SHMF: Interest Prediction Model with Social Hub Matrix Factorization

Algorithm 1

The process of predicting users’ interests.

Require:

 Dataset: ;

 The dimension of the latent feature: ;;

 Parameters: ;

 An updating parameter:

 Convergence parameter:

 The maximum number of iterations:

Ensure:

  The user-topic matrix in time segment :

(1)

(2) for    do

(3)   initialize ;

(4)    if  then

(5)    Compute the mean matrices

(6)   end if

(7)   for    do

(8)     compute the gradient descent in Eq. (15) (16) (17) (18);

(9)    updating in Eq. (14);

(10)    compute in Eq. (13);

(11)    if    then

(12)     break

(13)    end  if

(14)     if   then

(15)     

(16)    end  if

(17)   end for

(18) ,

(19) end for

(20) predict using