Reza Mortazavi

Due to recent advances in data collection and processing, data publishing has emerged by some organizations for scientific and commercial purposes. Published data should be anonymized such that staying useful while the privacy of data respondents is preserved. Microaggregation is a popular mechanism for data anonymization, but naturally operates on numerical datasets. However, the type of data in the real world is usually mixed i.e., there are both numeric and categorical attributes together. In this paper, we propose a novel transformation based method for microaggregation of mixed data called TBM. The method uses multidimensional scaling to generate a numeric equivalent from mixed dataset. The partitioning step of microaggregation is performed on the equivalent dataset but the aggregation step on the original data. TBM can microaggregate large mixed datasets in a short time with low information loss. Experimental results show that the proposed method attains better trade-off between data utility and privacy in a shorter time in comparison with the traditional methods. © 2016, The Author(s).

Microaggregation is known as a successful perturbative mechanism to realize k-anonymity. The method partitions the dataset into groups of at least k members and then aggregates the group members. These aggregated values are published instead of the original ones. In conventional microaggregation methods, it is desired to produce a protected dataset similar to the original one, so close data records are grouped into the same cluster. Accordingly, the aggregation phase of the algorithms are designed to minimize the sum of within-group squared error (SSE), and therefore a simple arithmetic mean in each group is utilized within the aggregation phase to compute the centroids. However, this trivial approach does not consider the proximity of the published values to the original ones, so intruders are able to limit the range of the original values with respect to published data. In this paper, a proximity-aware microaggregation post-processing algorithm is proposed that revisits the aggregation step to remedy this deficiency. Additionally, it is possible to consider different levels of minimum required distances between original record values and their corresponding published ones. Empirical results confirm the superiority of the proposed method in achieving a better trade-off point between disclosure risk and information loss in comparison with similar microaggregation techniques. © 2017, University of Skovde. All rights reserved.

Microaggregation is a masking mechanism to protect confidential data in a public release. This technique can produce a k-anonymous dataset where data records are partitioned into groups of at least k members. In each group, a representative centroid is computed by aggregating the group members and is published instead of the original records. In a conventional microaggregation algorithm, the centroids are computed based on simple arithmetic mean of group members. This naïve formulation does not consider the proximity of the published values to the original ones, so an intruder may be able to guess the original values. This paper proposes a disclosure-aware aggregation model, where published values are computed in a given distance from the original ones to attain a more protected and useful published dataset. Empirical results show the superiority of the proposed method in achieving a better trade-off point between disclosure risk and information loss in comparison with other similar anonymization techniques. © 2015, The Author(s).