Self-supervised Heterogeneous Graph Learning: a Homophily and Heterogeneity View

This repository contains the reference code for the manuscript ``Self-supervised Heterogeneous Graph Learning: a Homophily and Heterogeneity View"

Contents

0. Installation
1. Preparation
2. Training

Installation

• pip install -r requirements.txt
• Unzip the datasets (the heterogeneous graph datasets can be found in ./dataset/, and the homogeneous graph datasets will be automatically downloaded from the public website)

Preparation

Important args:

• --use_pretrain Test checkpoints to reproduce the results
• --dataset Heterogeneous graph: ACM, Yelp, DBLP, Aminer || Homogeneous graph: photo, computers, cs, physics
• --custom_key Node: node classification

Training

python main.py