Gaussian Noise Multiplicative Privacy for Data Perturbation Under Multi Level Trust

Abstract

Data mining is the technique of exploring and analyzing huge blocks of information to find significant trends and patterns. Perturbation is a mechanism that has been introduced in the fields of celestial mechanics and mathematical physics. Each characteristic has a weight that represents how accurate and comprehensive it is. Database and data security administrators are forced to perform a difficult balancing act when it comes to granting employees access to organizational data. for this To use multiplicative data perturbation in conjunction with single level and multilayer trust the geometric type of multiplicative data perturbation will be carried out in this method, as well. When generating the perturbed copy, geometric perturbation involves the orthonormal matrix, translational matrix, and a random generated Gaussian noise vector, among other things. In the beginning, the orthonormal matrix will be used to perform the rotation perturbation, and then the translational matrix and Gaussian noise components will be added to it for the final perturbed copy. We can say that under single level trust, additive Gaussian data perturbation produces perturbed copies using uniform Gaussian noise. Regardless of their trust ratings, all data miners receive the same perturbed copy. Additive Gaussian data perturbation at multi level trust is studied for data miners at various trust levels. A common randomization technique that guarantees confidentiality and trustworthy data mining findings is data perturbation.