Hybrid Solution for Privacy-Preserving Data Mining on the Cloud Computing

Abstract

The continuous evolution of the cloud has attracted many enterprises to outsource their data to the cloud to accomplish data mining tasks and other services. So that data owners can enjoy the benefits of the cloud without fear from violations of data privacy, PPDM approaches came to protect the privacy of data while preserving the usefulness of data. This article reviews the most popular models of PPDM over the cloud along with their strengths and the weaknesses to conclude the research gap. Some of the current PPDM models still vulnerable to various types of attacks like the K-anonymity model whereby surface from background knowledge attack, homogeneity attack, similarity attack and probabilistic inference attack, others of them consume great computational complexity, for example, the methods which depended on cryptography. The hybrid solution was proposed to protect the data privacy and overcome the current problems of PPDM include identify and attributes disclosures whereby preservation the privacy of data before to outsourced on the cloud is the main focus. K-Anonymization is combined with homomorphic encryption to rid from their limitations and take their advantage together to enhance data privacy-preserving and maintain the data utility of outsourced data.