Secure Anonymization for Incremental Datasets

Abstract

Abstract. Data anonymization techniques based on the k-anonymity model have been the focus of intense research in the last few years. Although the k-anonymity model and the related techniques provide valuable solutions to data privacy, current solutions are limited only to static data release (i.e., the entire dataset is assumed to be available at the time of release). While this may be acceptable in some applications, today we see databases continuously growing everyday and even every hour. In such dynamic environments, the current techniques may suffer from poor data quality and/or vulnerability to inference. In this paper, we analyze various inference channels that may exist in multiple anonymized datasets and discuss how to avoid such inferences. We then present an approach to securely anonymizing a continuously growing dataset in an efficient manner while assuring high data quality.

Non-Centralized Distinct L-Diversity

“…These research works can be classified into serial data publishing on a dynamic dataset and serial data publishing on a static dataset. Most of the works focus on the cases for a dynamic dataset (or multiple instances of data), which includes an incremental dataset [4][5] (i.e. allowing insertions of new tuples), a dataset with insertions and deletions [6], a dataset with insertions, deletions, and updates [7], a dataset with sensitive values that change over time [8].…”

“…can be expressed as an expression containing terms that can be calculated by simple counting on the published tables (such an expression is the general case for Equation (4). By the definition of conditional probability, it can be expressed as…”

“…Recently, many researchers have shown that if there are multiple published tables created from a centralized dataset (which can be either dynamic or static), privacy may not be preserved even if every published table satisfies a certain privacy preserving requirement individually [4][5] [6][7] [8]. This problem is called serial data publishing (on a centralized dataset).…”

An Attack on the Privacy of Sanitized Data that Fuses the Outputs of Multiple Data Miners

“…Fusion techniques have also been considered in privacypreserving data re-publishing, and for analyzing knowledge merged from sequential releases of data [15], [16]. To the best of our knowledge, fusion of data mining results has not been considered before in data privacy research.…”

Anonymizing Temporal Data

“…Motivations In this paper, we consider temporal data of the form D [1], D [2], …, where each D[i] is the snapshot containing the records collected at the time instance i. Examples of temporal data include sensor data, Internet traffic, financial tickers, web logs, fraudulent credit card transaction detection, network intrusion detection, financial record auditing, and telephone call records, inpatient discharge data, criminal reports, and population based disease monitoring data.…”