基于图计算的企业员工异常网络行为检测方法

The Method of Detecting Abnormal Network Behavior of Employees Based on Graph Computing

通过高效、自动化的方式检测企业员工的异常网络行为是企业保护敏感信息以及规范员工网络行为的有效途径之一。针对这一场景,通过可视化机器学习平台Sophon,提出一种风险传播算法结合IsolationForest 算法的异常网络行为检测方法。实验结果表明,该方法在查准率、查全率等多项性能指标上均有较好的表现。

Using efficient and automatic methods to detect abnormal network behavior of employees is one of the most effective ways to protect sensitive information of enterprises and standardize network behavior of employees . To tackle this scenario, a detection method which combines a risk propagation algorithm and the Isolation Forest algorithm is proposed based on the visualized machine learning platform Sophon. The experimental results illustrate that it achieves a high performance in the sense of precision, recall and other metrics.