A Real-time Detection Scheme of User Behavior Anomaly for Management Information System

Abstract

The management information system is a person-oriented system, which uses computer hardware, software, network communication equipment and other office equipment to collect, transmit, process, store, update, expand and maintain information. Its security is very important to the country and the society. Aiming at the illegal operation of the legitimate users and the illegal user camouflaging into the system, a targeted user behavior anomaly detection scheme is proposed. The user behavior data is obtained by the Web log and click stream. And the user behavior is divided into the role attribute label and behavior habit label. The user behavior data under two kinds of labels are modeled respectively, the user's role behavior contour and user's customary behavior contour are obtained. The two behavior profiles are fused to get the final user behavior model as the basis for the pattern matching and the real time detection of abnormal behavior. And the real-time detection system of user abnormal behavior is designed to meet the special requirements of the management information system, to realize the timely discovery of illegal operations and illegal users and to realize real-time early warning, which could enhance the security guarantee of the management information system.