A Prediction Model of Insider Threat Based on Multi-agent

Current intrusion detection systems are almost unable to repel the threats from internal users of organizations. In order to mitigate insider threat availably, in the paper we present a multi-agent based predicting and detecting model and describe implementation algorithm of related agents. It can continually report the attack probability of an attacker to system administrators, and provide a quantitative approach to help administrators make sound decision