Research On Data Perturbation Privacy Preserving Method For Distributed

As computation, communication and digital storage technology advances, the amountof data is rapidly growing. Organizations make extensive use of data mining: commercialorganizations rely on data mining to improve the business strategy to obtain greater profits;medical institutions rely on data mining to seek better treatment and medical research;national security authorities to safeguard national security by digging criminal records. Withthe wide application of data mining, privacy issues draw more and more attention.Therefore, the privacy preserving data mining (PPDM) has become a new research directionwith new methods emerging continuously.The present study describes how to build a highquality mining model without the leakage of the privacy.With research of the existing data mining methods of privacy protection, it is foundthat a major challenge of privacy protection methods is to balance privacy protection anddata availability which usually is a pair of contradictory factors. Privacy preserving datamining needs to balance the degree of privacy protection and data availability.Many existing methods have many deficiencies, such as high maintenance costs, hugecommunication amount, low data availability and poor anti-attack capability. In order tosolve the above problems, this thesis presents ART (Axis Rotational Based Transformation)method which efficiently solves the leakage of distributed data. Therefore, this paperfocuses on maximizing data privacy protection degrees on the basis of data availability. Inaddition, the algorithm also reduces the communication cost and computation amount in adistributed environment into. This paper also proposes a modified ART method to settle thecollusion attack in a distributed environment.Finally, the theory and the experiment showed that the ART method and the modifiedART method are effective to ensure the availability of data and to improve the capability ofanti-attack.