Skip to content

Privacy Metrics Engine of the PriEval-Protect Framework, responsible for computing data-level privacy risks using a suite of statistical, probabilistic, and information-theoretic metrics.

License

Notifications You must be signed in to change notification settings

Chebil-Ilef/Privacy-Engine

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

33 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PriEval-Protect: Privacy Metrics Engine

This repository contains the Privacy Metrics Engine of the PriEval-Protect Framework, responsible for computing data-level privacy risks using a suite of statistical, probabilistic, and information-theoretic metrics.

Overview

This module performs privacy risk evaluation on structured datasets, identifying potential re-identification threats, uncertainty, and data disclosure risks. It supports automated classification of dataset attributes into:

  • Quasi-identifiers (QIDs)
  • Sensitive Attributes (SAs)
  • Non-sensitive attributes (NSAs)

It is designed for use in e-health privacy audits and compliance scoring pipelines.

Key Features

  • Attribute classification using algorithmic and LLM-based techniques
  • Risk estimation: k-anonymity, l-diversity, entropy, mutual information
  • Adversary success likelihood & delta presence analysis
  • Modular utils for integration with scoring or compliance modules
  • FastAPI-compatible

Project Structure

Privacy-Engine/
├── data/
│   └── test/                            # Sample test datasets
│
├── src/
│   ├── utils/                           # Core metric calculators
│   │   ├── adversary_success.py         # Adversary success and delta presence
│   │   ├── algorithmic_attribute_classification.py
│   │   ├── data_similarity.py           # k-anonymity, l-diversity
│   │   ├── info_gain_loss.py            # Mutual information, entropy
│   │   ├── llm_attribute_classification.py
│   │   └── uncertainty.py               # Entropy metrics
│   └── main.py                   # Entry point for metric pipeline
│
├── .env.example                  # Environment variable template
├── .gitignore
├── README.md
└── requirements.txt              # Python dependencies

Getting Started

  1. Clone the repository:
git clone https://github.com/PriEval-Protect-Framework/Privacy-Engine.git
cd Privacy-Engine
  1. Install dependencies:
pip install -r requirements.txt
  1. Run :
uvicorn src.main:app --reload 

Privacy Metrics Computed

  • k-Anonymity, l-Diversity, α-k Anonymity
  • Mutual Information, Conditional Entropy
  • Delta Presence, Adversary Success Rate
  • Normalized Shannon Entropy
  • Risk-Based Attribute Classification (QID, SA, NSA)

License

This module is licensed under the MIT License.

Authors

Developed by Ilef Chebil and Asma ElHadj Supervised by [EFREI Paris] and [INSAT Tunisia]

Related Modules

About

Privacy Metrics Engine of the PriEval-Protect Framework, responsible for computing data-level privacy risks using a suite of statistical, probabilistic, and information-theoretic metrics.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%