Static semantic and matrix-based approach of abnormal transaction recognition
Lichen Zhang · Jisuanji gongcheng · 2004
Using intrusion detection to find abnormal transaction is common nowadays. But many signature detection and non-signature detection approaches are at the operating system level not at the DBMS level ,and cannot guarantee the prompt response for all abnormal behaviors. A new approach for abnormal transaction recognition was presented. We used static semantic analyzing the read, pre-write and post-write set of data objects modified by transactions. If they were inconsistent with the relative transaction operation,the transaction was confirmed abnormal transaction. Then we calculated the real-matrix, write-matrix and damaged-data-vector and found the damaged data objects by malicious transaction.