Research And Implementation Of Web Vulnerability Intelligent Mining Technology

Web technology is developing rapidly with the popularity of e-commerce and e-government,and website security is becoming an important consideration for maintaining the stable operation of websites,so it is very meaningful to mine and analyze website vulnerabilities.Web vulnerability mining technology can be divided into source code based vulnerability mining technology and target code based vulnerability mining technology.In this thesis,we focus on source-based vulnerability mining technology(code audit).The development of traditional vulnerability mining technology is gradually entering a bottleneck.By studying Web vulnerability intelligent mining technology to find the vulnerability of Web applications faster and more accurately and repair optimization,is the current trend of vulnerability mining technology development.This thesis introduces the CNN-LSTM algorithm into the code audit of PHP scripting language,proposes a CNN-LSTM network-based Web vulnerability intelligent mining method and designs the related software,and compares the performance with other deep learning algorithms to confirm the correctness of the system design in principle and implementation technology.The thesis firstly introduces the research background of the topic,introduces the development direction and prospect of Web vulnerability intelligence mining technology,and briefly introduces the structure and content of this thesis.Secondly,the relevant concepts and theoretical foundations are studied and discussed,including the PHP scripting language and its vulnerability formation causes,followed by an introduction to the vulnerability mining objectives explored in this thesis,the causes,hazards and their avoidance methods of SQL injection vulnerabilities and XSS vulnerabilities,and then an introduction to the relevant content of the deep learning algorithm,so as to facilitate the development and conduct of subsequent research.Finally,the design and implementation process of the vulnerability intelligence mining software is introduced in detail.The preparation of the software design is completed by describing how the Web vulnerability data is processed to form the dataset available for the network model in terms of data collection,pre-processing and feature representation.Then it describes how to use the use of CNN-LSTM network to train a classification model that can accurately discriminate the vulnerability data.This study chose to combine CNN and LSTM to build a CNN-LSTM network model,which has the advantages of both CNN and LSTM algorithms,extracting the features of the data through CNN network,and then using LSTM network to classify the Web vulnerability data,making the CNN-LSTM network model not only has the model training speed comparable to CNN network,but also retains the The CNN-LSTM network model is not only faster than the CNN network,but also retains the powerful classification ability of the LSTM network.The experimental results prove that the intelligent mining method researched and designed in this thesis basically meets the requirements of intelligent Web vulnerability mining,and has certain reference value for the improvement of the intelligence level of present-day Web vulnerability mining technology.