CTR Prediction on MovieLens 20M (test)
97.9AUCKGNN-LS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| KGNN-LS2019.05 | 97.9 | — | |
| KGCN-sumaggregator=sum2019.03 | 97.8 | 93.2 | |
| KGCN-concataggregator=concat2019.03 | 97.7 | 93.1 | |
| KGCN-neighboraggregator=neighbor2019.03 | 97.7 | 93.2 | |
| KGCN-avgaggregator=average2019.03 | 97.5 | 92.9 | |
| RippleNet2019.03 | 96.8 | 91.2 | |
| LibFM + TransEKnowledge Graph embedding=TransE2019.03 | 96.6 | 91.7 | |
| LibFM + TransETransE dimension=322019.05 | 96.6 | — | |
| SVDtype=unbiased2019.03 | 96.3 | 91.9 | |
| SVD2019.05 | 96.3 | — | |
| RippleNetd=8, H=2, lambda1=10^-6, lambda2=0.01, eta=0.012019.05 | 96 | — | |
| LibFM2019.03 | 95.9 | 90.6 | |
| LibFMdimension={1, 1, 8}, training epochs=502019.05 | 95.9 | — | |
| CKE2019.03 | 92.4 | 87.1 | |
| CKEembedding dimension=64, KG training weight=0.12019.05 | 92.4 | — | |
| PERfeatures=meta-path based2019.03 | 83.2 | 78.8 | |
| PERmeta-paths=user-movie-director-movie, user-movie-genre-movie, user-movie-star-movie2019.05 | 83.2 | — |