Abstract

<p>At present, insider threat detection requires a series of complex projects, and has certain limitations in practical applications; in order to reduce the complexity of the model, most studies ignore the timing of user behavior and fail to identify internal attacks that last for a period of time. In addition, companies usually categorize the behavior data generated by all users and store them in different databases. How to collaboratively process large-scale heterogeneous log files and extract characteristic data that accurately reflects user behavior is a difficult point in current research. In order to optimize the parameter selection of the DBSCAN algorithm, this paper proposes a Psychometric Data & Attack Threat Density Based Spatial Clustering of Applications with Noise algorithm (PD&AT-DBSCAN). This algorithm can improve the accuracy of clustering results. The simulation results show that this algorithm is better than the traditional DBSCAN algorithm in terms of Rand index and normalized mutual information.</p> <p> </p>

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call