Differentially Private Release of the Distribution of Clustering Coefficients across Communities

Algorithm 1

The DPLM algorithm.

Input: the graph dataset G, the privacy budget

Output: a private partition

(1) do until no change about all communities

(2)  for each node i in graph G

(3)   find the neighboring communities

(4)   compute the gains ,

(5)   remain with

(6)   select community C with probability

(7)   move node i into community C

(8)  end

(9) end

(10) aggregate the found communities to build a new weighted graph

(11) repeat lines ~ until no improvement of Q

(12) return