Reproduce “SoK: Practical Aspects of Releasing Differentially Private Graphs” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nicholas D'Silva, Surya Nepal, Salil S. Kanhere
A structured systemisation and a practical framework were developed for evaluating differentially private graph release mechanisms, revealing that many edge-DP methods exhibit empirical privacy loss despite theoretical guarantees and that current node-DP approaches offer inadequate utility for real-world applications.
Key insights
- Edge-DP mechanisms often fail to mask global and multi-hop structural correlations, leading to significant empirical privacy loss despite theoretical guarantees, due to unaddressed data independence assumptions.
- Current node-DP methods, designed to protect individual entities, largely provide insufficient utility for practical applications, producing severely distorted graph structures.
- The research landscape is heavily skewed towards centralised trust models and edge-DP, neglecting more complex graph types (temporal, attributed) and advanced dependency-aware privacy definitions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.