On Learning Cluster Coefficient of Private Networks

Provided by: University of North Carolina at Chapel Hill (Kenan-Flagler) Topic: Security Date Added: Jun 2012 Format: PDF
Enabling accurate analysis of social network data while preserving differential privacy has been challenging since graph features such as clustering coefficient or modularity often have high sensitivity, which is different from traditional aggregate functions (e.g., count and sum) on tabular data. In this paper, the authors treat a graph statistics as a function f and develop a divide and conquer approach to enforce differential privacy. The basic procedure of this approach is to first decompose the target computation f into several less complex unit computations f1, fm connected by basic mathematical operations (e.g., addition, subtraction, multiplication, division), then perturb the output of each fi with laplace noise derived from its own sensitivity value.

Find By Topic