Graph-Based Collaborative Filtering with MLP

Algorithm 1

GKM.

Input: Rating Matrix: R, The dimension of feature: K

Procedure:

Decompose matrix with SVD method. ,

For all user do

Receive the vector of

Compute the similarity between users with Equation (6)

end for

Through the activation function and compute the adjacency matrix A

Establish graph

For all user do

Traverse graph and receive the neighbors w

Compute the feature vector of with , w

end for

 Output: The feature vector of user: The vector of item: