The Research On Database Safe Detection Based On Data Mining

As the wild spread use of information technology, database has gotten extensive application in many fields. People are also paid more attention on data information safety. This thesis mainly makes some works on read and safety detection of database log.In the information system currently, we have adopted various measures to protect the safety of information system. Most of them aiming at network and virus, but the supervisal for database are comparatively rare, especially on the supervisal of database operator (e.g. DBA). Since database log is faithful recorded all of the database operation, we start with database log, to find some needful information based on the analyses and mining of the log.This paper aiming at database of oracle8 edition above, have studied the management and storage rules of oracle database, and realized the local or remote read real time of database log. It is convenient for the supervisor of database department, to manage database, the system can track the source of data change in time and offset the shortage of data management.We have adopted the method of data mining for database log detection, used decision tree and association analysis for it. After the log drawn from oracle, the system firstly carry out data preparation and preprocessing. because of there are some relatives between all kinds of the tables in the database, therefore, we first carries out association data mining for the data after preprocessing, find strong relation set, we use the relation set as basis for data train. Decision tree is set up for training set to get the decision tree normally.After detection model is established, it can be real time to analyze and forecast the log in normally operation.This paper establishment based on Oracle9i & PowerBuilder can realize the analysis and data mining for oracle database basically.