Searching for better randomized response schemes for privacy-preserving data mining

Zhengli Huang; Wenliang Du; Zhouxuan Teng

doi:10.1007/978-3-540-74976-9_50

Searching for better randomized response schemes for privacy-preserving data mining

Zhengli Huang, Wenliang Du, Zhouxuan Teng

Department of Electrical Engineering & Computer Science

Research output: Chapter in Book/Entry/Poem › Conference contribution

2 Scopus citations

Abstract

To preserve user privacy in Privacy-Preserving Data Mining (PPDM), the randomized response (RR) technique is widely used for categorical data. Although various RR schemes have been proposed, there is no study to systematically compare them in order to find optimal RR schemes. In the paper, we choose the R-U (Risk-Utility) confidentiality map to compare different randomization schemes. Using the R-U map as our metric, we present an optimal RR scheme for binary data, which helps us find an optimal class of RR matrices. From this optimal scheme, we have discovered several heuristic rules among the elements in the optimal class. We generalize these rules to find optimal class of RR matrices for categorical data. Based on these rules, we propose an RR scheme to find a class of RR matrices for categorical data. Our experimental results have shown that our scheme has much better performance than the existing RR schemes.

Original language	English (US)
Title of host publication	Knowledge Discovery in Database
Subtitle of host publication	PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings
Publisher	Springer Verlag
Pages	487-497
Number of pages	11
ISBN (Print)	9783540749752
DOIs	https://doi.org/10.1007/978-3-540-74976-9_50
State	Published - 2007
Event	11th European Conference on Principles and Practice of Knowledge Discovery in Databases, PKDD 2007 - Warsaw, Poland Duration: Sep 17 2007 → Sep 21 2007

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	4702 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Other

Other	11th European Conference on Principles and Practice of Knowledge Discovery in Databases, PKDD 2007
Country/Territory	Poland
City	Warsaw
Period	9/17/07 → 9/21/07

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-540-74976-9_50

Cite this

Huang, Z., Du, W., & Teng, Z. (2007). Searching for better randomized response schemes for privacy-preserving data mining. In Knowledge Discovery in Database: PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings (pp. 487-497). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4702 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-540-74976-9_50

Searching for better randomized response schemes for privacy-preserving data mining. / Huang, Zhengli; Du, Wenliang; Teng, Zhouxuan.
Knowledge Discovery in Database: PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings. Springer Verlag, 2007. p. 487-497 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4702 LNAI).

Research output: Chapter in Book/Entry/Poem › Conference contribution

Huang, Z, Du, W & Teng, Z 2007, Searching for better randomized response schemes for privacy-preserving data mining. in Knowledge Discovery in Database: PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4702 LNAI, Springer Verlag, pp. 487-497, 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, PKDD 2007, Warsaw, Poland, 9/17/07. https://doi.org/10.1007/978-3-540-74976-9_50

Huang Z, Du W, Teng Z. Searching for better randomized response schemes for privacy-preserving data mining. In Knowledge Discovery in Database: PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings. Springer Verlag. 2007. p. 487-497. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-540-74976-9_50

Huang, Zhengli ; Du, Wenliang ; Teng, Zhouxuan. / Searching for better randomized response schemes for privacy-preserving data mining. Knowledge Discovery in Database: PKDD 2007 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, Proceedings. Springer Verlag, 2007. pp. 487-497 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{0c3fa044c85b4486b6c2fd3b3dc71682,

title = "Searching for better randomized response schemes for privacy-preserving data mining",

abstract = "To preserve user privacy in Privacy-Preserving Data Mining (PPDM), the randomized response (RR) technique is widely used for categorical data. Although various RR schemes have been proposed, there is no study to systematically compare them in order to find optimal RR schemes. In the paper, we choose the R-U (Risk-Utility) confidentiality map to compare different randomization schemes. Using the R-U map as our metric, we present an optimal RR scheme for binary data, which helps us find an optimal class of RR matrices. From this optimal scheme, we have discovered several heuristic rules among the elements in the optimal class. We generalize these rules to find optimal class of RR matrices for categorical data. Based on these rules, we propose an RR scheme to find a class of RR matrices for categorical data. Our experimental results have shown that our scheme has much better performance than the existing RR schemes.",

author = "Zhengli Huang and Wenliang Du and Zhouxuan Teng",

year = "2007",

doi = "10.1007/978-3-540-74976-9_50",

language = "English (US)",

isbn = "9783540749752",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "487--497",

booktitle = "Knowledge Discovery in Database",

note = "11th European Conference on Principles and Practice of Knowledge Discovery in Databases, PKDD 2007 ; Conference date: 17-09-2007 Through 21-09-2007",

}

TY - GEN

T1 - Searching for better randomized response schemes for privacy-preserving data mining

AU - Huang, Zhengli

AU - Du, Wenliang

AU - Teng, Zhouxuan

PY - 2007

Y1 - 2007

N2 - To preserve user privacy in Privacy-Preserving Data Mining (PPDM), the randomized response (RR) technique is widely used for categorical data. Although various RR schemes have been proposed, there is no study to systematically compare them in order to find optimal RR schemes. In the paper, we choose the R-U (Risk-Utility) confidentiality map to compare different randomization schemes. Using the R-U map as our metric, we present an optimal RR scheme for binary data, which helps us find an optimal class of RR matrices. From this optimal scheme, we have discovered several heuristic rules among the elements in the optimal class. We generalize these rules to find optimal class of RR matrices for categorical data. Based on these rules, we propose an RR scheme to find a class of RR matrices for categorical data. Our experimental results have shown that our scheme has much better performance than the existing RR schemes.

AB - To preserve user privacy in Privacy-Preserving Data Mining (PPDM), the randomized response (RR) technique is widely used for categorical data. Although various RR schemes have been proposed, there is no study to systematically compare them in order to find optimal RR schemes. In the paper, we choose the R-U (Risk-Utility) confidentiality map to compare different randomization schemes. Using the R-U map as our metric, we present an optimal RR scheme for binary data, which helps us find an optimal class of RR matrices. From this optimal scheme, we have discovered several heuristic rules among the elements in the optimal class. We generalize these rules to find optimal class of RR matrices for categorical data. Based on these rules, we propose an RR scheme to find a class of RR matrices for categorical data. Our experimental results have shown that our scheme has much better performance than the existing RR schemes.

UR - http://www.scopus.com/inward/record.url?scp=38049173463&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38049173463&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-74976-9_50

DO - 10.1007/978-3-540-74976-9_50

M3 - Conference contribution

AN - SCOPUS:38049173463

SN - 9783540749752

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 487

EP - 497

BT - Knowledge Discovery in Database

PB - Springer Verlag

T2 - 11th European Conference on Principles and Practice of Knowledge Discovery in Databases, PKDD 2007

Y2 - 17 September 2007 through 21 September 2007

ER -

Searching for better randomized response schemes for privacy-preserving data mining

Abstract

Publication series

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this