Research And Implementation Of Probability Graph Classification Method For Telecom User Behavior

With the continuous development of information technology,telephone communication has become an indispensable social method in people’s life.The telecommunication fraud is constantly invading people’s normal life along with the development of telephone communication.As a new way of crime,telecommunication fraud often makes people unable to guard against it because of its non-contact and diverse methods,which has caused huge property losses to the society.Due to the low proportion of victim data,and criminals’imitation of normal users’behavior to escape from interception,the effect of traditional means to identify and intercept telecommunication fraud is not as efficient as expected.Therefore,it is urgent to put forward more efficient methods to address on telecommunication fraud.Based on the above situation,this study proposed a probability graph classification method of Telecommunication user behavior.Through the combination of probability theory and graph structure,telecommunication fraud users in communication social network can be effectively identified.This probability graph classification method based on social communication not only makes use of the call characteristics of telecommunication users,but also takes into account the social relationship among users.So that telecommunication fraud and normal users be classified and detected based on different social behaviors and social objects.Firstly,the study analyzes the key difficulties of applying telecommunication behavior analysis and an probability graph classification method.It proposes a probability graph classification model of telecommunication user behavior according to the analysis results,and the communication characteristics and social characteristics of users are integrated into the communication social network structure.Then the fraud probability of user nodes is calculated through the confidence propagation algorithm between users.Secondly,the probability graph classification method is implemented and tested,and the feasibility of this method is verified by the comparison on the real communication data set.Finally,based on the probability graph classification method,the application system in real life is designed in detail.The effectiveness of the system is verified by testing the function of the system.