Abstract:
A new label propagation algorithm based on node intimacy is proposed for the randomness and instability of overlapping community discovery algorithms based on tag propagation. Firstly, based on the module degree increment, it is coarsely clustered the nodes in the network using the local information of the network. Then, it is defined the node affinity function for tag update and selection. The algorithm is validated on artificial and real networks. The results show that the algorithm can effectively improve the accuracy and stability of large-scale overlapping community detection, and has near linear time complexity.