New Benchmarks for Learning on Non-Homophilous Graphs
This work addresses the need for better benchmarks in graph machine learning for non-homophilous data, which is incremental as it builds on existing datasets and measures.
The authors tackled the problem of limited and inadequate datasets for evaluating graph machine learning methods in non-homophilous settings by introducing improved graph datasets and a new measure of homophily, benchmarking various methods to draw new insights.
Much data with graph structures satisfy the principle of homophily, meaning that connected nodes tend to be similar with respect to a specific attribute. As such, ubiquitous datasets for graph machine learning tasks have generally been highly homophilous, rewarding methods that leverage homophily as an inductive bias. Recent work has pointed out this particular focus, as new non-homophilous datasets have been introduced and graph representation learning models better suited for low-homophily settings have been developed. However, these datasets are small and poorly suited to truly testing the effectiveness of new methods in non-homophilous settings. We present a series of improved graph datasets with node label relationships that do not satisfy the homophily principle. Along with this, we introduce a new measure of the presence or absence of homophily that is better suited than existing measures in different regimes. We benchmark a range of simple methods and graph neural networks across our proposed datasets, drawing new insights for further research. Data and codes can be found at https://github.com/CUAI/Non-Homophily-Benchmarks.