A Study of Data Perturbation in Outsourced Databases for Preserving Privacy

• Main Authors: Li-Cheng Yang, 楊立誠
• Other Authors: 林詠章
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/03532796162135693753

碩士 === 國立中興大學 === 資訊管理學系所 === 102 === The cloud computing technique rises in these years. The concept of database as a service has been proposed. The inside data in the organization may increase rapidly with time. In order to reduce cost of organization, they may chose third-party storage provider to store entire data. There is a leakage crisis when provider is untrusted. For another instance, a dealer collects all transaction data and publishes to the data analysis company for marketing purpose. It may reveal privacy when the company is malicious. For these reason, preserving privacy in database becomes very important. However, it is hard to address since database security is wide issues. This paper only concerns prediction disclosure risk in numerical database. We present an efficient noise generation which relies on Huffman coding algorithm, and also consider occurrence probability of records. We also build a noise matrix that can inject intuitively noise to original value. Moreover, we adopt clustering technique before generating noise to enhance the speed of the process. In the experiment, we examine the running time of noise generation and also examine the data quality after replaced value. The experimental results show the running time of noise generation of clustering scheme is the fastest. And information loss for all data is not exceeding to 50 percentages. Finally, we conclude our proposed scheme and discuss future works for this study.