Analyzing and Securing Social Networks

194 Analyzing and Securing Social Networks

be extended to a standard distributed architecture. For DC, we simply distribute the adjacency list

for each node to a separate processor. For information gain-based partitioning (IGBP), when cal-

culating the information gain for each attribute, we distribute the processing of each attribute to a

separate processor.

Algorithm 17.1: CC_Multicore()

1: for

v

i

∈

do

2:

NE

ijij

vvv

←∈

{(,)

}

|

{rst-degree neighbors of v

i

}

3: r

i

= FindPossibleTriples(

v

ii

,

)

4: end for

5: for (v

i

, v

j

, v

k

) ∈ r

i

do {H is a Hashmap}

6: IncrementCount(Hash(v

i

, v

j

, v

k

), H)

7: end for

8: for v

i

∈ {initialize counters}

9: c

i

= 0

10: end for

11: for (k, v) ∈ H do

12: if v = 3 then

13: for v

i

∈ k do

14: c

i

++

15: end for

16: end if

17: ...