基于k匿名数据集的鲁棒性水印技术研究

doi:10.3969/j.issn.1671-1122.2022.09.002

摘要/Abstract

摘要：

在大数据时代，安全可控的数据发布变得越来越重要。数据持有者在发布数据前，出于保护用户隐私的目的，通常对数据集进行k匿名处理;而出于对数据版权的保护，则需要对数据集添加水印。因此，在k匿名数据集上嵌入水印信息具有现实意义。文章以在k匿名数据集上嵌入水印为研究目标，针对k匿名数据集缺少主键和具有受限的水印空间的问题，提出了一种基于k匿名数据集的鲁棒性水印方案，方案使用准标识符属性替代主键属性作为水印定位函数的种子，在k匿名数据集中的非敏感属性上嵌入水印信息，并在水印检测阶段采用两次多数投票机制纠正水印错误。该方案在不影响k匿名隐私目标实现的前提下，实现了数据版权信息嵌入，达到了数据隐私和数据版权的双重保护。实验证明，文章提出的水印方案具有良好的鲁棒性和执行效率。

关键词: 数据库水印, 版权保护, k匿名, 隐私保护

Abstract:

In the era of big data, secure and controlled data publishing becomes increasingly vital. When data holders publish dataset to data demanders, data holders often anonymize user’s data by k-anonymity for privacy purpose and embed watermarking in published dataset for protecting data copyright. Hence, there is a realistic demand for watermarking k-anonymity dataset. The main purpose of this paper is to embed watermark in k-anonymity dataset. However, there are two important problems for k-anonymity dataset to be addressed: the lack of primary key and the limited watermark space. In this paper, we try to address above problems by proposing a robust watermarking scheme based on k-anonymity dataset. This scheme used quasi-identifier attribute as the seed of watermark location function instead of primary key, embedded watermark information on non-sensitive attributes, and corrected error by twice majority voting in watermark detection phase. This scheme did not affect the effect of k-anonymity and realize the dual protection of privacy and copyright. Experiments showed that the watermarking scheme proposed in this paper has good robustness and efficiency.

Key words: database watermarking, copyright protection, k-anonymity, privacy protection

中图分类号:

TP309

于晶, 袁曙光, 袁煜琳, 陈驰. 基于k匿名数据集的鲁棒性水印技术研究[J]. 信息网络安全, 2022, 22(9): 11-20.

YU Jing, YUAN Shuguang, YUAN Yulin, CHEN Chi. A Robust Watermarking Technology Based on k-Anonymity Dataset[J]. Netinfo Security, 2022, 22(9): 11-20.

图/表 11

表1

表2

表3

图1

图2

表4

表5

表6

图3

表7

图4

参考文献 20

[1]	ZHOU Xiang, HUANG Min, PENG Zhiyong. An Additive-Attack-Proof Watermarking Mechanism for Databases’ Copyrights Protection Using Image[C]// ACM. Proceedings of the 2007 ACM Symposium on Applied Computing. New York: ACM, 2007: 254-258.
[2]	WANG Haiqing, CUI Xinchun, CAO Zaihui. A Speech Based Algorithm for Watermarking Relational Databases[C]// IEEE. 2008 International Symposiums on Information Processing. New York: IEEE, 2008: 603-606.
[3]	SION R, ATALLAH M, PRABHAKAR S. Rights Protection for Relational Data[J]. IEEE Trans. Knowl. Data Eng. 2004, 16(12): 1509-1525.
[4]	SHEHAB M, BERTINO E, GHAFOOR A. Watermarking Relational Databases Using Optimization-Based Techniques[J]. IEEE Trans. Knowl. Data Eng., 2007, 20(1): 116-129.
[5]	FRANCO-CONTRERAS J, COATRIEUX G. Robust Watermarking of Relational Databases with Ontology-Guided Distortion Control[J]. IEEE Trans. Inf. Forensics Secur., 2015, 10(9): 1939-1952.
[6]	CUI Xinchun, QIN Xiaolin, SHENG Gang. A Weighted Algorithm for Watermarking Relational Databases[J]. Wuhan University Journal of Natural Sciences, 2007, 12(1): 79-82.
[7]	SCHRITTWIESER S, KIESEBERG P, ECHIZEN I, et al. Using Generalization Patterns for Fingerprinting Sets of Partially Anonymized Microdata in the Course of Disasters[C]// IEEE. Sixth International Conference on Availability, Reliability and Security. New York: IEEE, 2011: 645-649.
[8]	SCHRITTWIESER S, KIESEBERG P, ECHIZEN I, et al. An Algorithm for k-anonymity-based Fingerprinting[C]// Springer. Digital Forensics and Watermarking-10th International Workshop. Berlin: Springer, 2011: 439-452.
[9]	KIESEBERG P, SCHRITTWIESER S, MULAZZANI M, et al. An Algorithm for Collusion-Resistant Anonymization and Fingerprinting of Sensitive Microdata[J]. Electronic Markets, 2014, 24(2): 113-124.
[10]	GAMBS S, LOLIVE J, ROBERT J. Entwining Sanitization and Personalization on Databases[C]// ACM. Proceedings of the 2018 on Asia Conference on Computer and Communications Security. New York: ACM, 2018: 207-219.
[11]	AGRAWAL R, KIERNAN J. Watermarking Relational Databases[C]// VLDB. Proceedings of 28th International Conference on Very Large Data Bases. CA: Morgan Kaufmann, 2002: 155-166.
[12]	AGRAWAL R, HAAS P J, KIERNAN J. Watermarking Relational Data: Framework, Algorithms and Analysis[J]. The VLDB Journal, 2003, 12(2): 157-169.
[13]	GUO Fei, WANGJianmin, LIDeyi. Fingerprinting Relational Databases[C]// ACM. Proceedings of the 2006 ACM Symposium on Applied Computing. New York: ACM, 2006: 487-492.
[14]	SAMARATI P. Protecting Respondents Identities in Microdata Release[J]. IEEE transactions on Knowledge and Data Engineering, 2001, 13(6): 1010-1027.
[15]	SWEENEY L. Achieving k-Anonymity Privacy Protection Using Generalization and Suppression[J]. International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2002, 10(5): 571-588.
[16]	LEFEVRE K, DEWITT D J, RAMAKRISHNAN R. Incognito: Efficient Full-Domain k-Anonymity[C]// ACM. Proceedings of the 2005 ACM SIGMOD International Conference on Management of Data. New York: ACM, 2005: 49-60.
[17]	SWEENEY L. Guaranteeing Anonymity When Sharing Medical Data, the Datafly System[C]// AMIA. American Medical Informatics Association Annual Symposium. TN: AMIA, 1997: 51-55.
[18]	FUNG B C, WANG K, YU P S. Top-Down Specialization for Information and Privacy Preservation[C]// IEEE. 21st International Conference on Data Engineering (ICDE’05). New York: IEEE, 2005: 205-216.
[19]	LEFEVRE K, DEWITT D J, RAMAKRISHNAN R. Mondrian Multidimensional k-Anonymity[C]// IEEE. 22nd International Conference on Data Engineering (ICDE’06). New York: IEEE, 2006: 25-28.
[20]	WANG K, YU P S, CHAKRABORTY S. Bottom-Up Generalization: A Data Mining Solution to Privacy Protection[C]// IEEE. Proceedings of the 4th IEEE International Conference on Data Mining (ICDM 2004). New York: IEEE, 2004: 249-256.

姓名	年龄	邮编	疾病	工资	分数
Alex	24	53712	心脏病	5000	98
Beth	25	53711	心脏病	6000	87
Carl	30	53711	流感	10000	80
Ellen	30	53711	癌症	5000	92
Glen	32	53712	心脏病	4000	79
Helen	32	53713	癌症	8000	96

年龄	邮编	疾病	工资	分数
[24-32]	[53712-53713]	心脏病	5000	98
[25-30]	53711	心脏病	6000	87
[25-30]	53711	流感	10000	80
[25-30]	53711	癌症	5000	92
[24-32]	[53712-53713]	心脏病	4000	79
[24-32]	[53712-53713]	癌症	8000	96

符号	描述	符号	描述
T	原数据集	k	匿名参数
T_A，t_A	匿名数据集及元组	γ	水印密度控制参数
T_AW，t_AW	嵌入水印的匿名数据集及元组	QI₁,…,QI_p	QI属性
T'_AW	被攻击的嵌入水印的匿名数据集	θ，η	NSA属性的个数，LSB的位数
W，w_k，l	水印，第k个水印比特，水印比特序列的长度	ρ	水印检测准确率
K_w	水印密钥	τ	水印检测阈值

DOBMM	DOBYY	YEAR	RACEAX	RACEBX	RACEWX	ACEIHN	EDUCYEAR	POVERTY	INCOME
1	1975	2003	3	3	1	4	12	4	32021
2	1975	2003	3	3	1	1	11	3	3700
1	1975	2003	3	3	1	1	12	3	33300
2	1975	2003	3	3	1	4	12	4	18000
3	1975	2003	3	3	1	4	12	5	20178
……
7	1950	2003	3	3	1	1	14	3	12000
8	1950	2003	3	3	1	4	16	4	27200
5	1950	2003	3	3	1	1	9	1	9434
6	1950	2003	3	3	1	4	14	5	15
7	1950	2003	3	3	1	1	12	5	21549
……

DOBMM	DOBYY	YEAR	RACEAX	RACEBX	RACEWX	ACEIHN	EDUCYEAR	POVERTY	INCOME
1:4	1975:1976	20032005	3	3	1	4	12	4	32021
1:4	1975:1976	20032005	3	3	1	1	11	3	3700
1:4	1975:1976	20032005	3	3	1	1	12	3	33300
1:4	1975:1976	20032005	3	3	1	4	12	4	18000
1:4	1975:1976	20032005	3	3	1	4	12	5	20178
……
5:8	1950:1951	20032005	3	3	1	1	14	3	12000
5:8	1950:1951	20032005	3	3	1	4	16	4	27200
5:8	1950:1951	20032005	3	3	1	1	9	1	9434
5:8	1950:1951	20032005	3	3	1	4	14	5	15
5:8	1950:1951	20032005	3	3	1	1	12	5	21549
……