Figure 1: EmbeddingVis consists of (1) control panel, (2) graph view, (3) cluster transition view, (4) pairwise ranking view and (5) structural view. Using EmbeddingVis to verify preserved metrics at the instance level: (a) users lasso a cluster of nodes and the corresponding nodes are highlighted and linked across different embedding spaces (b1-e1). Clicking one "hub" node in this cluster (label 20) in the graph view generates four neighbor ranking lists of this "hub" node: (b2) the graph space, (c2) DeepWalk, (d2) struc2vec, and (e2) node2vec. The highlighted rectangles (c2, d2, e2) indicate the preserved metrics by each embedding. (f) Users can click "Filter" to highlight the nodes of each ranking list in the cluster transition view (b3-e3). (g) The Euclidean distance between adjacent nodes in struc2vec is more volatile compared with that in the other two spaces. ABSTRACTConstructing latent vector representation for nodes in a network through embedding models has shown its practicality in many graph analysis applications, such as node classification, clustering, and link prediction. However, despite the high efficiency and accuracy of learning an embedding model, people have little clue of what information about the original network is preserved in the embedding vectors. The abstractness of low-dimensional vector representation, stochastic nature of the construction process, and non-transparent hyper-parameters all obscure understanding of network embedding results. Visualization techniques have been introduced to facilitate embedding vector inspection, usually by projecting the embedding *

No abstract