Amazon-Book Top-K (Recall@20) amazon-book-recall20 Leaderboard
Top-K collaborative-filtering ranking on the Amazon-Book implicit-feedback benchmark (the canonical LightGCN split). Recall@20 over all users; higher is better. Transcribed from the SimGCL paper Table 5; LightGCN/SGL rows match those papers exactly, confirming a shared split. · Metric: Recall@20 (higher is better)
| # | Model | Recall@20 | Paper |
|---|---|---|---|
| 1 | SimGCL | 0.05 | link |
| 2 | MixGCF | 0.05 | link |
| 3 | SGL | 0.05 | link |
| 4 | DNN+SSL | 0.04 | link |
| 5 | LightGCN | 0.04 | link |
| 6 | Mult-VAE | 0.04 | link |