PKGem: Secure Enrichment of Personal Knowledge Graphs

Source code for PKGem, a system that provides an end-to-end secure solution to enrich personal knowledge graphs in mobile environments. The system is implemented as an Android application and supports a variety of real-world usage scenarios.

The paper is accepted by CIKM '25, the video is available at Youtube.

Dependencies

System Requirements

• Android SDK: API level 26 (Android 8.0) or higher
• Target SDK: API level 34 (Android 14)
• Gradle: Version 8.12 (included in project)
• Kotlin: Version 1.8.0
• Python: Version 3.8 (for Chaquopy integration)

Key Dependencies

• Chaquopy: Python runtime for Android (version 16.1.0)
• TenSEAL: Homomorphic encryption library (version 0.3.15)
• PyTorch: Machine learning framework (version 2.0.1)
• Sentence Transformers: Text embedding models (version 4.1.0)
• ONNX Runtime: Model inference optimization

Python Packages

The following Python packages are automatically installed via Chaquopy:

• numpy==1.24.4
• scipy==1.15.3
• sentence_transformers==4.1.0
• tenseal==0.3.16
• torch==2.0.1
• torch_xla==2.7.0
• tqdm==4.66.1
• transformers==4.30.2
• psutil==5.9.5

Installation

Prerequisites

1. Install Android Studio
2. Install Android SDK with API level 26+ and target API level 34
3. Install Android NDK
4. Ensure you have at least 2GB of available RAM

Clone Repository

git clone https://github.com/golden-eggs-lab/pkgem.git
cd pkgem

Acknowledgement and License

PKGem is built on top of TenSEAL, and the Python components are wrapped by Chaquopy.

MIT LICENSE

Questions / Help / Bug Reports

If you encounter any problems while using PKGem and need our help, please click here to report the problem.