A new network node similarity measure method and its applications

Network node similarity measure has been paid particular attention in the field of statistical physics. In this paper, we utilize the concept of information and information loss to measure the node similarity. The whole model is based on this idea that if two nodes are more similar than the others, then the information loss of seeing them as the same is less. The present new method has low algorithm complexity so that it can save much time and energy to deal with the large scale real-world network. We illustrate the availability of this approach based on two artificial examples and computer-generated networks by comparing its accuracy with the other selected approaches. The above tests demonstrate that the new method can provide more reasonable results consistent with our human common judgment. The new similarity measure method is also applied to predict the network evolution and predict unknown nodes' attributions in the two application examples.