Asma Alnemari

Abstract

Preserving privacy while publishing data for analysis by researchers is an issue which has considerable attention recently. ε-differential privacy is a solution that guarantees the privacy while publishing sets of data or some information about them. This work aimed to develop a mechanism that answers a given workload of range queries efficiently under differential privacy. Therefore, an algorithm was implemented during this project to satisfy differential privacy by controlling the noise added to the answers according to the input data and the given workload. The algorithm first produces two partitions of the data domain. The first partition will be produced according to the relationships between the input data while the second partition will be produced according to the ranges of the given queries. When the domain is partitioned into buckets, the counts of each bucket are calculated privately and split among the vector’s positions to answer the given query set. The performances of the proposed mechanisms were evaluated using different workloads over different attributes, and the algorithm produces satisfactory results in most cases.

Library of Congress Subject Headings

Data protection; Database security; Data encryption (Computer science)

Publication Date

6-2016

Document Type

Thesis

Student Type

Graduate

Degree Name

Computer Science (MS)

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Rajendra K. Raj

Advisor/Committee Member

Carol Romanowski

Advisor/Committee Member

Zack Butler

Comments

Physical copy available from RIT's Wallace Library at QA76.9.A25 A56 2016

Campus

RIT – Main Campus

Plan Codes

COMPSCI-MS

Download

Share

COinS