Differential Privacy for Edge Weights in Social Networks

Algorithm 1

Merging Barrels and Consistency Inference (MB-CI) algorithm.

Input: Raw weighted-graph database , privacy budget , parameter

Output: Disturbed weighted-graph database

// and contain three column vectors , , , and , , , respectively.

// represents the starting points of edges. indicates the ends.

// and store the original edge weights and the disturbed ones.

(1) Scan once to compute three vectors , , :

 , , .

(2)

(3)

(4) for to

(5)  if then

(6)   

(7)  else

(8)   

(9)  end if

(10)   end for

(11) if then

(12) for to

(13)  

(14) end for

(15) while

(16)  

(17)  while

(18)   if or then

(19)    

(20)   else

(21)    

(22)   end if

(23)  end while

(24)  

(25) end while

(26) for to

(27)  

(28) end for

(29) return