR, DT, and AdaBoost, using the XSS attack detection dataset. Accuracy, sensitivity, precision, specificity and F1-score performance criteria have been used to identify the best ML model. The AdaBoost model, with an accuracy rating of 99.92%, is the most accurate, preceded by the NB, LR, and DT models, with accuracy ratings of 88.89%, 99.85% and 99.89%, respectively.