Satisfying privacy requirements : one step before anonymization

Date

2010

Authors

Sun, X.
Wang, H.
Li, J.

Editors

Zaki, M.J.
Yu, J.X.
Ravindran, B.
Pudi, V.

Advisors

Journal Title

Journal ISSN

Volume Title

Type:

Conference paper

Citation

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2010 / Zaki, M.J., Yu, J.X., Ravindran, B., Pudi, V. (ed./s), vol.6118 LNAI, iss.PART 1, pp.181-editor

Statement of Responsibility

Conference Name

14th Pacific Asia Conference on Knowledge Discovery and Data Mining (21 Jun 2010 - 24 Jun 2010 : Hyderabad, India)

Abstract

In this paper, we study a problem of privacy protection in large survey rating data. The rating data usually contains both ratings of sensitive and non-sensitive issues, and the ratings of sensitive issues include personal information. Even when survey participants do not reveal any of their ratings, their survey records are potentially identifiable by using information from other public sources. We propose a new (k,ε,l)-anonymity model, in which each record is required to be similar with at least k − 1 others based on the non-sensitive ratings, where the similarity is controlled by ε, and the standard deviation of sensitive ratings is at least l. We study an interesting yet nontrivial satisfaction problem of the (k,ε,l)-anonymity, which is to decide whether a survey rating data set satisfies the privacy requirements given by users. We develop a slice technique for the satisfaction problem and the experimental results show that the slicing technique is fast, scalable and much more efficient in terms of execution time than the heuristic pairwise method.

School/Discipline

Dissertation Note

Provenance

Description

Access Status

Rights

Copyright 2010 Springer-Verlag Berlin Heidelberg

License

Call number

Persistent link to this record