Amazon Ratings
Description
Amazon Ratings
Dataset Overview
Lifting Methods
Structural-based Liftings
- Cellular: Cycle-based lifting
- Simplicial: Clique complex lifting
- Hypergraph: k-hop lifting
Feature Lifting
- Projected sum
Model Performance
| Model | Accuracy | Std Dev |
|---|---|---|
| GCN | 49.56 | 0.55 |
| GIN | 49.16 | 1.02 |
| GAT | 50.17 | 0.59 |
| AST | 50.50 | 0.27 |
| EDGNN | 48.18 | 0.09 |
| UniGNN2 | 49.06 | 0.08 |
| CWN | 51.90 | 0.15 |
| CCCN | 50.26 | 0.17 |
| SCCNN | OOM | OOM |
| SCN | OOM | OOM |
Key Insights
- CWN achieves the best performance with 51.90 ± 0.15 accuracy
- Most models perform in the 48–51% range, with CWN outperforming others by a notable margin
- GIN shows relatively high variability (±1.02)
- EDGNN shows the most stable results with lowest std dev (±0.09)
- SCCNN and SCN ran out of memory (OOM)
Reproducibility
python -m topobench model=cell/cwn dataset=graph/amazon_ratings optimizer.parameters.lr=0.001 model.feature_encoder.out_channels=128 model.backbone.n_layers=4 model.readout.readout_name=PropagateSignalDown model.feature_encoder.proj_dropout=0.25 dataset.dataloader_params.batch_size=1 transforms.graph2cell_lifting.max_cell_length=10 dataset.split_params.data_seed=0,3,5,7,9 trainer.max_epochs=1000 trainer.min_epochs=50 trainer.check_val_every_n_epoch=1 callbacks.early_stopping.patience=50 logger.wandb.project=TopoBench --multirun