关键词: Web 攻击检测, 机器学习, 深度学习, 防火墙, 蜜罐

Abstract: Web application plays a crucial role in facilitating business operation and user service for numerous organizations. Data interaction and sharing are achieved through web browsers, while due to the involvement of sensitive infor鄄 mation and business data, websites and databases often become targets of various high-frequency Web attack. With the continuous emergence of new types of Web attack, traditional Web attack detection technologies have be鄄 come inadequate, and domestic and foreign scholars have started to adopt novel approaches for attack detection re鄄 search. Based on an extensive review of domestic and foreign literature from the past three years, this paper firstly describes the process of Web attack and related concept, as well as comprehensively analyzes the current state of Web application attack. Subsequently, the latest detection methods are summarized from three perspectives of ma鄄 chine learning and deep learning, optimization of Web application firewall (WAF), as well as Web honeypot de鄄 ception technology, and the performance and advantages of different models are compared and analyzed. Finally, the challenges associated with detecting Web attack are summarized, and prospectsfor future research are provided.

Key words: Web attack detection, machine learning, deep learning, firewall, honeypot