This is the official repository for the paper "GraphLand: Evaluating Graph Machine Learning Models on Diverse Industrial Data".
GraphLand datasets are available at Zenodo.
This repository can be used to run experiments with GNNs on many graph datasets including those from the GraphLand benchmark.
How to use it:
If you want to use datasets from the GraphLand benchmark, download them from Zenodo and put them in the data directory. If you want to use other supported datasets, they can be downloaded by the Dataset class automatically.
If you simply want to get access to GraphLand and other datasets in a unififed format, you can use the Dataset class from the dataset.py file — it is self-contained.
If you want to run experiments in this codebase, install the dependencies from environment.yml and run main.py.
Executing main.py runs a single experiment which might include hyperparameter search and multiple runs with the best hyperparameters to compute the mean and standard deviation of model performance. main.py can accept a number of arguments, see get_args function from args.py for a full list. A simple example is:
python main.py --name example_experiment --dataset hm-categories --split RL --transductive True --model GT --lr 3e-4 --dropout 0.1 --device cuda:0
Many arguments can accept a list of values. Passing a list of values rather than a single value will trigger grid search over these values. For example, to run grid search over learning rate and dropout probability, run the following command:
python main.py --name example_experiment_with_hparam_search --dataset hm-categories --split RL --transductive True --model GT --lr 3e-5 1e-4 3e-4 1e-3 3e-3 --dropout 0 0.1 0.2 --device cuda:0
The directory scripts contains commands to reproduce all the GNN experiments in our paper. Specifically, there are 4 file, one for each setting in our paper: run_experiments_RL.sh, run_experiments_RH.sh, run_experiments_TH.sh, run_experiments_THI.sh. Each file contains commands to reproduce all GNN experiments for the corresponding setting (RL, RH, TH, THI). Each line runs a single experiment (with a single model and a single dataset) including hyperparameter search. Note that running all the experiments takes considerable time (multiple days on a single A100 GPU), although experimenting with smaller datasets is much faster than with larger ones.
If you found GraphLand datasets or this repository useful, please cite the following work:
@article{bazhenov2025graphland,
title={GraphLand: Evaluating graph machine learning models on diverse industrial data},
author={Bazhenov, Gleb and Platonov, Oleg and Prokhorenkova, Liudmila},
journal={arXiv preprint},
year={2025}
}