Research on Abnormal Behavior Detection Method Based on User Access Sequence Analysis

With the rapid development of enterprise informatization, traditional perimeter security protection technologies have gradually exposed their limitations in dealing with unknown threats. In response to this issue, this article proposes an anomaly behavior detection method based on user access sequence analysis. This method constructs a multi-level and multi-dimensional adaptive baseline model through deep data modeling and machine learning algorithms. By introducing sequence correlation analysis and spatiotemporal distribution feature analysis, comprehensive modeling was conducted from three dimensions: individual-level behavior patterns, position-level behavior patterns, and group-level behavior patterns, significantly improving the perception ability of covert network threats. The experimental results indicate that the multidimensional sequence baseline weighted scoring mechanism proposed in this study can effectively identify potential abnormal access behaviors. This provides a new technological support method for enterprise network security protection system.