TA_DA: Target-Aware Data Anonymization

More and more scenarios rely today on data analysis of massive amount of data, possibly contributed from multiple parties (data controllers). Data may, however, contain information that is sensitive, company-confidential, or that should be protected (e.g., since it exposes identities of the data subjects) and cannot simply be freely shared and used for analysis. Business rules, restrictions from individuals (data subjects to which data refer), as well as privacy regulations demand data to be sanitized before being released or shared with others. Unfortunately, such protection typically comes with a loss of utility of the released data, for analytics tasks operating on the data. In this paper, we present $\mathsf{TA\_DA}$, a target-aware data anonymization approach that aims at protecting (anonymizing) data while preserving as much as possible the utility of the released data for the data analytics task operating downstream. Our approach does not replace anonymization solutions, it operates prior to anonymization catering data for the anonymization process. The idea is to partition data in groups (constructed based on the data analytics task to which the data are fed) on which anonymization then operates. With anonymization achieved through data generalization (to provide $k$-anonymity and $\ell$-diversity guarantees), the goal of having anonymization operate independently on groups is to limit the impact of anonymization on the attributes and values that should be preserved for the downstream data analytics task. Our experimental evaluation confirms the effectiveness of our approach.