CPSC_583_Final_Project_KG

This is the github repository for Fall 23 CPSC 583 Final Project by Bowen Duanmu, Evan Shi, Andrew Yi.

Project Overview

Our project can be devided into three main part:

Traditional GNN methods' on tackling link prediction tasks on complex heterogeneous KGs.
More cutting-edge methods (KG Embeddings / ensemble learning) for the same link prediction tasks.
Genetic Algorithm.

Please first select a version and install torch with CUDA

Then install pytorch geometric

pip install torch_geometric

Make sure the installation is performed correctly

python -c "import torch; print(torch.version.cuda)"

And then install other pyg/torch related dependencies

pip install pyg_lib torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-${TORCH}+${CUDA}.html

Where TORCH and CUDA are replaced by the specific versions according to the official documentation https://pytorch-geometric.readthedocs.io/en/latest/install/installation.html

Then, you can simply do

pip install -r requirements.txt

For the rest of the dependencies

Some code requires ogb==1.2.4 while others work fine with the newest version

Please refer to the individual README.md files in the MovieLens and obgl-biokg directories for separate instructions