The Algorithm of Link Prediction on Social Network

Algorithm 2

KatzGF algorithm.

Input: social network graph 𝐺=⟨𝑉,𝐸⟩, node 𝑥, node 𝑦

Output: the similarity of node 𝑥 and node 𝑦

The description of the algorithm:

(1) Find all paths between node 𝑥 and node 𝑦 which length are less then 6. Put all the

   nodes in the paths into the set 𝑥𝑦.path_nodes. Record the length of each path.

(2) Extract the sub-graph which contains the tested node pair and all the nodes in their paths.

(3) for (from the first path to the last path){

(4)   while (the nodes set of the path is not null){

(5)   Calculate the degree of node 𝑣, and get 𝑣.degree. Calculate the degree of node 𝑣

    in the extracted sub-graph in the Step 2, get 𝑣.path_degree.

      Calculate the guidance capability of this node

    Guidance-capability. 𝑣=𝑣.path_degree/log(𝑣.degree).

(6)   Calculate the weight of this path Guidance_capability.path+ = Guidance_capability.𝑣}.

(7) Calculate the similarity of node 𝑥 and node 𝑦 Similarity.𝑥𝑦+ = 𝛽^𝑙 Guidance_capability.path}