Measuring and Improving the Use of Graph Information in Graph Neural Networks

TL;DR

This paper introduces a framework and metrics to quantify how effectively GNNs utilize graph data, and proposes a new model, CS-GNN, that improves performance by enhancing the use of graph information.

Contribution

It presents a novel framework with smoothness metrics to measure graph information use and introduces CS-GNN, a model that leverages these metrics for improved performance.

Findings

CS-GNN outperforms existing GNN methods on real graph datasets.

The smoothness metrics effectively quantify the quality and quantity of graph information used.

The framework provides insights into how GNNs utilize graph structures.

Abstract

Graph neural networks (GNNs) have been widely used for representation learning on graph data. However, there is limited understanding on how much performance GNNs actually gain from graph data. This paper introduces a context-surrounding GNN framework and proposes two smoothness metrics to measure the quantity and quality of information obtained from graph data. A new GNN model, called CS-GNN, is then designed to improve the use of graph information based on the smoothness values of a graph. CS-GNN is shown to achieve better performance than existing methods in different types of real graphs.