Compute the User similarity matrix , Item similarity matrix , User-item attention matrix , Item-user attention matrix according to equations (3), (6), (8) and (9).
//Community detection in double-layer network
for each node do
Apply vector dynamic evolution clustering equations (13) and (15) to find the appropriate community for each node.
If and then all nodes stop iterating
end if
end for
//Calculate the dynamic similarity
for each user community and item community do
Apply equation (17) to calculate the similarity matrix in the same community.
end for
//Prediction score
for each target user and target item do
Find the neighbor set of the target user by similarity sort.
Compute the prediction score according to equation (19)
end for
//Select the Top-N items as recommendation list for target user