Hiding Sensitive Association Rules by Distorting RHS Items
Parvin Shirrouhi, Mohammad Naderi Dehkordi, Faramarz Safi · 2013
Association rules is a data mining technique which extracts useful patterns in the form of laws. One of the major problems in applying this technique on a Dataset is the disclosure of sensitive information which would endanger their security and confidentiality. Privacy-preserving data mining is to preserve the privacy of the personal data identified by the data mining techniques. Hiding association rules is one of the methods in privacy-preserving. In this article, one hiding association rules algorithm has been discussed. In the proposed algorithm for hiding sensitive rules, Data Distortion Techniques— based on reduction of confidence rules— has been used. Reduction of confidence in sensitive rules, through the reduction of right side Items Set support, and working on a series of transactions, which totally support the sensitive rules, by choosing a transaction with the least amount of items. Our proposed algorithm with two reference algorithms on both compacted and non-compacted Dataset has been implemented, we observed that the execution time of the proposed algorithms are compared with both reference on both Dataset has been considerably reduced. Also, the number of lost rules, the non-compacted Dataset, the proposed algorithm is more efficient than the two reference algorithms. The effectiveness of the proposed algorithm has been analyzed through the implementation and comparing of the obtained results with the reference algorithms. The results indicate that the proposed algorithm is effective.