scottshufe / hyobscure

The code repository for paper HyObscure.

HyObscure

Minimizing privacy leakage and ensuring data utility is a critical problem to social media platforms (SMPs) for data publishing, namely privacy preserving data publishing problem. Due to data correlations, it is challenging to achieve an optimal privacy-utility tradeoff with hybrid obscuring on heterogeneous data. HyObscure is a novel hybrid protection mechanism focusing on heterogeneous data publishing on SMPs, which cross-iteratively optimizes the generalization and obfuscation operations to maximize privacy protection under a certain utility guarantee.

Software Environment

• MATLAB and CVX package with Mosek solver
• Python3.6. All necessary packages are listed in file requirements.txt.
• MATLAB Engine API for Python

Python Packages Installation

pip install -r requirements.txt

MATLAB Engine API for Python Installation

See https://www.mathworks.com/help/matlab/matlab_external/install-the-matlab-engine-for-python.html for details.