Web based Privacy Disclosure Threats and Control Techniques

Web based Privacy Disclosure Threats and Control Techniques

Tithi Hunka (KIIT University, India), Sital Dash (KIIT University, India) and Prasant Kumar Pattnaik (KIIT University, India)
Copyright: © 2016 |Pages: 8
DOI: 10.4018/978-1-4666-9764-5.ch014
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Due to advancement of internet technologies, web based applications are gaining popularity day by day. Many organizations maintain large volumes of web site based data about individuals that may carry information that cannot be revealed to the public or researchers. While web-based applications are becoming increasingly pervasive by nature, they also present new security and privacy challenges. However, privacy threats effects negatively on sensitive data and possibly leads to the leakage of confidential information. More ever, privacy preserving data mining techniques allow us to protect the sensitive data before it gets published to the public by changing the original micro-data format and contents. This chapter is intended to undertake an extensive study on some ramified disclosure threats to the privacy and PPDM (privacy preserving data mining) techniques as a unified solution to protect against threats.
Chapter Preview
Top

Privacy1

The American Institute of Certified Public Accountants (AICPA) and Canadian Institute of Charted Accountants (CICA) define that, “Privacy is the right and the obligation of individuals and organizations with respect to the collection, use, retention, and disclosure of personal information” (Joseph, Daniel, & Vasanthi, 2013, p. 1). Privacy aims to less risk of disclosure of confidential value in a micro-data file, less information loss and high utility of micro-data for analysis purpose.

Privacy Disclosure Threats

There can be four types of privacy disclosure threats commonly identified while publishing the micro-data 1) Identity Disclosure 2) Membership Disclosure 3) Attribute Disclosure 4) Statistical Disclosure. A novel technique should protect against all the disclosure while maintaining the trade-off between data utility, information loss and risk of disclosure.

Table 1.
Control techniques for Privacy Disclosure threats
Serial No.Privacy Disclosure ThreatDescriptionPrivacy Disclosure Control Techniques
1Identity Disclosure (Templ, Meindl, & Kowarik, 2013)Identity disclosure occurs when adversaries are able to recognize a particular person in the released data set.• k-anonymity
• l-diversity
• t-closeness with k-anonymity
• slicing
2Statistical Disclosure
(Templ, Meindl, & Kowarik, 2013)
In this, if adversaries are able to estimate the confidential data from the released statics than it is said that statistical disclosure has been taking place. For disclosure control either data are modified (Perturbation of data) or reduced (broad banding of data) to an acceptable level. The method chosen depends upon the data to be released.• Global Recoding
• Local Suppression
• Top/Bottom Coding
• Adding Noise
• Swapping
• PRAM(Post Randomization method)
• Micro aggregation
• Resampling
3Membership Disclosure (Templ, Meindl, & Kowarik, 2013)In a specific database (such as dataset containing cancer patients) If someone is unable to decide whether the record of any individual is present in the dataset or not, then the dataset is free from membership disclosure. In some cases it is better to use the identity disclosure control technique when the adversary is unknown about the membership of individuals, and in cases when adversary knows the individual’s record than membership disclosure control technique is not sufficient.• Slicing
4Attribute Disclosure (Templ, Meindl, & Kowarik, 2013)Attribute Disclosure occurs, when any additional information about an individual is revealed from the released data set. Identity Disclosure leads to Attribute Disclosure .Attribute Disclosure can occur with or without identity disclosure (If someone knows the identity of a person in a dataset, he can easily find out the sensitive information or for all matching tuples there are the same sensitive attribute).• Randomization
• Bucketization
• l-diversity
• Slicing

Complete Chapter List

Search this Book:
Reset