A Method for Multi-source Heterogeneous Log Correlation Based on Conditional Independence
Liyue Chen, Chen Luo, Xianfu Zhou, X.Q. Wang, Dong Yan, Sijia Wei · 2024
This study explores the alignment of database audit logs with network server logs to uncover potential system relationships. We propose a method combining temporal correlation screening and conditional independence testing. Pattern access lists are generated from HTTP and SQL logs, followed by temporal correlation to identify relationships. The chi-square test's p-value quantifies these associations. Results validate the method's efficacy in correlating HTTP requests with SQL operations, offering novel insights for heterogeneous log alignment, aiding system monitoring, and optimization.