A Robust Data-obfuscation Approach for Privacy Preservation of Clustered Data

Privacy is defined as the freedom from unauthorized intrusion. The availability of personal information through online databases such as government records, medical records, and voters’ lists poses a threat to personal privacy. Intelligent search engines and data mining techniques further exacerbate the problem of privacy by simplifying access and retrieval of personal records. Data Obfuscation (DO) techniques distort data in order to hide information. One application area for DO is privacy preservation. Many data obfuscation techniques have been suggested and implemented for privacy preserving data mining applications. However, existing approaches are either not robust to privacy attacks or they do not preserve data clusters, thereby making it difficult to apply data mining techniques. The absence of a standard for measuring the privacy provided by the various data obfuscation techniques makes it hard to compare the robustness of the techniques. The main contributions of this paper are (1) to propose a data obfuscation technique called Nearest Neighbor Data Substitution (NeNDS), that has strong privacy-preserving properties and maintains data clusters, (2) to define a property called Reversibility for the categorization and comparison of data obfuscation techniques, in terms of their resilience to reverse engineering, and (3) to formally prove that cluster preserving geometric transformations, by themselves are extremely easy to reverse engineer.