Affair increasing and reducing method for hiding association rules

Abstract

The invention provides a hidden affair increase and decrease method used for an association rule which belongs to the data mining and knowledge hidden field. The method combines a process for adding weak relative affair with a process for removing strong relative affair based on contribution degree of affair to the mining rule. The method needs not increase and decrease item, only needs to amend one item, and gives an adjusted factor for updating the affair amount with various size to analyse various hidden effect. An experiment with single rule hidden, multi-rule hidden and anti-support attack is carried out by using the method, the experiment result shows that if only the adjusted factor reaches 0.5, the hidden rate can be kept in 100%; in side effect, the rule loss rate and the rule increase rate are less than 15%; the method can fight back the support attack in large range, and when the adjusted factor reaches above 0.07, the hidden rate is kept in 100%. The method has little amending amount to a database affair, is suitable for random combination hidden with multi-rule, and has strong superiority in hidden effect, stability and practicability.

Description

technical field [0001] The invention relates to a method for hiding association rules, belonging to the fields of data mining and knowledge hiding. Background technique [0002] Association rule extraction is one of the important methods of data mining technology, which can discover the knowledge that users are interested in from a large amount of data. With the continuous demand for network information sharing and publishing, users hope to provide important references for enterprise competition through in-depth analysis of these data. But on the other hand, information security issues are increasingly exposed. In many fields such as business, medical care and insurance, people don't want their private information to be known by others. Therefore, how to share data without revealing sensitive information and finally mine useful information for users has become a key issue in data mining research. [0003] Association rules describe the potential relationships that exist b...

AI Technical Summary

Problems solved by technology

Therefore, the application of the law is more limited

[0013] (4) Reverse the BOOL item value: This method is easy to understand, but it is only applicable to the item value of BOOL type, and it is not suitable for the value type

However, due to the complexity of the interval division itself, this processing method is difficult to be practically applied

[0014] In addition, the results of these methods in terms of attack resistance have not been reported

Images