Method for automatically discovering auditing rules and abnormal data

The invention discloses a method for automatically discovering auditing rules and abnormal data. The method comprises the steps: dividing an audited data set into classification attributes and numerical attributes according to attribute types, and enabling a user to give the minimum support rate of the classification attributes and the offset rate of the numerical attributes, and then selecting candidate classification attributes and attribute values with the support rate higher than the minimum support rate from the classification attributes; generating a candidate attribute set according tothe attribute type so as to obtain an auditing rule; and finally, according to the obtained auditing rules, screening the collected auditing data sets in sequence, and selecting data which meets condition constraints in the auditing rules and has derived attributes of which attribute values are not in a normal value range, namely abnormal data. The method for automatically discovering the auditingrule and the abnormal data can effectively improve the auditing efficiency and reduce the auditing cost.