Investigating the Effectiveness of Graph-Based Algorithms in Identifying Insider Threats and Collusion: A Machine Learning Approach for Anomaly Detection in HR Networks

Abstract

Data security and organizational integrity are seriously threatened by insider threats and collusion in the rapidly changing cybersecurity landscape. Because they are unable to analyze complex and dynamic human resource (HR) networks, traditional approaches of detecting such risks frequently prove to be inadequate. In order to discover anomalies in HR networks, this study examines how well graph-based algorithms detect insider threats and collusion using a machine learning technique. Using organizational data, the study first builds HR networks, which depict workers and their connections as nodes and edges within a graph structure. These networks are analyzed using a variety of graph-based methods, such as Graph Neural Networks (GNNs) specifically Local Outlier Factor (LOF), and community discovery techniques specifically Centrality Measures. The algorithms’ goal is to find trends and abnormalities that point to possible collusion and insider threats.