Research On Simulation Technology For User's Email Behavior Based On BP Neural Network And Agent Model

As it operates conveniently and transfers message rapidly,email has become an important tool for internet users to communicate with each other.Delivering information by email is so convenient that internet is filled with spam.Spam always contains URL in its content,but network security issues that hidden in these URLs are likely to threaten user's host security.So,the internet user's operation on email may directly affect the security of its own host and the whole computer system.We call the internet user's operation on email “user's email behavior”,including email reading,sending,deleting and so on.The hidden network security threat can be found through the analysis and simulation of user's email behavior.At the same time,simulation of user's behavior on email can also be applied to simulate network scenario in the cyber range or network simulation platform,which makes the simulation of network environment more real.In this paper,BP-BDI model is proposed to simulate the behavior of user,the experiment proved that the model has a high accuracy for the simulation of user email behavior.The main work of this paper is as follows.(1)The pattern of users' email behavior is mined by analyzing internet user's behavior data.In this paper,we analyze user's email behavior from the aspects of user's email network,user interest topic and time pattern of the user's behavior.(2)The user's email behavior is simulated according to the characteristic of users' email behavior.In this paper,BP-BDI model is proposed to simulate the user's email behavior.The mechanism of auto-update belief set and user's email behavior learning which is derived from BP neural network is built in BP-BDI model based on BDI model.So,the accuracy of behavior simulation is improved.(3)A prototype system is realized for simulation of the user's email behavior.The user's email behavior simulation system has realized the function modules such as email data parse,user's behavior data storage,behavior analysis and user's email behavior simulation.