Quantification of Network Structural Dissimilarities Based on Graph Embedding

Identifying and quantifying structural dissimilarities between complex networks is a fundamental challenging problem in network science. Previous comparison methods are based on the features, such as length of shortest path, degree graphlet, which may only contain part topological information. Therefore, we propose an efficient method embedding, i.e., \textit{DeepWalk}, considers global In detail, calculate distance nodes through vector extracted by \textit{DeepWalk} quantify dissimilarity spectral entropy Jensen-Shannon divergences distribution node distances. Experiments both synthetic empirical data show that our outperforms baseline can distinguish perfectly using embedding distribution. addition, capture properties, e.g., average path link density. Moreover, experiments modularity further implies functionality method.