An Unsupervised Framework for Comparing Graph Embeddings The goal of many machine learning applications is to make predictions or discover new patterns using graph-structured data as feature information. In order to extract useful structural information from graphs, one might want