Research On Detection Method Of Phishing Web Page Based On Deep Learning

With the rapid development of information technology,the Internet has become an important part of people's daily-lives.At the same time,network security issues are also emerging.Phishing is one of the problems.It deceives users to steal sensitive information to realize phishing attacks,causing economic losses to users.Phishing has seriously threatened network security,and how to effectively contain phishing attacks has become an important research task in the field of network security.For the problem that the existing webpage features cannot quickly and effectively detect new phishing webpages and the complex model calculation,a phishing webpage detection algorithm based on new feature synthesis and main and auxiliary models is proposed.The algorithm extracts 39 features from the webpage URL(Uniform Resource Locator)and HTML(HyperText Markup Language)source code,including two new features to detect phishing web pages.At the same time,the main feature and auxiliary feature set are constructed according to the importance of the feature,and the MACB(Main-Auxiliary-CNN-BiLSTM)main and auxiliary model is constructed to process the main feature and auxiliary feature,so as to reduce the calculation complexity of the model and improve the unknown webpage discriminate efficiency.Experimental results show that the proposed algorithm has better detection efficiency and detection accuracy.To solve the difficulty of manual feature engineering in traditional phishing webpage detection methods and the inability to deal with hidden abstract features,a phishing webpage detection algorithm based on representation learning is proposed.The algorithm parses the web page into three parts:URL,Document Object Model(DOM)and HTML text content to comprehensively analyze the phishing webpage,and characterize the webpage without feature loss to the matrix space,avoiding complex feature engineering problems.At the same time,a deep learning model is constructed to mine internal details and contextual information of the web page,and extract deep-level implicit abstract features.And add attention mechanism to enhance the recognition ability of feature information learning process,highlight the role of important information,improve detection accuracy,and finally build a MICBA(MultilInput-CNN-BiLSTM-Attention)multi-data input attention model to realize detection of phishing webpages.The experimental results show that the proposed algorithm can efficiently deal with and complex problems,and has achieved good performance in the comparison of multiple experimental evaluation indicators.On the basis of the proposed algorithm theory,a phishing webpage detection system is implemented.The system is applied to the browser by combining with the browser plug-in.For the visited webpage,the system can detect whether it is a phishing webpage in real time.