M-PRIVACY FOR COLLABORATIVE DATA PUBLISHING: COLLABORATIVE DATA PUBLISHING PROBLEM

Abstract

The collaborative data publishing problem for anonymizing horizontally partitioned data at multiple data providers is considered. A new type of “insider attack” by colluding data providers who may use their own data records (a subset of the overall data) in addition to the external background knowledge to infer the data records contributed by other data providers. This new threat and makes several contributions. The notion of m-privacy, which guarantees that the anonymized data satisfies a given privacy constraint against any group of up to m colluding data providers. A heuristic algorithms exploiting the equivalence group monotonicity of privacy constraints and adaptive ordering techniques for efficiently checking m-privacy given a set of records is presented. A data provider-aware anonymization algorithm is presented with adaptive m- privacy checking strategies to ensure high utility and m-privacy of anonymized data with efficiency. Experiments on real-life datasets suggest that this approach achieves better or comparable utility and efficiency than existing and baseline algorithms while providing m-privacy guarantee.

The goal is to publish an anonymized view of the integrated data such that a data recipient including the data providers will not be able to compromise the privacy of the individual records provided by other parties.

Key words: Anonymization, Adversary, Algorithm.