A Method for Hiding Association rules with MinimumChanges in Database

Privacy preserving data mining is a continues way for touse data mining, without disclosing private information. Toprevent disclosure of sensitive information by data miningtechniques, it is necessary to make changes to the data base.Association rules are important and efficient data miningtechnique. In order to achieve this algorithm is proposed,that as well as hiding sensitive association rules, having thelowest side effects on the original data set. Proposedalgorithm by removing selective item, among items ofantecedent sensitive rule (L.H.S.), causes to decreaseconfidence of sensitive rule below less them threshold andhide the sensitive rule. Also keeps sensitive rules until theend of securing process is reduce the failure hiding, andbecause the internal clustering, hiding sensitive rulesperformed synchronic takes insensitive rules to reduce theloss. This algorithm is compared with basic algorithm, ondense and sparse data base. The results with criteria ofhiding failure, is indicates 41.6% improvement in densedata base and 28% in made with software data base. Withcriteria of lost rule, is indicates 70%, 57.1% and 83.3%improvement over the base algorithm. Which indicates theproposed algorithm is efficient

کلیدواژه ها:

Privacy Preserving Data Mining ، Association Rules ، Hiding Sensitive Rules ، The security data base