Privacy Characterization and Quantification in Data Publishing

Aim:  propose a novel multi-variable privacy characterization and quantification model.

 Existing system: Privacy-Preserving Data Publishing (PPDP) techniques (k-anonymity, l-diversity and t-closeness.) have been proposed in literature. However, they lack a proper privacy characterization and measurement. all the existing PPDP schemes have limitations in privacy characterization

 Proposed System: PUBLISHING MODEL goals is to quantify privacy loss of individuals having other attribute values (other diseases) within the same class. PRIVACY CHARACTERIZATION proposes two privacy metrics that can measure privacy leakage from two different perspectives.

 Methodology:

The proposed framework consists of two steps. First, model attributes in a dataset as a multi-variable model. Based on first model, are able to re-define the prior and posterior adversarial belief about attribute values of individuals.  Then characterize privacy of these individuals based on the privacy risks attached with combining different attributes. Model is indeed a more precise model to describe privacy risk of publishing datasets.

·         PUBLISHING MODEL AND PRIVACY CHARACTERIZATION:

To obtain a meaningful definition of data privacy, it is necessary to characterize and quantify the knowledge about sensitive attributes that the adversary gains from observing the published dataset taking into consideration the combinational relation of different attributes. to characterize privacy, employ a multi-dimensional scheme of privacy risk analysis attached with combining different attributes.

·         PRIVACY QUANTIFICATION:

a new set of privacy quantification metrics to measure the gap between prior information belief and posterior information belief of an adversary, from both local and global perspectives. Specifically, we introduce two privacy leakage measurements: distribution leakage and entropy leakage.

Future Enhancement:

Optimization of the chosen set of quasi-identifiers with an objective of minimizing distribution and entropy leakages within the published table or specific classes of higher privacy concerns.

Conclusion:

Introduced comprehensive characterization and novel quantification methods of privacy to deal with the problem of privacy quantification in privacy preserving data publishing. In order to consider the privacy loss of combined attributes, presented data publishing as a multi-relational model. re-defined the prior and posterior beliefs of the adversary. The proposed model and adversarial beliefs contribute to a more precise privacy characterization and quantification.

For additional details comment below with requirements.