Covert Network Analysis for Key Player Detection and Event Prediction Using a Hybrid Classifier

Algorithm 1

Algorithm to find outliers.

Procedure:  Find Outliers

Input:  𝑘, the number of nearest neighbors;

𝑛, the number of outliers to be returned;

𝐷, the set of data points.

Output:  𝑂, the set of outliers.

Begin

  𝑐=0 (𝑐 is the cutoff threshold)

  𝑂={}

  for each 𝑑 in 𝐷  do

     Neighbors(𝑑)={}

     for each 𝑏 in 𝐷 such that 𝑏 <> 𝑑  do

        if  Neighbors(𝑑)<𝑘 or Distance(𝑏,𝑑) < Maxdist(𝑑, Neighbors(𝑑))

        then

         Neighbors(𝑑) = Closest(𝑑,Neighbors(𝑑)⋃𝑏,𝑘)

        end if

        if  Neighbors(𝑑) ≥ 𝑘 and 𝑐 > Distance(𝑏,𝑑)) then

         Break

        end if

     end for

     𝑂 = TopOutliers(𝑂⋃𝑏,𝑛)

     𝑐 = MaxThreshold(𝑂)

  end for

End