Based On The Mrf Model Network Trading Behavior Analysis

Although the vigorous development of network transactions brings convenience to the people, it also brings security risks. How to detect and warn the abnormal behaviors of network transactions are the key challenges of the online network transactions. In this paper, we focus on the behavior monitoring and behavior analysis of network transaction two key technologies.Firstly, in order to obtain the basic behavioral data of network transaction analysis, we establish a "monitoring agent-monitoring center-information manager" three layer structure frame mechanism when the network trading software is running. On this mechanism, we use the reflection technology to obtain relevant event information, based dynamic aspect oriented programming monitor to flexibly monitor information, and adopts a method of pull-and-push to collect monitoring information. The realizing the dynamic monitoring the interactive behavior the software provides a data basis for the follow-up behavior analysis.Secondly, according to the key factors that affect the behavior analysis of network transactions, we proposed a behavior intention analysis method based on Markov random field model. In this method, transactions between users can use the graphical modeling, priori probability between points can be determined, and proposed a Multi-level factors algorithm to combine the transaction key factors of the model, then using belief propagation algorithm to solve the model value and analyze the user’s behavior intention. In the simulation, this method can effectively realize the transactions between users, and can better identify transaction abnormal behavior.Finally, we use Matlab to simulate the user behavior analysis method based on MRF model and to evaluate it’s performance, the testing and analysis demonstrate that the proposed method is feasible and effective.