Research Article

A Personalized -Anonymity Model of Social Network for Protecting Privacy

Algorithm 1

Personalized -anonymity algorithm.
Inputs:
Initial anonymous graph G = (V, E),
Sensitivity parameters: k’(k’ ≥2); l(2 ≤ l ≤ k’); m(l ≤ m ≤ |V|);
Node attributes table: AS = {viS,viN(1),…,viN(s),viC(1),…,viC(t)};
Edge attributes table: AS = {vjS,vjN(1),…,vjN(s),vjC(1),…,vjC(t)};
All the classified attribute inheritance tree HC
Input parameters α =0 and β
Outputs:
Anonymous graph Gp = {g1, g2, …, ge};
The whole VCS, ECS and their attribute information;
Steps:
1 anonymous graph Gp, groups VCS and ECS are caught;
2 read α to judge the generalizing type;
3 got the group number:NVCS = |NVP|/k, NECS = |NEP|/k;
4 fori =1 to NVCSdo//QI attributes generalization
5 forj =1 to sdo//numeric type attributes generalization
6  gen(VCSi)[Nj] = [min{v1N(j), …,vkN(j)},max{v1N(j), …,vkN(j)}]
7  end for
8  forj =1 to tdo//t type QI attributes generalization
9   gen(VCSi)[Cj] = {v1C(j),…,vkC(j)}
10  end for
11  whiledo
12   if(sensitive attributes are classified) then
13    forj =0 to kdo
14     vjC is replaced by its parents node in classified inheritance tree of sensitive attribute
15     ifthen jump while loop
16    end for
17   Else
18    forj =1 to kdo
19     the interval of vjN is changed to its neighborhood;
20     ifthen jump while loop
21    end for
22   end if
23  end while
24 end for
25 fori =1 to NECSdo
26  forj =1 to pdo
27  gen(ECSi)[Nj] = [min{v1N(j),…,vkN(j)},max{v1N(j), …,vkN(j)}]
28  end for
29  forj =1 to qdo
30   gen(ECSi)[Cj] = {e1C(j),…,ekC(j)}
31  end for
32  whiledo
33   if(sensitive attributes are classified) then
34    forj =1 to kdo
35     vjC is replaced by its parents node in classified inheritance tree of sensitive attribute
36     ifthen jump while loop
37    end for
38   Else
39    forj =1 to kdo
40     the interval of njN is changed to its neighborhood;
41     ifthen jump while loop
42    end for
43   end if
44  end while
45 end for
46 anonymous graph Gp is published; all the VCS nodes, ECS edges and their attribute information are published