Similarity Propagation Based Node Matching between Complex Networks

A similarity propagation based algorithm is proposed for the node matching problem between complex networks.In order to fully exploit the similarity information provided by the limited numbers of revealed matching node pairs,a node similarity propagation process is introduced,which makes the initial similarity information propagate along the network topology globally.The steady distribution of this propagation process is equivalent to the principle eigenvector of a large matrix,which can be efficiently solved by the power method in an iterative way.The final matching node pairs are extracted by the KM(Kuhn-Munkres) algorithm in graph theory.The proposed method is applied to solving the node matching problems among four different types of networks,respectively.It is revealed that,the proposed algorithm can significantly improve the node matching precision.