Multityped Community Discovery in Time-Evolving Heterogeneous Information Networks Based on Tensor Decomposition

Algorithm 1

SOSComm.

Input: the tensor representation of the current network

, the number of multi-typed communities ,

temporally regularization parameter , the community

membership matrices for the previous network snapshot

, and maximum iterations .

Output: community membership matrices for the current

network snapshot and multi-typed

communities .

(1) Set  ;

(2) Set ;

(3) repeat

(4) for    to    do

(5) Set ;

(6) Update according to (21);

(7) Normalize according to (22);

(8) end for

(9) Set ;

(10) until   unchanged or

(11) Recover the discrete community membership matrices

from (optional).

(12) Extract the multi-typed communities according to (3).