User account risk identification model for web applications

With the continuous development of the Internet era, the information security environment faced by Internet users is becoming more and more severe. In view of the intensified user account theft on the Internet, this paper analyzes the user behavior habits by collecting user behavior information and application log, and proposes a user account risk identification algorithm based on user behavior. In order to improve the accuracy of user account risk identification, Firstly, use the Kmeans algorithm to cluster user accounts based on user behavior data. In the clustering process, the PSO (Particle Swarm Optimization) algorithm is introduced to form an improved PSO_Kmeans clustering algorithm. Then, extend log data with imported external “threat intelligence” data, to classify the clustered data, using Random Forest, Decision Tree, Naive Bayesian machine learning classification algorithm. The experimental results show that the model can effectively identify the risk account.