Joint Use of Node Attributes and Proximity for Node Classification

Node classification aims to infer unknown node labels from known labels and other node attributes. Standard approaches for this task assume homophily, whereby a node’s label is predicted from the labels of other nodes nearby in the network. However, there are also cases of networks where labels are better predicted from the individual attributes of each node rather than the labels of nearby nodes. Ideally, node classification methods should flexibly adapt to a range of settings wherein unknown labels are predicted either from labels of nearby nodes, or individual node attributes, or partly both. In this paper, we propose a principled approach, JANE, based on a generative probabilistic model that jointly weighs the role of attributes and node proximity via embeddings in predicting labels. Experiments on multiple network datasets demonstrate that JANE exhibits the desired combination of versatility and competitive performance compared to baselines.