Design And Implementation Of Web Attack Detection And Analysis System Based On Machine Learning

With the development of communication technology and Internet,Web services have been more and more widely used.However,the attack technology for Web application is also upgrading.For organizations that provide Web services,Web attacks seriously threaten the security of their intranet,so it is necessary to detect and analyze Web attacks in real time in order to take timely Web protection measures.First of all,this paper realizes the detection of Web attack behavior by identifying malicious URL.Based on the text features of malicious URLs,a one-dimensional convolutional neural network model for URL multi classification is designed.Experiments show that the malicious URL detection model proposed in this paper can achieve 99.65%accuracy and recall rate.Compared with other common malicious URL detection models,it has more accurate detection effect and higher efficiency.Secondly,this paper constructs the knowledge graph of Web attack domain,and builds the taxonomy in the knowledge graph by introducing the concept of basic level category.Web attacks are represented as entities,and the graph has three levels:the concept,the entity and the attribute.Compared with common open-domain knowledge graphs,this knowledge graph contains more entities of Web security threats,and has a more detailed taxonomy.This domain knowledge graph can provide query and help with decision-making for Web attack analysis.Finally,we design and develop a Web attack detection and analysis system,which consists of five functional modules:traffic collection,attack detection,data storage,asset monitoring and result display.It can detect Web attack traffic in real time,and provide attack analysis function from four dimensions:knowledge graph,asset status,attack details and attackers' portrait.The system can help network administrators discover and respond to Web attacks in time,and ensure network security.