The Design And Implementation Of Phishing Website Auxiliary Identification System

Phishing is the most common criminal act on Internet.With the rapid development of Internet information technology,people are becoming increasingly reliant on the Internet.For the purpose of stealing online property,phishing is becoming more and more common.Phishing website is one of the forms of phishing crimes.After entering into the phishing website,one may mistakenly believes that he has entered the regular website and exposed his account password to the phisher,resulting in the hacking.As the most powerful form of phishing attack,phishing website can be spread around by means of mail,advertisements,etc.,and the penetration is extremely strong.Therefore,the establishment of a phishing website detection system with better recognition ability has high practicability and safety value.The main research contents are as follows:1)For the large number of URLs to be detected,and also for the system concurrency requirements,this thesis use the Redis database and black-white list to quickly filter out most of the URLs,and also use the lightweight URL detection module reduce the pressure of subsequent module detection.2)Using machine learning algorithm to learn and train datasets,this thesis uses cascade detection method to divide feature space into URL text features,WHOIS features and page features.Training the URL classifier with lightweight URL text features and WHOIS features,and training the page classifier with the page feature.The cascade detection method can effectively reduce the system pressure and ensure the system's ability to support concurrency under the premise of ensuring the overall detection accuracy.3)For the traditional detection problem,it can only distinguish whether it is a phishing site or not.This thesis adds a brand identification module.For samples that have been determined to be phishing websites,the module use convolutional neural network model to classify its favicon to identify its target brand for subsequent feedback notification and data analysis.Based on the above work,this thesis implements the phishing website detection system.Besides the normal work of the core algorithm,it also shows good usability and stability.This system has been applied to the back-end system of a security company and provided service for mail user and network disk user,and the actual detection result looks good.