Privacy utility trade off tool
Method(s) and system(s) for providing an optimal trade off point between privacy of a private data and utility of a utility application thereof are described. The method includes quantifying privacy content of a private data associated with a user based on uniqueness of information in the private data, where the private content comprises sensitive information about the user. The method further includes determining a privacy-utility trade off point model based on analytical analysis of the privacy content, a privacy requirement of the user, and a utility requirement of third party to which the private data is disclosed, where the privacy-utility trade off point model is indicative of optimal private data sharing technique with the third party. Furthermore, the method also includes identifying privacy settings for the user based on risk appetite of the third party, utilizing the determined privacy-utility tradeoff point model.
1. A computer implemented method for determining a privacy-utility trade off, the method comprising:
quantifying privacy content of a private data associated with a user based on uniqueness of information in the private data, wherein the quantifying comprises determining a theoretic randomness and uncertainty of information in the private data of the user, and wherein the private content comprises sensitive information about the user;
determining a privacy-utility trade off point model based on analytical analysis of the privacy content of a private data associated with a user, a privacy requirement of the user, and a utility requirement of third party utilizing the private data of the user, wherein the privacy-utility trade off point model is indicative of optimal private data sharing technique with the third party for determining optimal sharing point, and wherein the optimal sharing point is a difference between the privacy content of the private data and that of a transformed private data;
identifying privacy settings for the user based on risk appetite of the user, utilizing the determined privacy-utility tradeoff point model, wherein the risk appetite of the user is indicative of the user's utility requirement and private data sharing capabilities; and
sanitizing the private data of the user by perturbation based on the identified privacy settings for the user wherein the perturbation reduces a granularity of the private data, and wherein the sanitization further comprises:
ascertaining the privacy requirement of the user, and the utility requirement of third party based on the privacy setting of the user; and
determining the optimal sharing point based on the privacy-utility trade off point model and the ascertained privacy requirement of the user,
wherein the quantifying further comprises computing Kullback-Leibler divergence to determine relative entropy between the private data of the user and a standard non-private data, wherein the divergence is indicative of the privacy content of the private data.
2. A system for determining a privacy-utility trade off, the system comprising:
a processor;
a quantification module coupled to the processor, wherein the quantification module is configured to quantify privacy content of a private data associated with a user based on uniqueness of information in the private data, wherein the quantifying further comprises determining a theoretic randomness and uncertainty of information in the private data of the user, and wherein the private content comprises sensitive information about the user;
a modeling module coupled to the processor, wherein the modeling module is configured to determine a privacy-utility trade off point model based on analytical analysis of the privacy content of a private data associated with a user, a privacy requirement of the user, and a utility requirement of third party utilizing the private data of the user, wherein the privacy-utility trade off point model is indicative of optimal private data sharing technique with the third party for determining optimal sharing point, and wherein the optimal sharing point is a difference between the privacy content of the private data and that of a transformed private data;
a privacy setting module coupled to the processor, wherein the privacy setting module is configured to identify privacy settings for the user based on risk appetite of the user, utilizing the determined privacy-utility tradeoff point model, wherein the risk appetite of the user is indicative of the user's utility requirement and private data sharing capabilities; and
a sanitization module coupled to the processor, wherein the sanitization module is configured to sanitize the private data of the user by perturbation based on the identified privacy settings for the user wherein the perturbation reduces a granularity of the private data, and wherein to sanitize the private data of the user, the sanitization module is configured to:
ascertain the privacy requirement of the user, and the utility requirement of third party based on the privacy setting of the user; and
determine the optimal sharing point based on the privacy-utility trade off point model and the ascertained privacy requirement of the user;
wherein the quantification module is further configured to compute Kullback-Leibler divergence to determine relative entropy between the private data of the user and a standard non-private data, wherein the divergence is indicative of the privacy content of the private data.
3. A non-transitory computer-readable medium having embodied thereon a computer readable program code for executing a method comprising:
quantifying privacy content of a private data associated with a user based on uniqueness of information in the private data, wherein the quantifying comprises determining a theoretic randomness and uncertainty of information in the private data of the user, and wherein the private content comprises sensitive information about the user, and further comprises computing a Kullback-Leibler divergence to determine relative entropy between the private data of the user and a standard non-private data, wherein the divergence is indicative of the privacy content of the private data;
determining a privacy-utility trade off point model based on analytical analysis of the privacy content of a private data associated with a user, a privacy requirement of the user, and a utility requirement of third party utilizing the private data of the user, wherein the privacy-utility trade off point model is indicative of optimal private data sharing technique with the third party for determining optimal sharing point, and wherein the optimal sharing point is a difference between the privacy content of the private data and that of a transformed private data;
identifying privacy settings for the user based on risk appetite of the user, utilizing the determined privacy-utility tradeoff point model, wherein the risk appetite of the user is indicative of the user's utility requirement and private data sharing capabilities; and
sanitizing the private data of the user by perturbation based on the identified privacy settings for the user wherein the perturbation reduces a granularity of the private data, and wherein the sanitization further comprises:
ascertaining the privacy requirement of the user, and the utility requirement of third party based on the privacy setting of the user; and
determining the optimal sharing point based on the privacy-utility trade off point model and the ascertained privacy requirement of the user.