Space4HGNN: A Novel, Modularized and Reproducible Platform to Evaluate Heterogeneous Graph Neural Network

TL;DR

Space4HGNN is a modular, reproducible platform that unifies and evaluates various heterogeneous graph neural network designs, facilitating better understanding and comparison across architectures.

Contribution

It introduces a unified framework and a platform for standardized, reproducible evaluation of HGNNs, simplifying implementation and analysis.

Findings

Identified key design dimensions affecting HGNN performance

Provided a condensed design space for HGNN architecture choices

Validated the platform's effectiveness through experiments

Abstract

Heterogeneous Graph Neural Network (HGNN) has been successfully employed in various tasks, but we cannot accurately know the importance of different design dimensions of HGNNs due to diverse architectures and applied scenarios. Besides, in the research community of HGNNs, implementing and evaluating various tasks still need much human effort. To mitigate these issues, we first propose a unified framework covering most HGNNs, consisting of three components: heterogeneous linear transformation, heterogeneous graph transformation, and heterogeneous message passing layer. Then we build a platform Space4HGNN by defining a design space for HGNNs based on the unified framework, which offers modularized components, reproducible implementations, and standardized evaluation for HGNNs. Finally, we conduct experiments to analyze the effect of different designs. With the insights found, we distill a condensed design space and verify its effectiveness.