Research On Application Of Phishing Detection Based On BLSTM

The upgrade and iteration of information technology has facilitated people’s lives,but it is also accompanied by the emergence of issues such as user privacy and information security.As people surf the Internet from PC to mobile,criminals also focus on mobile phones.Criminals conduct a series of phishing attacks on smartphones,constantly stealing sensitive user information.Therefore,how to better identify phishing websites and prevent users from being attacked by phishing has become a hot topic in information security research.Although different phishing website detection methods have achieved good results,due to the update and iteration of phishing attack methods,current phishing detection still faces many challenges,and there is still room for improvement.Based on the deep learning detection algorithm,this thesis proposes corresponding improvement measures for the current problems in this field.The main work of this thesis is as follows:First,this thesis analyzes in detail the shortcomings of the methods of URL address segmentation based on characters and URL address segmentation based on words,and proposes ideas and suggestions for solving the problem.On the basis of combining the advantages of these two methods with the relevant features of URL addresses,this thesis proposes a method of URL address division based on sensitive words and word vector embedding based on URL-related features.Then,the Bidirectional Long Short-term Memory(BLSTM)network combined with the attention mechanism is used to implement the phishing detection algorithm.Starting from different word segmentation methods and different network models,this thesis extracts,integrates and trains URL addresses.Finally,through experimental comparison,it is concluded that the method of dividing URL addresses based on sensitive words and training through the BLSTM network can obtain better experimental results.Second,in order to meet the needs of users more comprehensively,this thesis designs and completes a phishing detection application system with complete functions and simple interface based on the Android system.The application system completes the legality detection of URL addresses through processes such as URL address acquisition,judgment,transmission,and query information feedback.The system innovatively uses functions such as clipboard monitoring and blacklist query to make up for the deficiencies of functional modules such as input query module and QR code scanning module.The system implements a series of system functions such as URL phishing detection,QR code phishing detection,security assessment,blacklist detection,and clipboard monitoring.