LDPCD: A Novel Method for Locally Differentially Private Community Detection
Algorithm 1
The implementation of truncated Laplace mechanism on the user side
Input: user i’s adjacent bit vector , the division of the group where user i is located , the truncated interval length given by the server , privacy budget
Output: perturbed degree vector of user i under the grouping
(1)
Calculate the true degree vector based on and ;
(2)
Generate and corresponding to and , respectively, according to , which is and ;
(3)
Find and in and , respectively, to satisfy that and ;
(4)
Set or , or , or , respectively;
(5)
if and then
(6)
Randomly chose a truncated range between and (let ) with equal probability, then perturb within it;
(7)
With , calculate and according to equations (7)–(10);
(8)
Calculate ,;
(9)
for do
(10)
Calculate ;
(11)
Sample a value according to the output probability of each integer in and assign it to the perturbed degree ;