PRIVACY PRESERVING USING ADDITIVE PERTURBATION BASED ON MULTILEVEL TRUST IN RELATIONAL STREAMING DATA

Abstract

Data perturbation is one of the very popular model that is used for privacy preserving data mining. The previous privacy preserving solutions were limited to only single level trust, which was not sufficient to preserve the privacy of information. So by expanding the scope from single level trust, here in the proposed system, multilevel trust solution for privacy preservation is applied in which data owner generates the different perturbed copies of same data for data miners of different trust levels. In the proposed system additive perturbation approach is used to generate the perturb copies of the relational streaming data. The data miner may mangle the different perturbed copies at different trust levels to collect the extra information about the original data, this is called the diversity attack. So to prevent from diversity attack Multilevel Trust Privacy Preserving Data Mining (MLT-PPDM) approach is used with the addition of Gaussian noise added to original relational data