A Method for Hiding Association rules with Minimum Changes in Database

Abstract

Privacy preserving data mining is a continues way for to use data mining, without disclosing private information. To prevent disclosure of sensitive information by data mining techniques, it is necessary to make changes to the data base. Association rules are important and efficient data mining technique. In order to achieve this algorithm is proposed, that as well as hiding sensitive association rules, having the lowest side effects on the original data set. Proposed algorithm by removing selective item, among items of antecedent sensitive rule (L.H.S.), causes to decrease confidence of sensitive rule below less them threshold and hide the sensitive rule. Also keeps sensitive rules until the end of securing process is reduce the failure hiding, and because the internal clustering, hiding sensitive rules performed synchronic takes insensitive rules to reduce the loss. This algorithm is compared with basic algorithm, on dense and sparse data base. The results with criteria of hiding failure, is indicates 41.6% improvement in dense data base and 28% in made with software data base. With criteria of lost rule, is indicates 70%, 57.1% and 83.3% improvement over the base algorithm. Which indicates the proposed algorithm is efficient.