The Extended Research On Web Vulnerability Defense Framework Based On SVM

With the development of Internet,network security has drawn more and more attention.Among them,web application vulnerabilities have become one of the most serious security risks.SQL injection vulnerabilities and cross-site scripting vulnerabilities are two of the most prominent ones.In the past,the defense technologies for XSS attack and SQL injection attack mainly include perfecting the filtering strategy and using secure coding.The detection methods mainly include static analysis,dynamic analysis and th e combination of the two methods.But in the face of increasingly diverse means of attack and massive logs,the previous defense means of web attack seems a bit inadequate.Machine learning method is a newer method of web attack detection in recent years.In view of the above problems,machine learning is an effective solution to the problems of an endless stream of attack tools and a large number of log s.In this thesis,we propose a web attack defense framework based on SVM and introduce an attack detection query model based on SVM inside of the framework to solve the problem of low efficiency of web attack detection.And we give the implementation of feature extraction,training and classification of support vector machine and detection model.This thesis introduces feature extraction algorithm and support vector machine into the web attack defense framework to expand the research.The defense framework can detect and defense attacks mainly for cross-site scripting attacks and SQL injection attacks.Because of the specificity of Dom-XSS attacks,traditional defenses against cro ss-site scripting attacks are generally only applicable to the reflected and storage-based cross-site scripting attacks.By modifying the script parsing engine,this thesis achieves protection against three types of cross-site scripting attacks.These are reflective and storage-based XSS attacks and DOM-based XSS attacks.Feature extraction algorithm helps to solve the problem of feature redundancy and improve the detection efficiency.Attack defense based on SVM can effectively solve the problem of the higher false positive rate and false negative rate of traditional detection and defense methods.In the end,the proposed web attack defense framework based SVM is simulated experimentally.In the case of a certain number of collected sample sets,we set up the experimental environment DVWA to test and get the experimental results.Compared with other defense models,the result shows that the proposed security defense framework has a higher detection rate,a lower false negative rate and false positive rate.