Research on Deep Learning Driven XSS Vulnerability Detection Technology: From Feature Extraction to Real-time Monitoring
TOTAL VIEWS: 223
Abstract
Cross-site scripting (XSS) vulnerabilities, characterized by their stealthy nature and adaptive capabilities, remain the most prevalent security threat in web applications. Conventional rule-based detection methods struggle to address frequent vulnerabilities. This study focuses on deep learning-based XSS detection, adopting a three-phase approach: feature extraction, model construction, and real-time monitoring. We propose multi-dimensional feature extraction methods including word embedding and syntax tree analysis, effectively overcoming limi-tations of traditional signature engineering in capturing evolving attack patterns. A hybrid neural network model combining CNN and LSTM is developed, where CNN identifies localized character patterns while LSTM analyzes contextual relationships, significantly enhancing detection capabilities against complex XSS scripts. A lightweight real-time monitoring framework utilizing edge computing is designed, achieving latency under 50 milliseconds-meeting stringent requirements for web system protection. Experimental results demonstrate 98.2% accuracy and 97.8% recall rate, with 32% fewer false positives compared to rule-based detection methods. This innovative approach provides an effective and intelligent defense mechanism against XSS vulnerabilities.
Keywords
Deep learning; Vulnerability mining of XSS; Feature extraction; CNN-LSTM mod-el
References
[1] Zhang M. Research on automated detection technology of industrial internet XSS vulnerabilities based on deep learning. Wireless Interconnect Technol. 2025;22(9):105-108.
[2] He ZY, He CW, Chen W, et al. Code vulnerability detection based on implicit flow analysis and deep learning. Comput Eng Des. 2025;46(7):1951-1958.
[3] Yang DF, Jiang XW. Research on XSS attack detection based on deep learning. Jiangsu Commun. 2023;39(3):94-102.
[4] Lin YB, Ling J. A method for XSS attack detection based on residual network and GRU. Comput Eng Appl. 2022;58(10):101-107.
[5] Jiang YM, Luo XY, Yu M, et al. A Web attack detection method based on bidirectional long short-term memory neural network. Inf Countermeas Technol. 2023;2(1):55-65.
[6] Pang B. Application of knowledge graph technology in vulnerability detection of computer network links. Inf Syst Eng. 2025;(12):61-64.
[7] Qiu B. Exploring the application of security vulnerability detection technology in computer software. Cybersecur Technol Appl. 2025;(10):73-75.
[8] Liang ZF. Research on security vulnerability detection and protection technology in the big data environment. Autom Appl. 2025;66(18):267-269.
[9] Host. Application of knowledge graph technology in computer network link vulnerability detection. Cybersecur Informatiz. 2025;(07):148-150.
How to cite this paper
Research on Deep Learning Driven XSS Vulnerability Detection Technology: From Feature Extraction to Real-time Monitoring
How to cite this paper: Huang Li, Amirrudin Kamsin. (2025) Research on Deep Learning Driven XSS Vulnerability Detection Technology: From Feature Extraction to Real-time Monitoring. Future Trends in AI Research, 2(1), 17-20.
DOI: http://dx.doi.org/10.26855/ftair.2025.06.004
More Articles
- Research on Deep Learning Driven XSS Vulnerability Detection Technology: From Feature Extraction to Real-time Monitoring
- Analysis of the “Golden Corners, Silver Edges, and Grass Belly” Principle in Go from a Game Theory Perspective
- Breaking the Boundary of the Concept of Inventor in Patent Law in the Age of Artificial Intelligence——The Limitations of Anthropocentric Legislation: A Look at the DABUS Case
- Semantic Enhancement Strategy for Knowledge Graph Completion